Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"

This reverts commit 98c37247d8.

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_abd.c

@@ -428,16 +428,14 @@ svuint64_t test_svabd_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svabd_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svabd_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_svabd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svabd_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_svabd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_svabd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svabd_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_svabd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_svabd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svabd_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_svabd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_svabd_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svabd_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svabd_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_svabd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svabd_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_svabd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_svabd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svabd_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_svabd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_svabd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svabd_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_svabd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_svabd_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svabd_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svabd_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_svabd_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svabd_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_svabd_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svabd_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_svabd_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svabd_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_svabd_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svabd_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svabd_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_svabd_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svabd_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_svabd_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svabd_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_svabd_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svabd_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_svabd_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svabd_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svabd_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_svabd_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svabd_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_svabd_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svabd_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_svabd_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svabd_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_svabd_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svabd_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svabd_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uabd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_svabd_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svabd_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_svabd_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svabd_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_svabd_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svabd_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -1059,7 +1011,6 @@ svfloat16_t test_svabd_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2)
   return SVE_ACLE_FUNC(svabd,_f16,_x,)(pg, op1, op2);
 }
 
-//
 // CHECK-LABEL: @test_svabd_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
@@ -1097,8 +1048,7 @@ svfloat64_t test_svabd_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svabd_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1106,8 +1056,7 @@ svfloat64_t test_svabd_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1120,8 +1069,7 @@ svfloat16_t test_svabd_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svabd_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1129,8 +1077,7 @@ svfloat16_t test_svabd_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1143,8 +1090,7 @@ svfloat32_t test_svabd_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svabd_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1152,8 +1098,7 @@ svfloat32_t test_svabd_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_f64_zu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1166,16 +1111,14 @@ svfloat64_t test_svabd_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svabd_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1187,16 +1130,14 @@ svfloat16_t test_svabd_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svabd_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1208,16 +1149,14 @@ svfloat32_t test_svabd_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svabd_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_f64_mu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -1229,16 +1168,14 @@ svfloat64_t test_svabd_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svabd_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fabd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1250,16 +1187,14 @@ svfloat16_t test_svabd_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svabd_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fabd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1271,16 +1206,14 @@ svfloat32_t test_svabd_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svabd_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabd_n_f64_xu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fabd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acge.c

@@ -74,8 +74,7 @@ svbool_t test_svacge_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svacge_n_f32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -83,8 +82,7 @@ svbool_t test_svacge_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z17test_svacge_n_f32u10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -97,8 +95,7 @@ svbool_t test_svacge_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svacge_n_f64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -106,8 +103,7 @@ svbool_t test_svacge_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z17test_svacge_n_f64u10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acgt.c

@@ -74,8 +74,7 @@ svbool_t test_svacgt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svacgt_n_f32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -83,8 +82,7 @@ svbool_t test_svacgt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z17test_svacgt_n_f32u10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -97,8 +95,7 @@ svbool_t test_svacgt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svacgt_n_f64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -106,8 +103,7 @@ svbool_t test_svacgt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z17test_svacgt_n_f64u10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_acle.c

@@ -74,8 +74,7 @@ svbool_t test_svacle_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svacle_n_f32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -83,8 +82,7 @@ svbool_t test_svacle_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z17test_svacle_n_f32u10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -97,8 +95,7 @@ svbool_t test_svacle_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svacle_n_f64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -106,8 +103,7 @@ svbool_t test_svacle_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z17test_svacle_n_f64u10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_aclt.c

@@ -74,8 +74,7 @@ svbool_t test_svaclt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svaclt_n_f32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -83,8 +82,7 @@ svbool_t test_svaclt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z17test_svaclt_n_f32u10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.facgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -97,8 +95,7 @@ svbool_t test_svaclt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svaclt_n_f64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -106,8 +103,7 @@ svbool_t test_svaclt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z17test_svaclt_n_f64u10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.facgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_add.c

@@ -428,16 +428,14 @@ svuint64_t test_svadd_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svadd_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svadd_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_svadd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svadd_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_svadd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_svadd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svadd_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_svadd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_svadd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svadd_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_svadd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_svadd_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svadd_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svadd_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_svadd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svadd_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_svadd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_svadd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svadd_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_svadd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_svadd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svadd_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_svadd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_svadd_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svadd_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svadd_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_svadd_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svadd_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_svadd_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svadd_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_svadd_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svadd_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_svadd_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svadd_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svadd_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_svadd_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svadd_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_svadd_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svadd_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_svadd_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svadd_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_svadd_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svadd_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svadd_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_svadd_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svadd_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_svadd_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svadd_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_svadd_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svadd_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_svadd_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svadd_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svadd_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.add.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_svadd_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svadd_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.add.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_svadd_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svadd_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.add.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_svadd_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svadd_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.add.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -1096,8 +1048,7 @@ svfloat64_t test_svadd_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svadd_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1105,8 +1056,7 @@ svfloat64_t test_svadd_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1119,8 +1069,7 @@ svfloat16_t test_svadd_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svadd_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1128,8 +1077,7 @@ svfloat16_t test_svadd_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1142,8 +1090,7 @@ svfloat32_t test_svadd_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svadd_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1151,8 +1098,7 @@ svfloat32_t test_svadd_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_f64_zu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1165,16 +1111,14 @@ svfloat64_t test_svadd_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svadd_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1186,16 +1130,14 @@ svfloat16_t test_svadd_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svadd_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1207,16 +1149,14 @@ svfloat32_t test_svadd_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svadd_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_f64_mu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -1228,16 +1168,14 @@ svfloat64_t test_svadd_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svadd_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fadd.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1249,16 +1187,14 @@ svfloat16_t test_svadd_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svadd_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fadd.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1270,16 +1206,14 @@ svfloat32_t test_svadd_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svadd_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadd_n_f64_xu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fadd.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_and.c

@@ -428,16 +428,14 @@ svuint64_t test_svand_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svand_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svand_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_svand_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svand_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_svand_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svand_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_svand_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svand_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_svand_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svand_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_svand_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svand_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_svand_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svand_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_svand_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svand_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svand_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_svand_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svand_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_svand_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svand_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_svand_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svand_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_svand_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svand_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_svand_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svand_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_svand_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svand_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_svand_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svand_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svand_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_svand_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svand_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svand_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_svand_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svand_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svand_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_svand_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svand_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svand_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_svand_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svand_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svand_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_svand_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svand_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svand_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_svand_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svand_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svand_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_svand_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svand_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svand_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_svand_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svand_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svand_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_svand_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svand_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svand_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_svand_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svand_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svand_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_svand_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svand_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svand_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_svand_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svand_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svand_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.and.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_svand_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svand_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svand_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.and.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_svand_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svand_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svand_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.and.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_svand_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svand_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svand_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.and.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_asr.c

@@ -223,8 +223,7 @@ svint64_t test_svasr_s64_x(svbool_t pg, svint64_t op1, svuint64_t op2)
 // CHECK-LABEL: @test_svasr_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -232,8 +231,7 @@ svint64_t test_svasr_s64_x(svbool_t pg, svint64_t op1, svuint64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svasr_n_s64_zu10__SVBool_tu11__SVInt64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -246,16 +244,14 @@ svint64_t test_svasr_n_s64_z(svbool_t pg, svint64_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svasr_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svasr_n_s64_mu10__SVBool_tu11__SVInt64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -267,16 +263,14 @@ svint64_t test_svasr_n_s64_m(svbool_t pg, svint64_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svasr_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svasr_n_s64_xu10__SVBool_tu11__SVInt64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.asr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -440,16 +434,14 @@ svint32_t test_svasr_wide_s32_x(svbool_t pg, svint32_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svasr_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svasr_n_s8_zu10__SVBool_tu10__SVInt8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -462,8 +454,7 @@ svint8_t test_svasr_n_s8_z(svbool_t pg, svint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svasr_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -471,8 +462,7 @@ svint8_t test_svasr_n_s8_z(svbool_t pg, svint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svasr_n_s16_zu10__SVBool_tu11__SVInt16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -485,8 +475,7 @@ svint16_t test_svasr_n_s16_z(svbool_t pg, svint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svasr_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -494,8 +483,7 @@ svint16_t test_svasr_n_s16_z(svbool_t pg, svint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svasr_n_s32_zu10__SVBool_tu11__SVInt32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -507,15 +495,13 @@ svint32_t test_svasr_n_s32_z(svbool_t pg, svint32_t op1, uint32_t op2)
 
 // CHECK-LABEL: @test_svasr_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svasr_n_s8_mu10__SVBool_tu10__SVInt8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -527,16 +513,14 @@ svint8_t test_svasr_n_s8_m(svbool_t pg, svint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svasr_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svasr_n_s16_mu10__SVBool_tu11__SVInt16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -548,16 +532,14 @@ svint16_t test_svasr_n_s16_m(svbool_t pg, svint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svasr_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svasr_n_s32_mu10__SVBool_tu11__SVInt32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -568,15 +550,13 @@ svint32_t test_svasr_n_s32_m(svbool_t pg, svint32_t op1, uint32_t op2)
 
 // CHECK-LABEL: @test_svasr_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svasr_n_s8_xu10__SVBool_tu10__SVInt8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -588,16 +568,14 @@ svint8_t test_svasr_n_s8_x(svbool_t pg, svint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svasr_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svasr_n_s16_xu10__SVBool_tu11__SVInt16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -609,16 +587,14 @@ svint16_t test_svasr_n_s16_x(svbool_t pg, svint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svasr_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svasr_n_s32_xu10__SVBool_tu11__SVInt32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -629,15 +605,13 @@ svint32_t test_svasr_n_s32_x(svbool_t pg, svint32_t op1, uint32_t op2)
 
 // CHECK-LABEL: @test_svasr_wide_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svasr_wide_n_s8_mu10__SVBool_tu10__SVInt8_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -649,16 +623,14 @@ svint8_t test_svasr_wide_n_s8_m(svbool_t pg, svint8_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svasr_wide_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s16_mu10__SVBool_tu11__SVInt16_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -670,16 +642,14 @@ svint16_t test_svasr_wide_n_s16_m(svbool_t pg, svint16_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svasr_wide_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s32_mu10__SVBool_tu11__SVInt32_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -690,16 +660,14 @@ svint32_t test_svasr_wide_n_s32_m(svbool_t pg, svint32_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svasr_wide_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svasr_wide_n_s8_zu10__SVBool_tu10__SVInt8_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -712,8 +680,7 @@ svint8_t test_svasr_wide_n_s8_z(svbool_t pg, svint8_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svasr_wide_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -721,8 +688,7 @@ svint8_t test_svasr_wide_n_s8_z(svbool_t pg, svint8_t op1, uint64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s16_zu10__SVBool_tu11__SVInt16_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -735,8 +701,7 @@ svint16_t test_svasr_wide_n_s16_z(svbool_t pg, svint16_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svasr_wide_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -744,8 +709,7 @@ svint16_t test_svasr_wide_n_s16_z(svbool_t pg, svint16_t op1, uint64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s32_zu10__SVBool_tu11__SVInt32_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -757,15 +721,13 @@ svint32_t test_svasr_wide_n_s32_z(svbool_t pg, svint32_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svasr_wide_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svasr_wide_n_s8_xu10__SVBool_tu10__SVInt8_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.asr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -777,16 +739,14 @@ svint8_t test_svasr_wide_n_s8_x(svbool_t pg, svint8_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svasr_wide_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s16_xu10__SVBool_tu11__SVInt16_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.asr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -798,16 +758,14 @@ svint16_t test_svasr_wide_n_s16_x(svbool_t pg, svint16_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svasr_wide_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z23test_svasr_wide_n_s32_xu10__SVBool_tu11__SVInt32_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.asr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfdot.c

@@ -57,15 +57,13 @@ svfloat32_t test_bfdot_lane_3_f32(svfloat32_t x, svbfloat16_t y, svbfloat16_t z)
 
 // CHECK-LABEL: @test_bfdot_n_f32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_bfdot_n_f32u13__SVFloat32_tu14__SVBFloat16_tu6__bf16(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfdot(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfmlalb.c

@@ -57,15 +57,13 @@ svfloat32_t test_bfmlalb_lane_7_f32(svfloat32_t x, svbfloat16_t y, svbfloat16_t
 
 // CHECK-LABEL: @test_bfmlalb_n_f32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_bfmlalb_n_f32u13__SVFloat32_tu14__SVBFloat16_tu6__bf16(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalb(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bfmlalt.c

@@ -57,15 +57,13 @@ svfloat32_t test_bfmlalt_lane_7_f32(svfloat32_t x, svbfloat16_t y, svbfloat16_t
 
 // CHECK-LABEL: @test_bfmlalt_n_f32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_bfmlalt_n_f32u13__SVFloat32_tu14__SVBFloat16_tu6__bf16(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[Z:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[Z:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.bfmlalt(<vscale x 4 x float> [[X:%.*]], <vscale x 8 x bfloat> [[Y:%.*]], <vscale x 8 x bfloat> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_bic.c

@@ -428,16 +428,14 @@ svuint64_t test_svbic_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svbic_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbic_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_svbic_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svbic_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_svbic_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_svbic_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svbic_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_svbic_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_svbic_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svbic_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_svbic_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_svbic_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svbic_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbic_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_svbic_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svbic_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_svbic_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_svbic_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svbic_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_svbic_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_svbic_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svbic_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_svbic_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_svbic_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svbic_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbic_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_svbic_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svbic_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_svbic_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svbic_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_svbic_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svbic_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_svbic_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svbic_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbic_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_svbic_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svbic_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_svbic_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svbic_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_svbic_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svbic_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_svbic_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svbic_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbic_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_svbic_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svbic_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_svbic_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svbic_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_svbic_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svbic_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_svbic_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svbic_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbic_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bic.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_svbic_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svbic_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bic.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_svbic_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svbic_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bic.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_svbic_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svbic_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbic_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bic.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpeq.c

@@ -161,8 +161,7 @@ svbool_t test_svcmpeq_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CHECK-LABEL: @test_svcmpeq_n_s64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpeq.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -170,8 +169,7 @@ svbool_t test_svcmpeq_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_s64u10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpeq.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -184,8 +182,7 @@ svbool_t test_svcmpeq_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmpeq_n_u64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpeq.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -193,8 +190,7 @@ svbool_t test_svcmpeq_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_u64u10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpeq.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -259,15 +255,13 @@ svbool_t test_svcmpeq_wide_s32(svbool_t pg, svint32_t op1, svint64_t op2)
 
 // CHECK-LABEL: @test_svcmpeq_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svcmpeq_n_s8u10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -279,8 +273,7 @@ svbool_t test_svcmpeq_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svcmpeq_n_s16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -288,8 +281,7 @@ svbool_t test_svcmpeq_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_s16u10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -302,8 +294,7 @@ svbool_t test_svcmpeq_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svcmpeq_n_s32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -311,8 +302,7 @@ svbool_t test_svcmpeq_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_s32u10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -324,15 +314,13 @@ svbool_t test_svcmpeq_n_s32(svbool_t pg, svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svcmpeq_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svcmpeq_n_u8u10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -344,8 +332,7 @@ svbool_t test_svcmpeq_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svcmpeq_n_u16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -353,8 +340,7 @@ svbool_t test_svcmpeq_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_u16u10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -367,8 +353,7 @@ svbool_t test_svcmpeq_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svcmpeq_n_u32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -376,8 +361,7 @@ svbool_t test_svcmpeq_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_u32u10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -447,8 +431,7 @@ svbool_t test_svcmpeq_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svcmpeq_n_f16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpeq.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -456,8 +439,7 @@ svbool_t test_svcmpeq_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_f16u10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpeq.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -470,8 +452,7 @@ svbool_t test_svcmpeq_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svcmpeq_n_f32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpeq.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -479,8 +460,7 @@ svbool_t test_svcmpeq_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_f32u10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpeq.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -493,8 +473,7 @@ svbool_t test_svcmpeq_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svcmpeq_n_f64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpeq.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -502,8 +481,7 @@ svbool_t test_svcmpeq_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpeq_n_f64u10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpeq.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -515,15 +493,13 @@ svbool_t test_svcmpeq_n_f64(svbool_t pg, svfloat64_t op1, float64_t op2)
 
 // CHECK-LABEL: @test_svcmpeq_wide_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svcmpeq_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -535,8 +511,7 @@ svbool_t test_svcmpeq_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmpeq_wide_n_s16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -544,8 +519,7 @@ svbool_t test_svcmpeq_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmpeq_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -558,8 +532,7 @@ svbool_t test_svcmpeq_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmpeq_wide_n_s32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -567,8 +540,7 @@ svbool_t test_svcmpeq_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmpeq_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpeq.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpge.c

@@ -161,8 +161,7 @@ svbool_t test_svcmpge_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CHECK-LABEL: @test_svcmpge_n_s64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpge.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -170,8 +169,7 @@ svbool_t test_svcmpge_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpge_n_s64u10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpge.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -184,8 +182,7 @@ svbool_t test_svcmpge_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmpge_n_u64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphs.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -193,8 +190,7 @@ svbool_t test_svcmpge_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpge_n_u64u10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphs.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -312,15 +308,13 @@ svbool_t test_svcmpge_wide_u32(svbool_t pg, svuint32_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svcmpge_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svcmpge_n_s8u10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -332,8 +326,7 @@ svbool_t test_svcmpge_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svcmpge_n_s16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -341,8 +334,7 @@ svbool_t test_svcmpge_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpge_n_s16u10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -355,8 +347,7 @@ svbool_t test_svcmpge_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svcmpge_n_s32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -364,8 +355,7 @@ svbool_t test_svcmpge_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpge_n_s32u10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -377,15 +367,13 @@ svbool_t test_svcmpge_n_s32(svbool_t pg, svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svcmpge_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svcmpge_n_u8u10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -397,8 +385,7 @@ svbool_t test_svcmpge_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svcmpge_n_u16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -406,8 +393,7 @@ svbool_t test_svcmpge_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpge_n_u16u10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -420,8 +406,7 @@ svbool_t test_svcmpge_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svcmpge_n_u32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -429,8 +414,7 @@ svbool_t test_svcmpge_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpge_n_u32u10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -500,8 +484,7 @@ svbool_t test_svcmpge_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svcmpge_n_f16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpge.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -509,8 +492,7 @@ svbool_t test_svcmpge_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpge_n_f16u10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpge.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -523,8 +505,7 @@ svbool_t test_svcmpge_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svcmpge_n_f32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -532,8 +513,7 @@ svbool_t test_svcmpge_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpge_n_f32u10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -546,8 +526,7 @@ svbool_t test_svcmpge_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svcmpge_n_f64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -555,8 +534,7 @@ svbool_t test_svcmpge_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpge_n_f64u10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -568,15 +546,13 @@ svbool_t test_svcmpge_n_f64(svbool_t pg, svfloat64_t op1, float64_t op2)
 
 // CHECK-LABEL: @test_svcmpge_wide_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svcmpge_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -588,8 +564,7 @@ svbool_t test_svcmpge_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmpge_wide_n_s16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -597,8 +572,7 @@ svbool_t test_svcmpge_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmpge_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -611,8 +585,7 @@ svbool_t test_svcmpge_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmpge_wide_n_s32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -620,8 +593,7 @@ svbool_t test_svcmpge_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmpge_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -633,15 +605,13 @@ svbool_t test_svcmpge_wide_n_s32(svbool_t pg, svint32_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svcmpge_wide_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svcmpge_wide_n_u8u10__SVBool_tu11__SVUint8_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -653,8 +623,7 @@ svbool_t test_svcmpge_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svcmpge_wide_n_u16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -662,8 +631,7 @@ svbool_t test_svcmpge_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmpge_wide_n_u16u10__SVBool_tu12__SVUint16_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -676,8 +644,7 @@ svbool_t test_svcmpge_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svcmpge_wide_n_u32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -685,8 +652,7 @@ svbool_t test_svcmpge_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmpge_wide_n_u32u10__SVBool_tu12__SVUint32_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpgt.c

@@ -161,8 +161,7 @@ svbool_t test_svcmpgt_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CHECK-LABEL: @test_svcmpgt_n_s64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpgt.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -170,8 +169,7 @@ svbool_t test_svcmpgt_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_s64u10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpgt.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -184,8 +182,7 @@ svbool_t test_svcmpgt_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmpgt_n_u64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphi.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -193,8 +190,7 @@ svbool_t test_svcmpgt_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_u64u10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphi.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -312,15 +308,13 @@ svbool_t test_svcmpgt_wide_u32(svbool_t pg, svuint32_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svcmpgt_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svcmpgt_n_s8u10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -332,8 +326,7 @@ svbool_t test_svcmpgt_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svcmpgt_n_s16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -341,8 +334,7 @@ svbool_t test_svcmpgt_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_s16u10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -355,8 +347,7 @@ svbool_t test_svcmpgt_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svcmpgt_n_s32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -364,8 +355,7 @@ svbool_t test_svcmpgt_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_s32u10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -377,15 +367,13 @@ svbool_t test_svcmpgt_n_s32(svbool_t pg, svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svcmpgt_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svcmpgt_n_u8u10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -397,8 +385,7 @@ svbool_t test_svcmpgt_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svcmpgt_n_u16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -406,8 +393,7 @@ svbool_t test_svcmpgt_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_u16u10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -420,8 +406,7 @@ svbool_t test_svcmpgt_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svcmpgt_n_u32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -429,8 +414,7 @@ svbool_t test_svcmpgt_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_u32u10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -500,8 +484,7 @@ svbool_t test_svcmpgt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svcmpgt_n_f16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpgt.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -509,8 +492,7 @@ svbool_t test_svcmpgt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_f16u10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpgt.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -523,8 +505,7 @@ svbool_t test_svcmpgt_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svcmpgt_n_f32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -532,8 +513,7 @@ svbool_t test_svcmpgt_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_f32u10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -546,8 +526,7 @@ svbool_t test_svcmpgt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svcmpgt_n_f64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -555,8 +534,7 @@ svbool_t test_svcmpgt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpgt_n_f64u10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -568,15 +546,13 @@ svbool_t test_svcmpgt_n_f64(svbool_t pg, svfloat64_t op1, float64_t op2)
 
 // CHECK-LABEL: @test_svcmpgt_wide_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svcmpgt_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -588,8 +564,7 @@ svbool_t test_svcmpgt_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmpgt_wide_n_s16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -597,8 +572,7 @@ svbool_t test_svcmpgt_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmpgt_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -611,8 +585,7 @@ svbool_t test_svcmpgt_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmpgt_wide_n_s32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -620,8 +593,7 @@ svbool_t test_svcmpgt_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmpgt_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -633,15 +605,13 @@ svbool_t test_svcmpgt_wide_n_s32(svbool_t pg, svint32_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svcmpgt_wide_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svcmpgt_wide_n_u8u10__SVBool_tu11__SVUint8_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -653,8 +623,7 @@ svbool_t test_svcmpgt_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svcmpgt_wide_n_u16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -662,8 +631,7 @@ svbool_t test_svcmpgt_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmpgt_wide_n_u16u10__SVBool_tu12__SVUint16_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -676,8 +644,7 @@ svbool_t test_svcmpgt_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svcmpgt_wide_n_u32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -685,8 +652,7 @@ svbool_t test_svcmpgt_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmpgt_wide_n_u32u10__SVBool_tu12__SVUint32_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmple.c

@@ -161,8 +161,7 @@ svbool_t test_svcmple_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CHECK-LABEL: @test_svcmple_n_s64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpge.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -170,8 +169,7 @@ svbool_t test_svcmple_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmple_n_s64u10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpge.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -184,8 +182,7 @@ svbool_t test_svcmple_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmple_n_u64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphs.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -193,8 +190,7 @@ svbool_t test_svcmple_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmple_n_u64u10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphs.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -312,15 +308,13 @@ svbool_t test_svcmple_wide_u32(svbool_t pg, svuint32_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svcmple_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svcmple_n_s8u10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpge.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -332,8 +326,7 @@ svbool_t test_svcmple_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svcmple_n_s16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -341,8 +334,7 @@ svbool_t test_svcmple_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmple_n_s16u10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpge.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -355,8 +347,7 @@ svbool_t test_svcmple_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svcmple_n_s32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -364,8 +355,7 @@ svbool_t test_svcmple_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmple_n_s32u10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpge.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -377,15 +367,13 @@ svbool_t test_svcmple_n_s32(svbool_t pg, svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svcmple_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svcmple_n_u8u10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphs.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -397,8 +385,7 @@ svbool_t test_svcmple_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svcmple_n_u16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -406,8 +393,7 @@ svbool_t test_svcmple_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmple_n_u16u10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphs.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -420,8 +406,7 @@ svbool_t test_svcmple_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svcmple_n_u32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -429,8 +414,7 @@ svbool_t test_svcmple_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmple_n_u32u10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphs.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -500,8 +484,7 @@ svbool_t test_svcmple_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svcmple_n_f16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpge.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -509,8 +492,7 @@ svbool_t test_svcmple_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmple_n_f16u10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpge.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -523,8 +505,7 @@ svbool_t test_svcmple_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svcmple_n_f32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -532,8 +513,7 @@ svbool_t test_svcmple_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmple_n_f32u10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpge.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -546,8 +526,7 @@ svbool_t test_svcmple_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svcmple_n_f64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -555,8 +534,7 @@ svbool_t test_svcmple_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmple_n_f64u10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpge.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -568,15 +546,13 @@ svbool_t test_svcmple_n_f64(svbool_t pg, svfloat64_t op1, float64_t op2)
 
 // CHECK-LABEL: @test_svcmple_wide_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmple.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svcmple_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmple.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -588,8 +564,7 @@ svbool_t test_svcmple_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmple_wide_n_s16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmple.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -597,8 +572,7 @@ svbool_t test_svcmple_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmple_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmple.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -611,8 +585,7 @@ svbool_t test_svcmple_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmple_wide_n_s32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmple.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -620,8 +593,7 @@ svbool_t test_svcmple_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmple_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmple.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -633,15 +605,13 @@ svbool_t test_svcmple_wide_n_s32(svbool_t pg, svint32_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svcmple_wide_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpls.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svcmple_wide_n_u8u10__SVBool_tu11__SVUint8_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpls.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -653,8 +623,7 @@ svbool_t test_svcmple_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svcmple_wide_n_u16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpls.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -662,8 +631,7 @@ svbool_t test_svcmple_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmple_wide_n_u16u10__SVBool_tu12__SVUint16_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpls.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -676,8 +644,7 @@ svbool_t test_svcmple_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svcmple_wide_n_u32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpls.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -685,8 +652,7 @@ svbool_t test_svcmple_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmple_wide_n_u32u10__SVBool_tu12__SVUint32_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpls.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmplt.c

@@ -161,8 +161,7 @@ svbool_t test_svcmplt_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CHECK-LABEL: @test_svcmplt_n_s64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpgt.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -170,8 +169,7 @@ svbool_t test_svcmplt_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmplt_n_s64u10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpgt.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -184,8 +182,7 @@ svbool_t test_svcmplt_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmplt_n_u64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphi.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -193,8 +190,7 @@ svbool_t test_svcmplt_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmplt_n_u64u10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmphi.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP1]], <vscale x 2 x i64> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -312,15 +308,13 @@ svbool_t test_svcmplt_wide_u32(svbool_t pg, svuint32_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svcmplt_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svcmplt_n_s8u10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpgt.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -332,8 +326,7 @@ svbool_t test_svcmplt_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svcmplt_n_s16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -341,8 +334,7 @@ svbool_t test_svcmplt_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmplt_n_s16u10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpgt.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -355,8 +347,7 @@ svbool_t test_svcmplt_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svcmplt_n_s32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -364,8 +355,7 @@ svbool_t test_svcmplt_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmplt_n_s32u10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpgt.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -377,15 +367,13 @@ svbool_t test_svcmplt_n_s32(svbool_t pg, svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svcmplt_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svcmplt_n_u8u10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmphi.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -397,8 +385,7 @@ svbool_t test_svcmplt_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svcmplt_n_u16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -406,8 +393,7 @@ svbool_t test_svcmplt_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmplt_n_u16u10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmphi.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP1]], <vscale x 8 x i16> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -420,8 +406,7 @@ svbool_t test_svcmplt_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svcmplt_n_u32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -429,8 +414,7 @@ svbool_t test_svcmplt_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmplt_n_u32u10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmphi.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP1]], <vscale x 4 x i32> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -500,8 +484,7 @@ svbool_t test_svcmplt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svcmplt_n_f16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpgt.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -509,8 +492,7 @@ svbool_t test_svcmplt_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmplt_n_f16u10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpgt.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP1]], <vscale x 8 x half> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -523,8 +505,7 @@ svbool_t test_svcmplt_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svcmplt_n_f32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -532,8 +513,7 @@ svbool_t test_svcmplt_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmplt_n_f32u10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpgt.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP1]], <vscale x 4 x float> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -546,8 +526,7 @@ svbool_t test_svcmplt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svcmplt_n_f64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -555,8 +534,7 @@ svbool_t test_svcmplt_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmplt_n_f64u10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpgt.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP1]], <vscale x 2 x double> [[OP1:%.*]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -568,15 +546,13 @@ svbool_t test_svcmplt_n_f64(svbool_t pg, svfloat64_t op1, float64_t op2)
 
 // CHECK-LABEL: @test_svcmplt_wide_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmplt.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svcmplt_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmplt.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -588,8 +564,7 @@ svbool_t test_svcmplt_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmplt_wide_n_s16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmplt.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -597,8 +572,7 @@ svbool_t test_svcmplt_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmplt_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmplt.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -611,8 +585,7 @@ svbool_t test_svcmplt_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmplt_wide_n_s32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmplt.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -620,8 +593,7 @@ svbool_t test_svcmplt_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmplt_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmplt.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -633,15 +605,13 @@ svbool_t test_svcmplt_wide_n_s32(svbool_t pg, svint32_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svcmplt_wide_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmplo.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svcmplt_wide_n_u8u10__SVBool_tu11__SVUint8_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmplo.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -653,8 +623,7 @@ svbool_t test_svcmplt_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svcmplt_wide_n_u16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmplo.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -662,8 +631,7 @@ svbool_t test_svcmplt_wide_n_u8(svbool_t pg, svuint8_t op1, uint64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmplt_wide_n_u16u10__SVBool_tu12__SVUint16_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmplo.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -676,8 +644,7 @@ svbool_t test_svcmplt_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svcmplt_wide_n_u32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmplo.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -685,8 +652,7 @@ svbool_t test_svcmplt_wide_n_u16(svbool_t pg, svuint16_t op1, uint64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmplt_wide_n_u32u10__SVBool_tu12__SVUint32_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmplo.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpne.c

@@ -161,8 +161,7 @@ svbool_t test_svcmpne_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CHECK-LABEL: @test_svcmpne_n_s64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -170,8 +169,7 @@ svbool_t test_svcmpne_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpne_n_s64u10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -184,8 +182,7 @@ svbool_t test_svcmpne_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmpne_n_u64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -193,8 +190,7 @@ svbool_t test_svcmpne_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpne_n_u64u10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -259,15 +255,13 @@ svbool_t test_svcmpne_wide_s32(svbool_t pg, svint32_t op1, svint64_t op2)
 
 // CHECK-LABEL: @test_svcmpne_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svcmpne_n_s8u10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -279,8 +273,7 @@ svbool_t test_svcmpne_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svcmpne_n_s16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -288,8 +281,7 @@ svbool_t test_svcmpne_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpne_n_s16u10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -302,8 +294,7 @@ svbool_t test_svcmpne_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svcmpne_n_s32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -311,8 +302,7 @@ svbool_t test_svcmpne_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpne_n_s32u10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -324,15 +314,13 @@ svbool_t test_svcmpne_n_s32(svbool_t pg, svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svcmpne_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svcmpne_n_u8u10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -344,8 +332,7 @@ svbool_t test_svcmpne_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svcmpne_n_u16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -353,8 +340,7 @@ svbool_t test_svcmpne_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpne_n_u16u10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -367,8 +353,7 @@ svbool_t test_svcmpne_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svcmpne_n_u32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -376,8 +361,7 @@ svbool_t test_svcmpne_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpne_n_u32u10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -447,8 +431,7 @@ svbool_t test_svcmpne_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svcmpne_n_f16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpne.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -456,8 +439,7 @@ svbool_t test_svcmpne_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpne_n_f16u10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpne.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -470,8 +452,7 @@ svbool_t test_svcmpne_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svcmpne_n_f32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpne.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -479,8 +460,7 @@ svbool_t test_svcmpne_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpne_n_f32u10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpne.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -493,8 +473,7 @@ svbool_t test_svcmpne_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svcmpne_n_f64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpne.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -502,8 +481,7 @@ svbool_t test_svcmpne_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpne_n_f64u10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpne.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -515,15 +493,13 @@ svbool_t test_svcmpne_n_f64(svbool_t pg, svfloat64_t op1, float64_t op2)
 
 // CHECK-LABEL: @test_svcmpne_wide_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svcmpne_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -535,8 +511,7 @@ svbool_t test_svcmpne_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmpne_wide_n_s16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -544,8 +519,7 @@ svbool_t test_svcmpne_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmpne_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -558,8 +532,7 @@ svbool_t test_svcmpne_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
 // CHECK-LABEL: @test_svcmpne_wide_n_s32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -567,8 +540,7 @@ svbool_t test_svcmpne_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svcmpne_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpuo.c

@@ -74,8 +74,7 @@ svbool_t test_svcmpuo_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svcmpuo_n_f16(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpuo.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -83,8 +82,7 @@ svbool_t test_svcmpuo_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpuo_n_f16u10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpuo.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -97,8 +95,7 @@ svbool_t test_svcmpuo_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svcmpuo_n_f32(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpuo.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -106,8 +103,7 @@ svbool_t test_svcmpuo_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpuo_n_f32u10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpuo.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -120,8 +116,7 @@ svbool_t test_svcmpuo_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svcmpuo_n_f64(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpuo.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]
@@ -129,8 +124,7 @@ svbool_t test_svcmpuo_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svcmpuo_n_f64u10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpuo.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP3]]

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_div.c

@@ -229,8 +229,7 @@ svuint64_t test_svdiv_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CHECK-LABEL: @test_svdiv_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -238,8 +237,7 @@ svuint64_t test_svdiv_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -252,8 +250,7 @@ svint32_t test_svdiv_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svdiv_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -261,8 +258,7 @@ svint32_t test_svdiv_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -275,8 +271,7 @@ svint64_t test_svdiv_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 // CHECK-LABEL: @test_svdiv_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -284,8 +279,7 @@ svint64_t test_svdiv_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -298,8 +292,7 @@ svuint32_t test_svdiv_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svdiv_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -307,8 +300,7 @@ svuint32_t test_svdiv_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -321,16 +313,14 @@ svuint64_t test_svdiv_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svdiv_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -342,16 +332,14 @@ svint32_t test_svdiv_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svdiv_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -363,16 +351,14 @@ svint64_t test_svdiv_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 // CHECK-LABEL: @test_svdiv_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -384,16 +370,14 @@ svuint32_t test_svdiv_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svdiv_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -405,16 +389,14 @@ svuint64_t test_svdiv_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svdiv_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -426,16 +408,14 @@ svint32_t test_svdiv_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svdiv_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -447,16 +427,14 @@ svint64_t test_svdiv_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 // CHECK-LABEL: @test_svdiv_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udiv.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -468,16 +446,14 @@ svuint32_t test_svdiv_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svdiv_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udiv.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -648,8 +624,7 @@ svfloat64_t test_svdiv_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svdiv_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -657,8 +632,7 @@ svfloat64_t test_svdiv_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -671,8 +645,7 @@ svfloat16_t test_svdiv_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svdiv_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -680,8 +653,7 @@ svfloat16_t test_svdiv_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -694,8 +666,7 @@ svfloat32_t test_svdiv_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svdiv_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -703,8 +674,7 @@ svfloat32_t test_svdiv_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_f64_zu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -717,16 +687,14 @@ svfloat64_t test_svdiv_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svdiv_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -738,16 +706,14 @@ svfloat16_t test_svdiv_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svdiv_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -759,16 +725,14 @@ svfloat32_t test_svdiv_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svdiv_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_f64_mu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -780,16 +744,14 @@ svfloat64_t test_svdiv_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svdiv_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdiv.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -801,16 +763,14 @@ svfloat16_t test_svdiv_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svdiv_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdiv.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -822,16 +782,14 @@ svfloat32_t test_svdiv_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svdiv_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdiv_n_f64_xu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdiv.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_divr.c

@@ -229,8 +229,7 @@ svuint64_t test_svdivr_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CHECK-LABEL: @test_svdivr_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -238,8 +237,7 @@ svuint64_t test_svdivr_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -252,8 +250,7 @@ svint32_t test_svdivr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svdivr_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -261,8 +258,7 @@ svint32_t test_svdivr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -275,8 +271,7 @@ svint64_t test_svdivr_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 // CHECK-LABEL: @test_svdivr_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -284,8 +279,7 @@ svint64_t test_svdivr_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -298,8 +292,7 @@ svuint32_t test_svdivr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svdivr_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -307,8 +300,7 @@ svuint32_t test_svdivr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -321,16 +313,14 @@ svuint64_t test_svdivr_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svdivr_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -342,16 +332,14 @@ svint32_t test_svdivr_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svdivr_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -363,16 +351,14 @@ svint64_t test_svdivr_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 // CHECK-LABEL: @test_svdivr_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -384,16 +370,14 @@ svuint32_t test_svdivr_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svdivr_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -405,16 +389,14 @@ svuint64_t test_svdivr_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svdivr_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -426,16 +408,14 @@ svint32_t test_svdivr_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svdivr_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -447,16 +427,14 @@ svint64_t test_svdivr_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 // CHECK-LABEL: @test_svdivr_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udivr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -468,16 +446,14 @@ svuint32_t test_svdivr_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svdivr_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udivr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -648,8 +624,7 @@ svfloat64_t test_svdivr_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svdivr_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -657,8 +632,7 @@ svfloat64_t test_svdivr_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -671,8 +645,7 @@ svfloat16_t test_svdivr_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svdivr_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -680,8 +653,7 @@ svfloat16_t test_svdivr_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -694,8 +666,7 @@ svfloat32_t test_svdivr_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svdivr_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -703,8 +674,7 @@ svfloat32_t test_svdivr_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_f64_zu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -717,16 +687,14 @@ svfloat64_t test_svdivr_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svdivr_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -738,16 +706,14 @@ svfloat16_t test_svdivr_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svdivr_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -759,16 +725,14 @@ svfloat32_t test_svdivr_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svdivr_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_f64_mu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -780,16 +744,14 @@ svfloat64_t test_svdivr_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svdivr_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fdivr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -801,16 +763,14 @@ svfloat16_t test_svdivr_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svdivr_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fdivr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -822,16 +782,14 @@ svfloat32_t test_svdivr_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svdivr_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdivr_n_f64_xu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fdivr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dot.c

@@ -76,15 +76,13 @@ svuint64_t test_svdot_u64(svuint64_t op1, svuint16_t op2, svuint16_t op3)
 
 // CHECK-LABEL: @test_svdot_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdot.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdot_n_s32u11__SVInt32_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sdot.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -95,15 +93,13 @@ svint32_t test_svdot_n_s32(svint32_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svdot_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdot.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdot_n_s64u11__SVInt64_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sdot.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -114,15 +110,13 @@ svint64_t test_svdot_n_s64(svint64_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svdot_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udot.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdot_n_u32u12__SVUint32_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.udot.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -133,15 +127,13 @@ svuint32_t test_svdot_n_u32(svuint32_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_svdot_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udot.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdot_n_u64u12__SVUint64_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.udot.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dup-bfloat.c

@@ -17,14 +17,12 @@
 
 // CHECK-LABEL: @test_svdup_n_bf16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svdup_n_bf16u6__bf16(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x bfloat> poison, bfloat [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.dup.x.nxv8bf16(bfloat [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x bfloat> [[DOTSPLATINSERT]], <vscale x 8 x bfloat> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    ret <vscale x 8 x bfloat> [[TMP0]]
 //
 svbfloat16_t test_svdup_n_bf16(bfloat16_t op) {
@@ -85,15 +83,13 @@ svbfloat16_t test_svdup_n_bf16_x(svbool_t pg, bfloat16_t op) {
 
 // CHECK-LABEL: @test_svdup_lane_bf16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.tbl.nxv8bf16(<vscale x 8 x bfloat> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x bfloat> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svdup_lane_bf16u14__SVBFloat16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x bfloat> @llvm.aarch64.sve.tbl.nxv8bf16(<vscale x 8 x bfloat> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x bfloat> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dup.c

@@ -16,14 +16,12 @@
 
 // CHECK-LABEL: @test_svdup_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
 //
 // CPP-CHECK-LABEL: @_Z15test_svdup_n_s8a(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
 //
 svint8_t test_svdup_n_s8(int8_t op)
@@ -33,14 +31,12 @@ svint8_t test_svdup_n_s8(int8_t op)
 
 // CHECK-LABEL: @test_svdup_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdup_n_s16s(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
 //
 svint16_t test_svdup_n_s16(int16_t op)
@@ -50,14 +46,12 @@ svint16_t test_svdup_n_s16(int16_t op)
 
 // CHECK-LABEL: @test_svdup_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP0]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdup_n_s32i(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP0]]
 //
 svint32_t test_svdup_n_s32(int32_t op)
@@ -67,14 +61,12 @@ svint32_t test_svdup_n_s32(int32_t op)
 
 // CHECK-LABEL: @test_svdup_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP0]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdup_n_s64l(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP0]]
 //
 svint64_t test_svdup_n_s64(int64_t op)
@@ -84,14 +76,12 @@ svint64_t test_svdup_n_s64(int64_t op)
 
 // CHECK-LABEL: @test_svdup_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
 //
 // CPP-CHECK-LABEL: @_Z15test_svdup_n_u8h(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP0]]
 //
 svuint8_t test_svdup_n_u8(uint8_t op)
@@ -101,14 +91,12 @@ svuint8_t test_svdup_n_u8(uint8_t op)
 
 // CHECK-LABEL: @test_svdup_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdup_n_u16t(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP0]]
 //
 svuint16_t test_svdup_n_u16(uint16_t op)
@@ -118,14 +106,12 @@ svuint16_t test_svdup_n_u16(uint16_t op)
 
 // CHECK-LABEL: @test_svdup_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP0]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdup_n_u32j(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP0]]
 //
 svuint32_t test_svdup_n_u32(uint32_t op)
@@ -135,14 +121,12 @@ svuint32_t test_svdup_n_u32(uint32_t op)
 
 // CHECK-LABEL: @test_svdup_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP0]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdup_n_u64m(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP0]]
 //
 svuint64_t test_svdup_n_u64(uint64_t op)
@@ -152,14 +136,12 @@ svuint64_t test_svdup_n_u64(uint64_t op)
 
 // CHECK-LABEL: @test_svdup_n_f16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP0]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdup_n_f16Dh(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP0]]
 //
 svfloat16_t test_svdup_n_f16(float16_t op)
@@ -169,14 +151,12 @@ svfloat16_t test_svdup_n_f16(float16_t op)
 
 // CHECK-LABEL: @test_svdup_n_f32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdup_n_f32f(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP0]]
 //
 svfloat32_t test_svdup_n_f32(float32_t op)
@@ -186,14 +166,12 @@ svfloat32_t test_svdup_n_f32(float32_t op)
 
 // CHECK-LABEL: @test_svdup_n_f64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdup_n_f64d(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP0]]
 //
 svfloat64_t test_svdup_n_f64(float64_t op)
@@ -752,15 +730,13 @@ svfloat64_t test_svdup_n_f64_x(svbool_t pg, float64_t op)
 
 // CHECK-LABEL: @test_svdup_lane_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[INDEX:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[INDEX:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.tbl.nxv16i8(<vscale x 16 x i8> [[DATA:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdup_lane_s8u10__SVInt8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[INDEX:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.tbl.nxv16i8(<vscale x 16 x i8> [[DATA:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -771,15 +747,13 @@ svint8_t test_svdup_lane_s8(svint8_t data, uint8_t index)
 
 // CHECK-LABEL: @test_svdup_lane_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.tbl.nxv8i16(<vscale x 8 x i16> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdup_lane_s16u11__SVInt16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.tbl.nxv8i16(<vscale x 8 x i16> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -790,15 +764,13 @@ svint16_t test_svdup_lane_s16(svint16_t data, uint16_t index)
 
 // CHECK-LABEL: @test_svdup_lane_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.tbl.nxv4i32(<vscale x 4 x i32> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdup_lane_s32u11__SVInt32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.tbl.nxv4i32(<vscale x 4 x i32> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -809,15 +781,13 @@ svint32_t test_svdup_lane_s32(svint32_t data, uint32_t index)
 
 // CHECK-LABEL: @test_svdup_lane_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.tbl.nxv2i64(<vscale x 2 x i64> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdup_lane_s64u11__SVInt64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.tbl.nxv2i64(<vscale x 2 x i64> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -828,15 +798,13 @@ svint64_t test_svdup_lane_s64(svint64_t data, uint64_t index)
 
 // CHECK-LABEL: @test_svdup_lane_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[INDEX:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[INDEX:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.tbl.nxv16i8(<vscale x 16 x i8> [[DATA:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svdup_lane_u8u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[INDEX:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.tbl.nxv16i8(<vscale x 16 x i8> [[DATA:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -847,15 +815,13 @@ svuint8_t test_svdup_lane_u8(svuint8_t data, uint8_t index)
 
 // CHECK-LABEL: @test_svdup_lane_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.tbl.nxv8i16(<vscale x 8 x i16> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdup_lane_u16u12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.tbl.nxv8i16(<vscale x 8 x i16> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -866,15 +832,13 @@ svuint16_t test_svdup_lane_u16(svuint16_t data, uint16_t index)
 
 // CHECK-LABEL: @test_svdup_lane_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.tbl.nxv4i32(<vscale x 4 x i32> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdup_lane_u32u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.tbl.nxv4i32(<vscale x 4 x i32> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -885,15 +849,13 @@ svuint32_t test_svdup_lane_u32(svuint32_t data, uint32_t index)
 
 // CHECK-LABEL: @test_svdup_lane_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.tbl.nxv2i64(<vscale x 2 x i64> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdup_lane_u64u12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.tbl.nxv2i64(<vscale x 2 x i64> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -904,15 +866,13 @@ svuint64_t test_svdup_lane_u64(svuint64_t data, uint64_t index)
 
 // CHECK-LABEL: @test_svdup_lane_f16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.tbl.nxv8f16(<vscale x 8 x half> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdup_lane_f16u13__SVFloat16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[INDEX:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.tbl.nxv8f16(<vscale x 8 x half> [[DATA:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP1]]
 //
@@ -923,15 +883,13 @@ svfloat16_t test_svdup_lane_f16(svfloat16_t data, uint16_t index)
 
 // CHECK-LABEL: @test_svdup_lane_f32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.tbl.nxv4f32(<vscale x 4 x float> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdup_lane_f32u13__SVFloat32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[INDEX:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.tbl.nxv4f32(<vscale x 4 x float> [[DATA:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //
@@ -942,15 +900,13 @@ svfloat32_t test_svdup_lane_f32(svfloat32_t data, uint32_t index)
 
 // CHECK-LABEL: @test_svdup_lane_f64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.tbl.nxv2f64(<vscale x 2 x double> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svdup_lane_f64u13__SVFloat64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[INDEX:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[INDEX:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.tbl.nxv2f64(<vscale x 2 x double> [[DATA:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP1]]
 //
@@ -961,14 +917,12 @@ svfloat64_t test_svdup_lane_f64(svfloat64_t data, uint64_t index)
 
 // CHECK-LABEL: @test_svdup_n_b8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i1> poison, i1 [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.dup.x.nxv16i1(i1 [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i1> [[DOTSPLATINSERT]], <vscale x 16 x i1> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
 //
 // CPP-CHECK-LABEL: @_Z15test_svdup_n_b8b(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i1> poison, i1 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.dup.x.nxv16i1(i1 [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i1> [[DOTSPLATINSERT]], <vscale x 16 x i1> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP0]]
 //
 svbool_t test_svdup_n_b8(bool op)
@@ -978,15 +932,13 @@ svbool_t test_svdup_n_b8(bool op)
 
 // CHECK-LABEL: @test_svdup_n_b16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i1> poison, i1 [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.dup.x.nxv8i1(i1 [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i1> [[DOTSPLATINSERT]], <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdup_n_b16b(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i1> poison, i1 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.dup.x.nxv8i1(i1 [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i1> [[DOTSPLATINSERT]], <vscale x 8 x i1> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -997,15 +949,13 @@ svbool_t test_svdup_n_b16(bool op)
 
 // CHECK-LABEL: @test_svdup_n_b32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i1> poison, i1 [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.dup.x.nxv4i1(i1 [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i1> [[DOTSPLATINSERT]], <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdup_n_b32b(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i1> poison, i1 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.dup.x.nxv4i1(i1 [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i1> [[DOTSPLATINSERT]], <vscale x 4 x i1> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
@@ -1016,15 +966,13 @@ svbool_t test_svdup_n_b32(bool op)
 
 // CHECK-LABEL: @test_svdup_n_b64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i1> poison, i1 [[OP:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.dup.x.nxv2i1(i1 [[OP:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i1> [[DOTSPLATINSERT]], <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdup_n_b64b(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i1> poison, i1 [[OP:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.dup.x.nxv2i1(i1 [[OP:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i1> [[DOTSPLATINSERT]], <vscale x 2 x i1> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_dupq.c

@@ -568,8 +568,9 @@ svfloat64_t test_svdupq_n_f64(float64_t x0, float64_t x1)
 // CHECK-NEXT:    [[TMP16:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
 // CHECK-NEXT:    [[TMP17:%.*]] = call <vscale x 16 x i8> @llvm.experimental.vector.insert.nxv16i8.v16i8(<vscale x 16 x i8> undef, <16 x i8> [[TMP15]], i64 0)
 // CHECK-NEXT:    [[TMP18:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dupq.lane.nxv16i8(<vscale x 16 x i8> [[TMP17]], i64 0)
-// CHECK-NEXT:    [[TMP19:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[TMP16]], <vscale x 16 x i8> [[TMP18]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CHECK-NEXT:    [[TMP19:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP19]]
+// CHECK-NEXT:    [[TMP20:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[TMP16]], <vscale x 16 x i8> [[TMP18]], <vscale x 2 x i64> [[TMP19]])
+// CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP20]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svdupq_n_b8bbbbbbbbbbbbbbbb(
 // CPP-CHECK-NEXT:  entry:
@@ -608,8 +609,9 @@ svfloat64_t test_svdupq_n_f64(float64_t x0, float64_t x1)
 // CPP-CHECK-NEXT:    [[TMP16:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.ptrue.nxv16i1(i32 31)
 // CPP-CHECK-NEXT:    [[TMP17:%.*]] = call <vscale x 16 x i8> @llvm.experimental.vector.insert.nxv16i8.v16i8(<vscale x 16 x i8> undef, <16 x i8> [[TMP15]], i64 0)
 // CPP-CHECK-NEXT:    [[TMP18:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dupq.lane.nxv16i8(<vscale x 16 x i8> [[TMP17]], i64 0)
-// CPP-CHECK-NEXT:    [[TMP19:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[TMP16]], <vscale x 16 x i8> [[TMP18]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CPP-CHECK-NEXT:    [[TMP19:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP19]]
+// CPP-CHECK-NEXT:    [[TMP20:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[TMP16]], <vscale x 16 x i8> [[TMP18]], <vscale x 2 x i64> [[TMP19]])
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP20]]
 //
 svbool_t test_svdupq_n_b8(bool x0, bool x1, bool x2, bool x3,
                           bool x4, bool x5, bool x6, bool x7,
@@ -641,9 +643,10 @@ svbool_t test_svdupq_n_b8(bool x0, bool x1, bool x2, bool x3,
 // CHECK-NEXT:    [[TMP16:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
 // CHECK-NEXT:    [[TMP17:%.*]] = call <vscale x 8 x i16> @llvm.experimental.vector.insert.nxv8i16.v8i16(<vscale x 8 x i16> undef, <8 x i16> [[TMP15]], i64 0)
 // CHECK-NEXT:    [[TMP18:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dupq.lane.nxv8i16(<vscale x 8 x i16> [[TMP17]], i64 0)
-// CHECK-NEXT:    [[TMP19:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP16]], <vscale x 8 x i16> [[TMP18]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CHECK-NEXT:    [[TMP19:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CHECK-NEXT:    [[TMP20:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP19]])
+// CHECK-NEXT:    [[TMP20:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP16]], <vscale x 8 x i16> [[TMP18]], <vscale x 2 x i64> [[TMP19]])
-// CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP20]]
+// CHECK-NEXT:    [[TMP21:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP20]])
+// CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP21]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svdupq_n_b16bbbbbbbb(
 // CPP-CHECK-NEXT:  entry:
@@ -666,9 +669,10 @@ svbool_t test_svdupq_n_b8(bool x0, bool x1, bool x2, bool x3,
 // CPP-CHECK-NEXT:    [[TMP16:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.ptrue.nxv8i1(i32 31)
 // CPP-CHECK-NEXT:    [[TMP17:%.*]] = call <vscale x 8 x i16> @llvm.experimental.vector.insert.nxv8i16.v8i16(<vscale x 8 x i16> undef, <8 x i16> [[TMP15]], i64 0)
 // CPP-CHECK-NEXT:    [[TMP18:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dupq.lane.nxv8i16(<vscale x 8 x i16> [[TMP17]], i64 0)
-// CPP-CHECK-NEXT:    [[TMP19:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP16]], <vscale x 8 x i16> [[TMP18]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CPP-CHECK-NEXT:    [[TMP19:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CPP-CHECK-NEXT:    [[TMP20:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP19]])
+// CPP-CHECK-NEXT:    [[TMP20:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP16]], <vscale x 8 x i16> [[TMP18]], <vscale x 2 x i64> [[TMP19]])
-// CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP20]]
+// CPP-CHECK-NEXT:    [[TMP21:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP20]])
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP21]]
 //
 svbool_t test_svdupq_n_b16(bool x0, bool x1, bool x2, bool x3,
                            bool x4, bool x5, bool x6, bool x7)
@@ -690,9 +694,10 @@ svbool_t test_svdupq_n_b16(bool x0, bool x1, bool x2, bool x3,
 // CHECK-NEXT:    [[TMP8:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
 // CHECK-NEXT:    [[TMP9:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.insert.nxv4i32.v4i32(<vscale x 4 x i32> undef, <4 x i32> [[TMP7]], i64 0)
 // CHECK-NEXT:    [[TMP10:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dupq.lane.nxv4i32(<vscale x 4 x i32> [[TMP9]], i64 0)
-// CHECK-NEXT:    [[TMP11:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[TMP10]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CHECK-NEXT:    [[TMP11:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CHECK-NEXT:    [[TMP12:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP11]])
+// CHECK-NEXT:    [[TMP12:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[TMP10]], <vscale x 2 x i64> [[TMP11]])
-// CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP12]]
+// CHECK-NEXT:    [[TMP13:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP12]])
+// CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP13]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svdupq_n_b32bbbb(
 // CPP-CHECK-NEXT:  entry:
@@ -707,9 +712,10 @@ svbool_t test_svdupq_n_b16(bool x0, bool x1, bool x2, bool x3,
 // CPP-CHECK-NEXT:    [[TMP8:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.ptrue.nxv4i1(i32 31)
 // CPP-CHECK-NEXT:    [[TMP9:%.*]] = call <vscale x 4 x i32> @llvm.experimental.vector.insert.nxv4i32.v4i32(<vscale x 4 x i32> undef, <4 x i32> [[TMP7]], i64 0)
 // CPP-CHECK-NEXT:    [[TMP10:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dupq.lane.nxv4i32(<vscale x 4 x i32> [[TMP9]], i64 0)
-// CPP-CHECK-NEXT:    [[TMP11:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[TMP10]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CPP-CHECK-NEXT:    [[TMP11:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CPP-CHECK-NEXT:    [[TMP12:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP11]])
+// CPP-CHECK-NEXT:    [[TMP12:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP8]], <vscale x 4 x i32> [[TMP10]], <vscale x 2 x i64> [[TMP11]])
-// CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP12]]
+// CPP-CHECK-NEXT:    [[TMP13:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP12]])
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP13]]
 //
 svbool_t test_svdupq_n_b32(bool x0, bool x1, bool x2, bool x3)
 {
@@ -726,9 +732,10 @@ svbool_t test_svdupq_n_b32(bool x0, bool x1, bool x2, bool x3)
 // CHECK-NEXT:    [[TMP4:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
 // CHECK-NEXT:    [[TMP5:%.*]] = call <vscale x 2 x i64> @llvm.experimental.vector.insert.nxv2i64.v2i64(<vscale x 2 x i64> undef, <2 x i64> [[TMP3]], i64 0)
 // CHECK-NEXT:    [[TMP6:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dupq.lane.nxv2i64(<vscale x 2 x i64> [[TMP5]], i64 0)
-// CHECK-NEXT:    [[TMP7:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP4]], <vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CHECK-NEXT:    [[TMP7:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CHECK-NEXT:    [[TMP8:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP7]])
+// CHECK-NEXT:    [[TMP8:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP4]], <vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]])
-// CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP8]]
+// CHECK-NEXT:    [[TMP9:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP8]])
+// CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP9]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svdupq_n_b64bb(
 // CPP-CHECK-NEXT:  entry:
@@ -739,9 +746,10 @@ svbool_t test_svdupq_n_b32(bool x0, bool x1, bool x2, bool x3)
 // CPP-CHECK-NEXT:    [[TMP4:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.ptrue.nxv2i1(i32 31)
 // CPP-CHECK-NEXT:    [[TMP5:%.*]] = call <vscale x 2 x i64> @llvm.experimental.vector.insert.nxv2i64.v2i64(<vscale x 2 x i64> undef, <2 x i64> [[TMP3]], i64 0)
 // CPP-CHECK-NEXT:    [[TMP6:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dupq.lane.nxv2i64(<vscale x 2 x i64> [[TMP5]], i64 0)
-// CPP-CHECK-NEXT:    [[TMP7:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP4]], <vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> shufflevector (<vscale x 2 x i64> insertelement (<vscale x 2 x i64> poison, i64 0, i32 0), <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer))
+// CPP-CHECK-NEXT:    [[TMP7:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 0)
-// CPP-CHECK-NEXT:    [[TMP8:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP7]])
+// CPP-CHECK-NEXT:    [[TMP8:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP4]], <vscale x 2 x i64> [[TMP6]], <vscale x 2 x i64> [[TMP7]])
-// CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP8]]
+// CPP-CHECK-NEXT:    [[TMP9:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP8]])
+// CPP-CHECK-NEXT:    ret <vscale x 16 x i1> [[TMP9]]
 //
 svbool_t test_svdupq_n_b64(bool x0, bool x1)
 {

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_eor.c

@@ -428,16 +428,14 @@ svuint64_t test_sveor_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_sveor_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveor_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_sveor_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_sveor_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_sveor_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_sveor_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_sveor_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_sveor_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_sveor_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_sveor_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_sveor_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_sveor_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_sveor_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveor_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_sveor_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_sveor_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_sveor_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_sveor_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_sveor_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_sveor_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_sveor_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_sveor_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_sveor_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_sveor_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_sveor_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveor_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_sveor_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_sveor_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_sveor_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_sveor_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_sveor_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_sveor_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_sveor_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_sveor_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveor_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_sveor_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_sveor_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_sveor_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_sveor_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_sveor_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_sveor_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_sveor_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_sveor_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveor_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_sveor_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_sveor_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_sveor_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_sveor_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_sveor_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_sveor_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_sveor_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_sveor_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveor_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_sveor_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_sveor_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_sveor_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_sveor_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_sveor_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_sveor_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveor_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_lsl.c

@@ -734,15 +734,13 @@ svuint32_t test_svlsl_wide_u32_x(svbool_t pg, svuint32_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svlsl_wide_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svlsl_wide_n_s8_mu10__SVBool_tu10__SVInt8_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -754,16 +752,14 @@ svint8_t test_svlsl_wide_n_s8_m(svbool_t pg, svint8_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svlsl_wide_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s16_mu10__SVBool_tu11__SVInt16_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -775,16 +771,14 @@ svint16_t test_svlsl_wide_n_s16_m(svbool_t pg, svint16_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svlsl_wide_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s32_mu10__SVBool_tu11__SVInt32_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -795,16 +789,14 @@ svint32_t test_svlsl_wide_n_s32_m(svbool_t pg, svint32_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svlsl_wide_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svlsl_wide_n_s8_zu10__SVBool_tu10__SVInt8_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -817,8 +809,7 @@ svint8_t test_svlsl_wide_n_s8_z(svbool_t pg, svint8_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svlsl_wide_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -826,8 +817,7 @@ svint8_t test_svlsl_wide_n_s8_z(svbool_t pg, svint8_t op1, uint64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s16_zu10__SVBool_tu11__SVInt16_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -840,8 +830,7 @@ svint16_t test_svlsl_wide_n_s16_z(svbool_t pg, svint16_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svlsl_wide_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -849,8 +838,7 @@ svint16_t test_svlsl_wide_n_s16_z(svbool_t pg, svint16_t op1, uint64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s32_zu10__SVBool_tu11__SVInt32_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -862,15 +850,13 @@ svint32_t test_svlsl_wide_n_s32_z(svbool_t pg, svint32_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svlsl_wide_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svlsl_wide_n_s8_xu10__SVBool_tu10__SVInt8_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsl.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -882,16 +868,14 @@ svint8_t test_svlsl_wide_n_s8_x(svbool_t pg, svint8_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svlsl_wide_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s16_xu10__SVBool_tu11__SVInt16_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsl.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -903,16 +887,14 @@ svint16_t test_svlsl_wide_n_s16_x(svbool_t pg, svint16_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svlsl_wide_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z23test_svlsl_wide_n_s32_xu10__SVBool_tu11__SVInt32_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsl.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_lsr.c

@@ -375,15 +375,13 @@ svuint32_t test_svlsr_wide_u32_x(svbool_t pg, svuint32_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svlsr_wide_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svlsr_wide_n_u8_mu10__SVBool_tu11__SVUint8_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -395,16 +393,14 @@ svuint8_t test_svlsr_wide_n_u8_m(svbool_t pg, svuint8_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svlsr_wide_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u16_mu10__SVBool_tu12__SVUint16_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -416,16 +412,14 @@ svuint16_t test_svlsr_wide_n_u16_m(svbool_t pg, svuint16_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svlsr_wide_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u32_mu10__SVBool_tu12__SVUint32_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -436,16 +430,14 @@ svuint32_t test_svlsr_wide_n_u32_m(svbool_t pg, svuint32_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svlsr_wide_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svlsr_wide_n_u8_zu10__SVBool_tu11__SVUint8_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -458,8 +450,7 @@ svuint8_t test_svlsr_wide_n_u8_z(svbool_t pg, svuint8_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svlsr_wide_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -467,8 +458,7 @@ svuint8_t test_svlsr_wide_n_u8_z(svbool_t pg, svuint8_t op1, uint64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u16_zu10__SVBool_tu12__SVUint16_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -481,8 +471,7 @@ svuint16_t test_svlsr_wide_n_u16_z(svbool_t pg, svuint16_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svlsr_wide_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -490,8 +479,7 @@ svuint16_t test_svlsr_wide_n_u16_z(svbool_t pg, svuint16_t op1, uint64_t op2)
 // CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u32_zu10__SVBool_tu12__SVUint32_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -503,15 +491,13 @@ svuint32_t test_svlsr_wide_n_u32_z(svbool_t pg, svuint32_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svlsr_wide_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z22test_svlsr_wide_n_u8_xu10__SVBool_tu11__SVUint8_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.lsr.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -523,16 +509,14 @@ svuint8_t test_svlsr_wide_n_u8_x(svbool_t pg, svuint8_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svlsr_wide_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u16_xu10__SVBool_tu12__SVUint16_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.lsr.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -544,16 +528,14 @@ svuint16_t test_svlsr_wide_n_u16_x(svbool_t pg, svuint16_t op1, uint64_t op2)
 // CHECK-LABEL: @test_svlsr_wide_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z23test_svlsr_wide_n_u32_xu10__SVBool_tu12__SVUint32_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.lsr.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mad.c

@@ -428,16 +428,14 @@ svuint64_t test_svmad_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2, svuint6
 
 // CHECK-LABEL: @test_svmad_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmad_n_s8_zu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_svmad_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CHECK-LABEL: @test_svmad_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_svmad_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_s16_zu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_svmad_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CHECK-LABEL: @test_svmad_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_svmad_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_s32_zu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_svmad_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CHECK-LABEL: @test_svmad_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_svmad_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_s64_zu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_svmad_n_s64_z(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
 
 // CHECK-LABEL: @test_svmad_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmad_n_u8_zu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_svmad_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CHECK-LABEL: @test_svmad_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_svmad_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_u16_zu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_svmad_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CHECK-LABEL: @test_svmad_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_svmad_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_u32_zu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_svmad_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CHECK-LABEL: @test_svmad_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_svmad_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_u64_zu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_svmad_n_u64_z(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
 
 // CHECK-LABEL: @test_svmad_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmad_n_s8_mu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_svmad_n_s8_m(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CHECK-LABEL: @test_svmad_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_s16_mu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_svmad_n_s16_m(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CHECK-LABEL: @test_svmad_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_s32_mu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_svmad_n_s32_m(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CHECK-LABEL: @test_svmad_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_s64_mu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_svmad_n_s64_m(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
 
 // CHECK-LABEL: @test_svmad_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmad_n_u8_mu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_svmad_n_u8_m(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CHECK-LABEL: @test_svmad_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_u16_mu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_svmad_n_u16_m(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CHECK-LABEL: @test_svmad_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_u32_mu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_svmad_n_u32_m(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CHECK-LABEL: @test_svmad_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_u64_mu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_svmad_n_u64_m(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
 
 // CHECK-LABEL: @test_svmad_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmad_n_s8_xu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_svmad_n_s8_x(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CHECK-LABEL: @test_svmad_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_s16_xu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_svmad_n_s16_x(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CHECK-LABEL: @test_svmad_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_s32_xu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_svmad_n_s32_x(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CHECK-LABEL: @test_svmad_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_s64_xu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_svmad_n_s64_x(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
 
 // CHECK-LABEL: @test_svmad_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmad_n_u8_xu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mad.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_svmad_n_u8_x(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CHECK-LABEL: @test_svmad_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_u16_xu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mad.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_svmad_n_u16_x(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CHECK-LABEL: @test_svmad_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_u32_xu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mad.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_svmad_n_u32_x(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CHECK-LABEL: @test_svmad_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_u64_xu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mad.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -1096,8 +1048,7 @@ svfloat64_t test_svmad_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
 // CHECK-LABEL: @test_svmad_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1105,8 +1056,7 @@ svfloat64_t test_svmad_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1119,8 +1069,7 @@ svfloat16_t test_svmad_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CHECK-LABEL: @test_svmad_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1128,8 +1077,7 @@ svfloat16_t test_svmad_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1142,8 +1090,7 @@ svfloat32_t test_svmad_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CHECK-LABEL: @test_svmad_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1151,8 +1098,7 @@ svfloat32_t test_svmad_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1165,16 +1111,14 @@ svfloat64_t test_svmad_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
 // CHECK-LABEL: @test_svmad_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1186,16 +1130,14 @@ svfloat16_t test_svmad_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CHECK-LABEL: @test_svmad_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1207,16 +1149,14 @@ svfloat32_t test_svmad_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CHECK-LABEL: @test_svmad_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -1228,16 +1168,14 @@ svfloat64_t test_svmad_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
 // CHECK-LABEL: @test_svmad_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1249,16 +1187,14 @@ svfloat16_t test_svmad_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CHECK-LABEL: @test_svmad_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1270,16 +1206,14 @@ svfloat32_t test_svmad_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CHECK-LABEL: @test_svmad_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmad_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_max.c

@@ -428,16 +428,14 @@ svuint64_t test_svmax_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svmax_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmax_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_svmax_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svmax_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_svmax_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_svmax_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svmax_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_svmax_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_svmax_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svmax_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_svmax_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_svmax_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svmax_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmax_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_svmax_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svmax_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_svmax_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_svmax_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svmax_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_svmax_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_svmax_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svmax_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_svmax_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_svmax_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svmax_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmax_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_svmax_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svmax_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_svmax_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svmax_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_svmax_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svmax_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_svmax_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svmax_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmax_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_svmax_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svmax_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_svmax_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svmax_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_svmax_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svmax_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_svmax_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svmax_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmax_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_svmax_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svmax_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_svmax_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svmax_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_svmax_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svmax_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_svmax_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svmax_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmax_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umax.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_svmax_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svmax_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umax.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_svmax_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svmax_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umax.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_svmax_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svmax_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umax.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -1096,8 +1048,7 @@ svfloat64_t test_svmax_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svmax_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1105,8 +1056,7 @@ svfloat64_t test_svmax_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1119,8 +1069,7 @@ svfloat16_t test_svmax_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svmax_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1128,8 +1077,7 @@ svfloat16_t test_svmax_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1142,8 +1090,7 @@ svfloat32_t test_svmax_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svmax_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1151,8 +1098,7 @@ svfloat32_t test_svmax_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_f64_zu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1165,16 +1111,14 @@ svfloat64_t test_svmax_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svmax_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1186,16 +1130,14 @@ svfloat16_t test_svmax_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svmax_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1207,16 +1149,14 @@ svfloat32_t test_svmax_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svmax_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_f64_mu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -1228,16 +1168,14 @@ svfloat64_t test_svmax_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svmax_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmax.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1249,16 +1187,14 @@ svfloat16_t test_svmax_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svmax_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmax.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1270,16 +1206,14 @@ svfloat32_t test_svmax_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svmax_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmax_n_f64_xu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmax.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_maxnm.c

@@ -176,8 +176,7 @@ svfloat64_t test_svmaxnm_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svmaxnm_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -185,8 +184,7 @@ svfloat64_t test_svmaxnm_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -199,8 +197,7 @@ svfloat16_t test_svmaxnm_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svmaxnm_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -208,8 +205,7 @@ svfloat16_t test_svmaxnm_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -222,8 +218,7 @@ svfloat32_t test_svmaxnm_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svmaxnm_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -231,8 +226,7 @@ svfloat32_t test_svmaxnm_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f64_zu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -245,16 +239,14 @@ svfloat64_t test_svmaxnm_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svmaxnm_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -266,16 +258,14 @@ svfloat16_t test_svmaxnm_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svmaxnm_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -287,16 +277,14 @@ svfloat32_t test_svmaxnm_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svmaxnm_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f64_mu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -308,16 +296,14 @@ svfloat64_t test_svmaxnm_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svmaxnm_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmaxnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -329,16 +315,14 @@ svfloat16_t test_svmaxnm_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svmaxnm_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmaxnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -350,16 +334,14 @@ svfloat32_t test_svmaxnm_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svmaxnm_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svmaxnm_n_f64_xu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmaxnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_min.c

@@ -428,16 +428,14 @@ svuint64_t test_svmin_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svmin_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmin_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_svmin_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svmin_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_svmin_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_svmin_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svmin_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_svmin_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_svmin_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svmin_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_svmin_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_svmin_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svmin_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmin_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_svmin_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svmin_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_svmin_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_svmin_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svmin_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_svmin_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_svmin_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svmin_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_svmin_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_svmin_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svmin_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmin_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_svmin_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svmin_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_svmin_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svmin_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_svmin_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svmin_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_svmin_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svmin_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmin_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_svmin_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svmin_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_svmin_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svmin_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_svmin_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svmin_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_svmin_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svmin_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmin_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_svmin_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svmin_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_svmin_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svmin_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_svmin_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svmin_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_svmin_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svmin_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmin_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umin.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_svmin_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svmin_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umin.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_svmin_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svmin_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umin.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_svmin_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svmin_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umin.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -1096,8 +1048,7 @@ svfloat64_t test_svmin_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svmin_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1105,8 +1056,7 @@ svfloat64_t test_svmin_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1119,8 +1069,7 @@ svfloat16_t test_svmin_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svmin_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1128,8 +1077,7 @@ svfloat16_t test_svmin_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1142,8 +1090,7 @@ svfloat32_t test_svmin_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svmin_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1151,8 +1098,7 @@ svfloat32_t test_svmin_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_f64_zu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1165,16 +1111,14 @@ svfloat64_t test_svmin_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svmin_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1186,16 +1130,14 @@ svfloat16_t test_svmin_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svmin_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1207,16 +1149,14 @@ svfloat32_t test_svmin_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svmin_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_f64_mu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -1228,16 +1168,14 @@ svfloat64_t test_svmin_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svmin_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmin.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1249,16 +1187,14 @@ svfloat16_t test_svmin_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svmin_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmin.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1270,16 +1206,14 @@ svfloat32_t test_svmin_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svmin_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmin_n_f64_xu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmin.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_minnm.c

@@ -176,8 +176,7 @@ svfloat64_t test_svminnm_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svminnm_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -185,8 +184,7 @@ svfloat64_t test_svminnm_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z20test_svminnm_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -199,8 +197,7 @@ svfloat16_t test_svminnm_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svminnm_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -208,8 +205,7 @@ svfloat16_t test_svminnm_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z20test_svminnm_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -222,8 +218,7 @@ svfloat32_t test_svminnm_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svminnm_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -231,8 +226,7 @@ svfloat32_t test_svminnm_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z20test_svminnm_n_f64_zu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -245,16 +239,14 @@ svfloat64_t test_svminnm_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svminnm_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svminnm_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -266,16 +258,14 @@ svfloat16_t test_svminnm_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svminnm_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svminnm_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -287,16 +277,14 @@ svfloat32_t test_svminnm_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svminnm_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svminnm_n_f64_mu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -308,16 +296,14 @@ svfloat64_t test_svminnm_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svminnm_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svminnm_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fminnm.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -329,16 +315,14 @@ svfloat16_t test_svminnm_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svminnm_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svminnm_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fminnm.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -350,16 +334,14 @@ svfloat32_t test_svminnm_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svminnm_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svminnm_n_f64_xu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fminnm.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mla.c

@@ -428,16 +428,14 @@ svuint64_t test_svmla_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2, svuint6
 
 // CHECK-LABEL: @test_svmla_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmla_n_s8_zu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_svmla_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CHECK-LABEL: @test_svmla_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_svmla_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_s16_zu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_svmla_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CHECK-LABEL: @test_svmla_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_svmla_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_s32_zu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_svmla_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CHECK-LABEL: @test_svmla_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_svmla_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_s64_zu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_svmla_n_s64_z(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
 
 // CHECK-LABEL: @test_svmla_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmla_n_u8_zu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_svmla_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CHECK-LABEL: @test_svmla_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_svmla_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_u16_zu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_svmla_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CHECK-LABEL: @test_svmla_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_svmla_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_u32_zu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_svmla_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CHECK-LABEL: @test_svmla_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_svmla_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_u64_zu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_svmla_n_u64_z(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
 
 // CHECK-LABEL: @test_svmla_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmla_n_s8_mu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_svmla_n_s8_m(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CHECK-LABEL: @test_svmla_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_s16_mu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_svmla_n_s16_m(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CHECK-LABEL: @test_svmla_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_s32_mu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_svmla_n_s32_m(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CHECK-LABEL: @test_svmla_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_s64_mu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_svmla_n_s64_m(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
 
 // CHECK-LABEL: @test_svmla_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmla_n_u8_mu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_svmla_n_u8_m(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CHECK-LABEL: @test_svmla_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_u16_mu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_svmla_n_u16_m(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CHECK-LABEL: @test_svmla_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_u32_mu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_svmla_n_u32_m(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CHECK-LABEL: @test_svmla_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_u64_mu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_svmla_n_u64_m(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
 
 // CHECK-LABEL: @test_svmla_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmla_n_s8_xu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_svmla_n_s8_x(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CHECK-LABEL: @test_svmla_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_s16_xu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_svmla_n_s16_x(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CHECK-LABEL: @test_svmla_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_s32_xu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_svmla_n_s32_x(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CHECK-LABEL: @test_svmla_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_s64_xu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_svmla_n_s64_x(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
 
 // CHECK-LABEL: @test_svmla_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmla_n_u8_xu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mla.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_svmla_n_u8_x(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CHECK-LABEL: @test_svmla_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_u16_xu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mla.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_svmla_n_u16_x(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CHECK-LABEL: @test_svmla_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_u32_xu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mla.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_svmla_n_u32_x(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CHECK-LABEL: @test_svmla_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_u64_xu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mla.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -1096,8 +1048,7 @@ svfloat64_t test_svmla_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
 // CHECK-LABEL: @test_svmla_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1105,8 +1056,7 @@ svfloat64_t test_svmla_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1119,8 +1069,7 @@ svfloat16_t test_svmla_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CHECK-LABEL: @test_svmla_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1128,8 +1077,7 @@ svfloat16_t test_svmla_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1142,8 +1090,7 @@ svfloat32_t test_svmla_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CHECK-LABEL: @test_svmla_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1151,8 +1098,7 @@ svfloat32_t test_svmla_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1165,16 +1111,14 @@ svfloat64_t test_svmla_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
 // CHECK-LABEL: @test_svmla_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1186,16 +1130,14 @@ svfloat16_t test_svmla_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CHECK-LABEL: @test_svmla_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1207,16 +1149,14 @@ svfloat32_t test_svmla_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CHECK-LABEL: @test_svmla_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -1228,16 +1168,14 @@ svfloat64_t test_svmla_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
 // CHECK-LABEL: @test_svmla_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1249,16 +1187,14 @@ svfloat16_t test_svmla_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CHECK-LABEL: @test_svmla_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1270,16 +1206,14 @@ svfloat32_t test_svmla_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CHECK-LABEL: @test_svmla_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmla_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mls.c

@@ -428,16 +428,14 @@ svuint64_t test_svmls_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2, svuint6
 
 // CHECK-LABEL: @test_svmls_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmls_n_s8_zu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_svmls_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CHECK-LABEL: @test_svmls_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_svmls_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_s16_zu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_svmls_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CHECK-LABEL: @test_svmls_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_svmls_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_s32_zu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_svmls_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CHECK-LABEL: @test_svmls_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_svmls_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_s64_zu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_svmls_n_s64_z(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
 
 // CHECK-LABEL: @test_svmls_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmls_n_u8_zu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_svmls_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CHECK-LABEL: @test_svmls_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_svmls_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_u16_zu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_svmls_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CHECK-LABEL: @test_svmls_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_svmls_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_u32_zu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_svmls_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CHECK-LABEL: @test_svmls_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_svmls_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_u64_zu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_svmls_n_u64_z(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
 
 // CHECK-LABEL: @test_svmls_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmls_n_s8_mu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_svmls_n_s8_m(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CHECK-LABEL: @test_svmls_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_s16_mu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_svmls_n_s16_m(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CHECK-LABEL: @test_svmls_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_s32_mu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_svmls_n_s32_m(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CHECK-LABEL: @test_svmls_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_s64_mu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_svmls_n_s64_m(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
 
 // CHECK-LABEL: @test_svmls_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmls_n_u8_mu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_svmls_n_u8_m(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CHECK-LABEL: @test_svmls_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_u16_mu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_svmls_n_u16_m(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CHECK-LABEL: @test_svmls_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_u32_mu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_svmls_n_u32_m(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CHECK-LABEL: @test_svmls_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_u64_mu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_svmls_n_u64_m(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
 
 // CHECK-LABEL: @test_svmls_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmls_n_s8_xu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_svmls_n_s8_x(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CHECK-LABEL: @test_svmls_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_s16_xu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_svmls_n_s16_x(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CHECK-LABEL: @test_svmls_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_s32_xu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_svmls_n_s32_x(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CHECK-LABEL: @test_svmls_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_s64_xu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_svmls_n_s64_x(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
 
 // CHECK-LABEL: @test_svmls_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmls_n_u8_xu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mls.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_svmls_n_u8_x(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CHECK-LABEL: @test_svmls_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_u16_xu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mls.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_svmls_n_u16_x(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CHECK-LABEL: @test_svmls_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_u32_xu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mls.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_svmls_n_u32_x(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CHECK-LABEL: @test_svmls_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_u64_xu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mls.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -1096,8 +1048,7 @@ svfloat64_t test_svmls_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
 // CHECK-LABEL: @test_svmls_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1105,8 +1056,7 @@ svfloat64_t test_svmls_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1119,8 +1069,7 @@ svfloat16_t test_svmls_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CHECK-LABEL: @test_svmls_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1128,8 +1077,7 @@ svfloat16_t test_svmls_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1142,8 +1090,7 @@ svfloat32_t test_svmls_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CHECK-LABEL: @test_svmls_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1151,8 +1098,7 @@ svfloat32_t test_svmls_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1165,16 +1111,14 @@ svfloat64_t test_svmls_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
 // CHECK-LABEL: @test_svmls_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1186,16 +1130,14 @@ svfloat16_t test_svmls_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CHECK-LABEL: @test_svmls_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1207,16 +1149,14 @@ svfloat32_t test_svmls_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CHECK-LABEL: @test_svmls_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -1228,16 +1168,14 @@ svfloat64_t test_svmls_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
 // CHECK-LABEL: @test_svmls_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1249,16 +1187,14 @@ svfloat16_t test_svmls_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CHECK-LABEL: @test_svmls_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1270,16 +1206,14 @@ svfloat32_t test_svmls_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CHECK-LABEL: @test_svmls_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmls_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_msb.c

@@ -428,16 +428,14 @@ svuint64_t test_svmsb_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2, svuint6
 
 // CHECK-LABEL: @test_svmsb_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmsb_n_s8_zu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_svmsb_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CHECK-LABEL: @test_svmsb_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_svmsb_n_s8_z(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_s16_zu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_svmsb_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CHECK-LABEL: @test_svmsb_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_svmsb_n_s16_z(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_s32_zu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_svmsb_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CHECK-LABEL: @test_svmsb_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_svmsb_n_s32_z(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_s64_zu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_svmsb_n_s64_z(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
 
 // CHECK-LABEL: @test_svmsb_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmsb_n_u8_zu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_svmsb_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CHECK-LABEL: @test_svmsb_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_svmsb_n_u8_z(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_u16_zu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_svmsb_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CHECK-LABEL: @test_svmsb_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_svmsb_n_u16_z(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_u32_zu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_svmsb_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CHECK-LABEL: @test_svmsb_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_svmsb_n_u32_z(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_u64_zu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_svmsb_n_u64_z(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
 
 // CHECK-LABEL: @test_svmsb_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmsb_n_s8_mu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_svmsb_n_s8_m(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CHECK-LABEL: @test_svmsb_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_s16_mu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_svmsb_n_s16_m(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CHECK-LABEL: @test_svmsb_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_s32_mu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_svmsb_n_s32_m(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CHECK-LABEL: @test_svmsb_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_s64_mu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_svmsb_n_s64_m(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
 
 // CHECK-LABEL: @test_svmsb_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmsb_n_u8_mu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_svmsb_n_u8_m(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CHECK-LABEL: @test_svmsb_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_u16_mu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_svmsb_n_u16_m(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CHECK-LABEL: @test_svmsb_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_u32_mu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_svmsb_n_u32_m(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CHECK-LABEL: @test_svmsb_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_u64_mu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_svmsb_n_u64_m(svbool_t pg, svuint64_t op1, svuint64_t op2, uint6
 
 // CHECK-LABEL: @test_svmsb_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmsb_n_s8_xu10__SVBool_tu10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_svmsb_n_s8_x(svbool_t pg, svint8_t op1, svint8_t op2, int8_t op3)
 // CHECK-LABEL: @test_svmsb_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_s16_xu10__SVBool_tu11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_svmsb_n_s16_x(svbool_t pg, svint16_t op1, svint16_t op2, int16_t
 // CHECK-LABEL: @test_svmsb_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_s32_xu10__SVBool_tu11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_svmsb_n_s32_x(svbool_t pg, svint32_t op1, svint32_t op2, int32_t
 // CHECK-LABEL: @test_svmsb_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_s64_xu10__SVBool_tu11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_svmsb_n_s64_x(svbool_t pg, svint64_t op1, svint64_t op2, int64_t
 
 // CHECK-LABEL: @test_svmsb_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmsb_n_u8_xu10__SVBool_tu11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.msb.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_svmsb_n_u8_x(svbool_t pg, svuint8_t op1, svuint8_t op2, uint8_t o
 // CHECK-LABEL: @test_svmsb_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_u16_xu10__SVBool_tu12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.msb.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_svmsb_n_u16_x(svbool_t pg, svuint16_t op1, svuint16_t op2, uint1
 // CHECK-LABEL: @test_svmsb_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_u32_xu10__SVBool_tu12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.msb.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_svmsb_n_u32_x(svbool_t pg, svuint32_t op1, svuint32_t op2, uint3
 // CHECK-LABEL: @test_svmsb_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_u64_xu10__SVBool_tu12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.msb.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -1096,8 +1048,7 @@ svfloat64_t test_svmsb_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
 // CHECK-LABEL: @test_svmsb_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1105,8 +1056,7 @@ svfloat64_t test_svmsb_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svfl
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1119,8 +1069,7 @@ svfloat16_t test_svmsb_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CHECK-LABEL: @test_svmsb_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1128,8 +1077,7 @@ svfloat16_t test_svmsb_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1142,8 +1090,7 @@ svfloat32_t test_svmsb_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CHECK-LABEL: @test_svmsb_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1151,8 +1098,7 @@ svfloat32_t test_svmsb_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1165,16 +1111,14 @@ svfloat64_t test_svmsb_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
 // CHECK-LABEL: @test_svmsb_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1186,16 +1130,14 @@ svfloat16_t test_svmsb_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CHECK-LABEL: @test_svmsb_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1207,16 +1149,14 @@ svfloat32_t test_svmsb_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CHECK-LABEL: @test_svmsb_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -1228,16 +1168,14 @@ svfloat64_t test_svmsb_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, fl
 // CHECK-LABEL: @test_svmsb_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1249,16 +1187,14 @@ svfloat16_t test_svmsb_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, fl
 // CHECK-LABEL: @test_svmsb_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1270,16 +1206,14 @@ svfloat32_t test_svmsb_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, fl
 // CHECK-LABEL: @test_svmsb_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmsb_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mul.c

@@ -428,16 +428,14 @@ svuint64_t test_svmul_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svmul_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmul_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_svmul_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svmul_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_svmul_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_svmul_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svmul_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_svmul_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_svmul_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svmul_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_svmul_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_svmul_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svmul_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmul_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_svmul_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svmul_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_svmul_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_svmul_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svmul_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_svmul_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_svmul_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svmul_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_svmul_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_svmul_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svmul_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmul_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_svmul_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svmul_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_svmul_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svmul_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_svmul_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svmul_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_svmul_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svmul_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmul_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_svmul_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svmul_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_svmul_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svmul_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_svmul_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svmul_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_svmul_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svmul_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmul_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_svmul_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svmul_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_svmul_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svmul_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_svmul_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svmul_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_svmul_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svmul_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svmul_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.mul.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_svmul_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svmul_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.mul.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_svmul_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svmul_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.mul.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_svmul_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svmul_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.mul.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -1096,8 +1048,7 @@ svfloat64_t test_svmul_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svmul_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1105,8 +1056,7 @@ svfloat64_t test_svmul_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1119,8 +1069,7 @@ svfloat16_t test_svmul_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svmul_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1128,8 +1077,7 @@ svfloat16_t test_svmul_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1142,8 +1090,7 @@ svfloat32_t test_svmul_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svmul_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1151,8 +1098,7 @@ svfloat32_t test_svmul_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_f64_zu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1165,16 +1111,14 @@ svfloat64_t test_svmul_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svmul_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1186,16 +1130,14 @@ svfloat16_t test_svmul_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svmul_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1207,16 +1149,14 @@ svfloat32_t test_svmul_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svmul_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_f64_mu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -1228,16 +1168,14 @@ svfloat64_t test_svmul_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svmul_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmul.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1249,16 +1187,14 @@ svfloat16_t test_svmul_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svmul_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmul.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1270,16 +1206,14 @@ svfloat32_t test_svmul_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svmul_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmul_n_f64_xu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmul.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mulh.c

@@ -428,16 +428,14 @@ svuint64_t test_svmulh_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svmulh_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmulh_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_svmulh_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svmulh_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_svmulh_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_svmulh_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svmulh_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_svmulh_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_svmulh_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svmulh_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_svmulh_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_svmulh_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svmulh_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmulh_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_svmulh_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svmulh_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_svmulh_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_svmulh_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svmulh_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_svmulh_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_svmulh_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svmulh_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_svmulh_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_svmulh_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svmulh_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmulh_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_svmulh_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svmulh_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_svmulh_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svmulh_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_svmulh_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svmulh_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_svmulh_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svmulh_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmulh_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_svmulh_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svmulh_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_svmulh_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svmulh_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_svmulh_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svmulh_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_svmulh_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svmulh_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmulh_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.smulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_svmulh_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svmulh_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_svmulh_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svmulh_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_svmulh_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svmulh_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_svmulh_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svmulh_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmulh_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.umulh.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_svmulh_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svmulh_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umulh.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_svmulh_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svmulh_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umulh.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_svmulh_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svmulh_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulh_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umulh.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_mulx.c

@@ -176,8 +176,7 @@ svfloat64_t test_svmulx_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svmulx_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -185,8 +184,7 @@ svfloat64_t test_svmulx_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svmulx_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -199,8 +197,7 @@ svfloat16_t test_svmulx_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svmulx_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -208,8 +205,7 @@ svfloat16_t test_svmulx_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svmulx_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -222,8 +218,7 @@ svfloat32_t test_svmulx_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svmulx_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -231,8 +226,7 @@ svfloat32_t test_svmulx_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svmulx_n_f64_zu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -245,16 +239,14 @@ svfloat64_t test_svmulx_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svmulx_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulx_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -266,16 +258,14 @@ svfloat16_t test_svmulx_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svmulx_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulx_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -287,16 +277,14 @@ svfloat32_t test_svmulx_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svmulx_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulx_n_f64_mu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -308,16 +296,14 @@ svfloat64_t test_svmulx_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svmulx_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulx_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fmulx.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -329,16 +315,14 @@ svfloat16_t test_svmulx_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svmulx_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulx_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmulx.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -350,16 +334,14 @@ svfloat32_t test_svmulx_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svmulx_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svmulx_n_f64_xu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fmulx.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmad.c

@@ -176,8 +176,7 @@ svfloat64_t test_svnmad_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
 // CHECK-LABEL: @test_svnmad_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -185,8 +184,7 @@ svfloat64_t test_svnmad_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
 // CPP-CHECK-LABEL: @_Z19test_svnmad_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -199,8 +197,7 @@ svfloat16_t test_svnmad_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CHECK-LABEL: @test_svnmad_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -208,8 +205,7 @@ svfloat16_t test_svnmad_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CPP-CHECK-LABEL: @_Z19test_svnmad_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -222,8 +218,7 @@ svfloat32_t test_svnmad_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CHECK-LABEL: @test_svnmad_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -231,8 +226,7 @@ svfloat32_t test_svnmad_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CPP-CHECK-LABEL: @_Z19test_svnmad_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -245,16 +239,14 @@ svfloat64_t test_svnmad_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
 // CHECK-LABEL: @test_svnmad_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmad_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -266,16 +258,14 @@ svfloat16_t test_svnmad_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CHECK-LABEL: @test_svnmad_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmad_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -287,16 +277,14 @@ svfloat32_t test_svnmad_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CHECK-LABEL: @test_svnmad_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmad_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -308,16 +296,14 @@ svfloat64_t test_svnmad_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
 // CHECK-LABEL: @test_svnmad_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmad_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmad.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -329,16 +315,14 @@ svfloat16_t test_svnmad_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CHECK-LABEL: @test_svnmad_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmad_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmad.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -350,16 +334,14 @@ svfloat32_t test_svnmad_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CHECK-LABEL: @test_svnmad_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmad_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmad.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmla.c

@@ -176,8 +176,7 @@ svfloat64_t test_svnmla_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
 // CHECK-LABEL: @test_svnmla_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -185,8 +184,7 @@ svfloat64_t test_svnmla_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
 // CPP-CHECK-LABEL: @_Z19test_svnmla_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -199,8 +197,7 @@ svfloat16_t test_svnmla_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CHECK-LABEL: @test_svnmla_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -208,8 +205,7 @@ svfloat16_t test_svnmla_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CPP-CHECK-LABEL: @_Z19test_svnmla_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -222,8 +218,7 @@ svfloat32_t test_svnmla_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CHECK-LABEL: @test_svnmla_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -231,8 +226,7 @@ svfloat32_t test_svnmla_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CPP-CHECK-LABEL: @_Z19test_svnmla_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -245,16 +239,14 @@ svfloat64_t test_svnmla_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
 // CHECK-LABEL: @test_svnmla_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmla_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -266,16 +258,14 @@ svfloat16_t test_svnmla_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CHECK-LABEL: @test_svnmla_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmla_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -287,16 +277,14 @@ svfloat32_t test_svnmla_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CHECK-LABEL: @test_svnmla_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmla_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -308,16 +296,14 @@ svfloat64_t test_svnmla_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
 // CHECK-LABEL: @test_svnmla_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmla_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmla.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -329,16 +315,14 @@ svfloat16_t test_svnmla_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CHECK-LABEL: @test_svnmla_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmla_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmla.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -350,16 +334,14 @@ svfloat32_t test_svnmla_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CHECK-LABEL: @test_svnmla_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmla_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmla.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmls.c

@@ -176,8 +176,7 @@ svfloat64_t test_svnmls_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
 // CHECK-LABEL: @test_svnmls_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -185,8 +184,7 @@ svfloat64_t test_svnmls_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
 // CPP-CHECK-LABEL: @_Z19test_svnmls_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -199,8 +197,7 @@ svfloat16_t test_svnmls_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CHECK-LABEL: @test_svnmls_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -208,8 +205,7 @@ svfloat16_t test_svnmls_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CPP-CHECK-LABEL: @_Z19test_svnmls_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -222,8 +218,7 @@ svfloat32_t test_svnmls_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CHECK-LABEL: @test_svnmls_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -231,8 +226,7 @@ svfloat32_t test_svnmls_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CPP-CHECK-LABEL: @_Z19test_svnmls_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -245,16 +239,14 @@ svfloat64_t test_svnmls_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
 // CHECK-LABEL: @test_svnmls_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmls_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -266,16 +258,14 @@ svfloat16_t test_svnmls_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CHECK-LABEL: @test_svnmls_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmls_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -287,16 +277,14 @@ svfloat32_t test_svnmls_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CHECK-LABEL: @test_svnmls_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmls_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -308,16 +296,14 @@ svfloat64_t test_svnmls_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
 // CHECK-LABEL: @test_svnmls_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmls_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmls.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -329,16 +315,14 @@ svfloat16_t test_svnmls_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CHECK-LABEL: @test_svnmls_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmls_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmls.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -350,16 +334,14 @@ svfloat32_t test_svnmls_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CHECK-LABEL: @test_svnmls_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmls_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmls.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_nmsb.c

@@ -176,8 +176,7 @@ svfloat64_t test_svnmsb_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
 // CHECK-LABEL: @test_svnmsb_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -185,8 +184,7 @@ svfloat64_t test_svnmsb_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2, svf
 // CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f16_zu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -199,8 +197,7 @@ svfloat16_t test_svnmsb_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CHECK-LABEL: @test_svnmsb_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -208,8 +205,7 @@ svfloat16_t test_svnmsb_n_f16_z(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f32_zu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -222,8 +218,7 @@ svfloat32_t test_svnmsb_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CHECK-LABEL: @test_svnmsb_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -231,8 +226,7 @@ svfloat32_t test_svnmsb_n_f32_z(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f64_zu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -245,16 +239,14 @@ svfloat64_t test_svnmsb_n_f64_z(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
 // CHECK-LABEL: @test_svnmsb_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f16_mu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -266,16 +258,14 @@ svfloat16_t test_svnmsb_n_f16_m(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CHECK-LABEL: @test_svnmsb_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f32_mu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -287,16 +277,14 @@ svfloat32_t test_svnmsb_n_f32_m(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CHECK-LABEL: @test_svnmsb_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f64_mu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -308,16 +296,14 @@ svfloat64_t test_svnmsb_n_f64_m(svbool_t pg, svfloat64_t op1, svfloat64_t op2, f
 // CHECK-LABEL: @test_svnmsb_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f16_xu10__SVBool_tu13__SVFloat16_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fnmsb.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -329,16 +315,14 @@ svfloat16_t test_svnmsb_n_f16_x(svbool_t pg, svfloat16_t op1, svfloat16_t op2, f
 // CHECK-LABEL: @test_svnmsb_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f32_xu10__SVBool_tu13__SVFloat32_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fnmsb.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -350,16 +334,14 @@ svfloat32_t test_svnmsb_n_f32_x(svbool_t pg, svfloat32_t op1, svfloat32_t op2, f
 // CHECK-LABEL: @test_svnmsb_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svnmsb_n_f64_xu10__SVBool_tu13__SVFloat64_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fnmsb.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_orr.c

@@ -428,16 +428,14 @@ svuint64_t test_svorr_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svorr_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svorr_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_svorr_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svorr_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_svorr_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_svorr_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svorr_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_svorr_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_svorr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svorr_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_svorr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_svorr_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svorr_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svorr_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_svorr_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svorr_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_svorr_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_svorr_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svorr_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_svorr_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_svorr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svorr_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_svorr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_svorr_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svorr_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svorr_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_svorr_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svorr_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_svorr_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svorr_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_svorr_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svorr_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_svorr_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svorr_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svorr_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_svorr_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svorr_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_svorr_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svorr_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_svorr_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svorr_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_svorr_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svorr_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svorr_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_svorr_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svorr_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_svorr_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svorr_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_svorr_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svorr_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_svorr_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svorr_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svorr_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.orr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_svorr_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svorr_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.orr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_svorr_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svorr_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.orr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_svorr_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svorr_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svorr_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.orr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_qadd.c

@@ -136,15 +136,13 @@ svuint64_t test_svqadd_u64(svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svqadd_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svqadd_n_s8u10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -155,15 +153,13 @@ svint8_t test_svqadd_n_s8(svint8_t op1, int8_t op2)
 
 // CHECK-LABEL: @test_svqadd_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svqadd_n_s16u11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -174,15 +170,13 @@ svint16_t test_svqadd_n_s16(svint16_t op1, int16_t op2)
 
 // CHECK-LABEL: @test_svqadd_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svqadd_n_s32u11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -193,15 +187,13 @@ svint32_t test_svqadd_n_s32(svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svqadd_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svqadd_n_s64u11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -212,15 +204,13 @@ svint64_t test_svqadd_n_s64(svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svqadd_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svqadd_n_u8u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -231,15 +221,13 @@ svuint8_t test_svqadd_n_u8(svuint8_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svqadd_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svqadd_n_u16u12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -250,15 +238,13 @@ svuint16_t test_svqadd_n_u16(svuint16_t op1, uint16_t op2)
 
 // CHECK-LABEL: @test_svqadd_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svqadd_n_u32u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -269,15 +255,13 @@ svuint32_t test_svqadd_n_u32(svuint32_t op1, uint32_t op2)
 
 // CHECK-LABEL: @test_svqadd_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svqadd_n_u64u12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_qsub.c

@@ -136,15 +136,13 @@ svuint64_t test_svqsub_u64(svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svqsub_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svqsub_n_s8u10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqsub.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -155,15 +153,13 @@ svint8_t test_svqsub_n_s8(svint8_t op1, int8_t op2)
 
 // CHECK-LABEL: @test_svqsub_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svqsub_n_s16u11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqsub.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -174,15 +170,13 @@ svint16_t test_svqsub_n_s16(svint16_t op1, int16_t op2)
 
 // CHECK-LABEL: @test_svqsub_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svqsub_n_s32u11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqsub.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -193,15 +187,13 @@ svint32_t test_svqsub_n_s32(svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svqsub_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svqsub_n_s64u11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqsub.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -212,15 +204,13 @@ svint64_t test_svqsub_n_s64(svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svqsub_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svqsub_n_u8u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqsub.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -231,15 +221,13 @@ svuint8_t test_svqsub_n_u8(svuint8_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svqsub_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svqsub_n_u16u12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqsub.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -250,15 +238,13 @@ svuint16_t test_svqsub_n_u16(svuint16_t op1, uint16_t op2)
 
 // CHECK-LABEL: @test_svqsub_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svqsub_n_u32u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqsub.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -269,15 +255,13 @@ svuint32_t test_svqsub_n_u32(svuint32_t op1, uint32_t op2)
 
 // CHECK-LABEL: @test_svqsub_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svqsub_n_u64u12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqsub.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_scale.c

@@ -176,8 +176,7 @@ svfloat64_t test_svscale_f64_x(svbool_t pg, svfloat64_t op1, svint64_t op2)
 // CHECK-LABEL: @test_svscale_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -185,8 +184,7 @@ svfloat64_t test_svscale_f64_x(svbool_t pg, svfloat64_t op1, svint64_t op2)
 // CPP-CHECK-LABEL: @_Z20test_svscale_n_f16_zu10__SVBool_tu13__SVFloat16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -199,8 +197,7 @@ svfloat16_t test_svscale_n_f16_z(svbool_t pg, svfloat16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svscale_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -208,8 +205,7 @@ svfloat16_t test_svscale_n_f16_z(svbool_t pg, svfloat16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z20test_svscale_n_f32_zu10__SVBool_tu13__SVFloat32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -222,8 +218,7 @@ svfloat32_t test_svscale_n_f32_z(svbool_t pg, svfloat32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svscale_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -231,8 +226,7 @@ svfloat32_t test_svscale_n_f32_z(svbool_t pg, svfloat32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z20test_svscale_n_f64_zu10__SVBool_tu13__SVFloat64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -245,16 +239,14 @@ svfloat64_t test_svscale_n_f64_z(svbool_t pg, svfloat64_t op1, int64_t op2)
 // CHECK-LABEL: @test_svscale_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svscale_n_f16_mu10__SVBool_tu13__SVFloat16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -266,16 +258,14 @@ svfloat16_t test_svscale_n_f16_m(svbool_t pg, svfloat16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svscale_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svscale_n_f32_mu10__SVBool_tu13__SVFloat32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -287,16 +277,14 @@ svfloat32_t test_svscale_n_f32_m(svbool_t pg, svfloat32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svscale_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svscale_n_f64_mu10__SVBool_tu13__SVFloat64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -308,16 +296,14 @@ svfloat64_t test_svscale_n_f64_m(svbool_t pg, svfloat64_t op1, int64_t op2)
 // CHECK-LABEL: @test_svscale_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svscale_n_f16_xu10__SVBool_tu13__SVFloat16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fscale.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -329,16 +315,14 @@ svfloat16_t test_svscale_n_f16_x(svbool_t pg, svfloat16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svscale_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svscale_n_f32_xu10__SVBool_tu13__SVFloat32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fscale.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -350,16 +334,14 @@ svfloat32_t test_svscale_n_f32_x(svbool_t pg, svfloat32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svscale_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svscale_n_f64_xu10__SVBool_tu13__SVFloat64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fscale.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_sub.c

@@ -428,16 +428,14 @@ svuint64_t test_svsub_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svsub_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svsub_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_svsub_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svsub_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_svsub_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_svsub_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svsub_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_svsub_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_svsub_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svsub_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_svsub_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_svsub_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svsub_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svsub_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_svsub_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svsub_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_svsub_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_svsub_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svsub_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_svsub_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_svsub_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svsub_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_svsub_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_svsub_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svsub_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svsub_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_svsub_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svsub_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_svsub_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svsub_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_svsub_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svsub_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_svsub_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svsub_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svsub_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_svsub_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svsub_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_svsub_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svsub_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_svsub_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svsub_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_svsub_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svsub_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svsub_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_svsub_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svsub_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_svsub_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svsub_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_svsub_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svsub_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_svsub_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svsub_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svsub_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_svsub_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svsub_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_svsub_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svsub_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_svsub_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svsub_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -1096,8 +1048,7 @@ svfloat64_t test_svsub_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svsub_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1105,8 +1056,7 @@ svfloat64_t test_svsub_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1119,8 +1069,7 @@ svfloat16_t test_svsub_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svsub_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1128,8 +1077,7 @@ svfloat16_t test_svsub_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1142,8 +1090,7 @@ svfloat32_t test_svsub_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svsub_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1151,8 +1098,7 @@ svfloat32_t test_svsub_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_f64_zu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1165,16 +1111,14 @@ svfloat64_t test_svsub_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svsub_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1186,16 +1130,14 @@ svfloat16_t test_svsub_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svsub_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1207,16 +1149,14 @@ svfloat32_t test_svsub_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svsub_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_f64_mu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -1228,16 +1168,14 @@ svfloat64_t test_svsub_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svsub_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsub.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1249,16 +1187,14 @@ svfloat16_t test_svsub_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svsub_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsub.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1270,16 +1206,14 @@ svfloat32_t test_svsub_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svsub_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsub_n_f64_xu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsub.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_subr.c

@@ -428,16 +428,14 @@ svuint64_t test_svsubr_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svsubr_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsubr_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -450,8 +448,7 @@ svint8_t test_svsubr_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svsubr_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -459,8 +456,7 @@ svint8_t test_svsubr_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -473,8 +469,7 @@ svint16_t test_svsubr_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svsubr_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -482,8 +477,7 @@ svint16_t test_svsubr_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -496,8 +490,7 @@ svint32_t test_svsubr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svsubr_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -505,8 +498,7 @@ svint32_t test_svsubr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -518,16 +510,14 @@ svint64_t test_svsubr_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svsubr_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsubr_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -540,8 +530,7 @@ svuint8_t test_svsubr_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svsubr_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -549,8 +538,7 @@ svuint8_t test_svsubr_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -563,8 +551,7 @@ svuint16_t test_svsubr_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svsubr_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -572,8 +559,7 @@ svuint16_t test_svsubr_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -586,8 +572,7 @@ svuint32_t test_svsubr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svsubr_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -595,8 +580,7 @@ svuint32_t test_svsubr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -608,15 +592,13 @@ svuint64_t test_svsubr_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svsubr_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsubr_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -628,16 +610,14 @@ svint8_t test_svsubr_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svsubr_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -649,16 +629,14 @@ svint16_t test_svsubr_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svsubr_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -670,16 +648,14 @@ svint32_t test_svsubr_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svsubr_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -690,15 +666,13 @@ svint64_t test_svsubr_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svsubr_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsubr_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -710,16 +684,14 @@ svuint8_t test_svsubr_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svsubr_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -731,16 +703,14 @@ svuint16_t test_svsubr_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svsubr_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -752,16 +722,14 @@ svuint32_t test_svsubr_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svsubr_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -772,15 +740,13 @@ svuint64_t test_svsubr_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svsubr_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsubr_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -792,16 +758,14 @@ svint8_t test_svsubr_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svsubr_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -813,16 +777,14 @@ svint16_t test_svsubr_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svsubr_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -834,16 +796,14 @@ svint32_t test_svsubr_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svsubr_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -854,15 +814,13 @@ svint64_t test_svsubr_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svsubr_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsubr_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.subr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -874,16 +832,14 @@ svuint8_t test_svsubr_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svsubr_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.subr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -895,16 +851,14 @@ svuint16_t test_svsubr_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svsubr_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.subr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -916,16 +870,14 @@ svuint32_t test_svsubr_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svsubr_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.subr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -1096,8 +1048,7 @@ svfloat64_t test_svsubr_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CHECK-LABEL: @test_svsubr_n_f16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1105,8 +1056,7 @@ svfloat64_t test_svsubr_f64_x(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_f16_zu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.sel.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[TMP2]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP3]]
@@ -1119,8 +1069,7 @@ svfloat16_t test_svsubr_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svsubr_n_f32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1128,8 +1077,7 @@ svfloat16_t test_svsubr_n_f16_z(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_f32_zu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.sel.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[TMP2]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP3]]
@@ -1142,8 +1090,7 @@ svfloat32_t test_svsubr_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svsubr_n_f64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1151,8 +1098,7 @@ svfloat32_t test_svsubr_n_f32_z(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_f64_zu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.sel.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[TMP2]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP3]]
@@ -1165,16 +1111,14 @@ svfloat64_t test_svsubr_n_f64_z(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svsubr_n_f16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_f16_mu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1186,16 +1130,14 @@ svfloat16_t test_svsubr_n_f16_m(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svsubr_n_f32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_f32_mu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1207,16 +1149,14 @@ svfloat32_t test_svsubr_n_f32_m(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svsubr_n_f64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_f64_mu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
@@ -1228,16 +1168,14 @@ svfloat64_t test_svsubr_n_f64_m(svbool_t pg, svfloat64_t op1, float64_t op2)
 // CHECK-LABEL: @test_svsubr_n_f16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_f16_xu10__SVBool_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.fsubr.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x half> [[TMP2]]
 //
@@ -1249,16 +1187,14 @@ svfloat16_t test_svsubr_n_f16_x(svbool_t pg, svfloat16_t op1, float16_t op2)
 // CHECK-LABEL: @test_svsubr_n_f32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_f32_xu10__SVBool_tu13__SVFloat32_tf(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.dup.x.nxv4f32(float [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fsubr.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP2]]
 //
@@ -1270,16 +1206,14 @@ svfloat32_t test_svsubr_n_f32_x(svbool_t pg, svfloat32_t op1, float32_t op2)
 // CHECK-LABEL: @test_svsubr_n_f64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svsubr_n_f64_xu10__SVBool_tu13__SVFloat64_td(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.dup.x.nxv2f64(double [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x double> @llvm.aarch64.sve.fsubr.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x double> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_sudot.c

@@ -29,15 +29,13 @@ svint32_t test_svsudot_s32(svint32_t x, svint8_t y, svuint8_t z) {
 
 // CHECK-LABEL: @test_svsudot_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[Z:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[Z:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usdot.nxv4i32(<vscale x 4 x i32> [[X:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[Y:%.*]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svsudot_n_s32u11__SVInt32_tu10__SVInt8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[Z:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[Z:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usdot.nxv4i32(<vscale x 4 x i32> [[X:%.*]], <vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[Y:%.*]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_usdot.c

@@ -29,15 +29,13 @@ svint32_t test_svusdot_s32(svint32_t x, svuint8_t y, svint8_t z) {
 
 // CHECK-LABEL: @test_svusdot_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[Z:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[Z:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usdot.nxv4i32(<vscale x 4 x i32> [[X:%.*]], <vscale x 16 x i8> [[Y:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svusdot_n_s32u11__SVInt32_tu11__SVUint8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[Z:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[Z:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.usdot.nxv4i32(<vscale x 4 x i32> [[X:%.*]], <vscale x 16 x i8> [[Y:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_aba.c

@@ -153,15 +153,13 @@ svuint64_t test_svaba_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
 
 // CHECK-LABEL: @test_svaba_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.saba.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z15test_svaba_n_s8u10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.saba.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -174,15 +172,13 @@ svint8_t test_svaba_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svaba_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saba.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svaba_n_s16u11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saba.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -195,15 +191,13 @@ svint16_t test_svaba_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svaba_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saba.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svaba_n_s32u11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saba.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -216,15 +210,13 @@ svint32_t test_svaba_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_svaba_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saba.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svaba_n_s64u11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saba.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -237,15 +229,13 @@ svint64_t test_svaba_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
 
 // CHECK-LABEL: @test_svaba_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uaba.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z15test_svaba_n_u8u11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uaba.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -258,15 +248,13 @@ svuint8_t test_svaba_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_svaba_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaba.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svaba_n_u16u12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaba.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -279,15 +267,13 @@ svuint16_t test_svaba_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_svaba_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaba.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svaba_n_u32u12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaba.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -300,15 +286,13 @@ svuint32_t test_svaba_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
 
 // CHECK-LABEL: @test_svaba_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaba.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svaba_n_u64u12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaba.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abalb.c

@@ -119,15 +119,13 @@ svuint64_t test_svabalb_u64(svuint64_t op1, svuint32_t op2, svuint32_t op3)
 
 // CHECK-LABEL: @test_svabalb_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabalb_n_s16u11__SVInt16_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint16_t test_svabalb_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svabalb_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabalb_n_s32u11__SVInt32_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint32_t test_svabalb_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svabalb_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabalb_n_s64u11__SVInt64_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint64_t test_svabalb_n_s64(svint64_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_svabalb_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabalb_n_u16u12__SVUint16_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint16_t test_svabalb_n_u16(svuint16_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_svabalb_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabalb_n_u32u12__SVUint32_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint32_t test_svabalb_n_u32(svuint32_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_svabalb_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabalb_n_u64u12__SVUint64_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abalt.c

@@ -119,15 +119,13 @@ svuint64_t test_svabalt_u64(svuint64_t op1, svuint32_t op2, svuint32_t op3)
 
 // CHECK-LABEL: @test_svabalt_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabalt_n_s16u11__SVInt16_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint16_t test_svabalt_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svabalt_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabalt_n_s32u11__SVInt32_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint32_t test_svabalt_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svabalt_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabalt_n_s64u11__SVInt64_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint64_t test_svabalt_n_s64(svint64_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_svabalt_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabalt_n_u16u12__SVUint16_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint16_t test_svabalt_n_u16(svuint16_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_svabalt_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabalt_n_u32u12__SVUint32_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint32_t test_svabalt_n_u32(svuint32_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_svabalt_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabalt_n_u64u12__SVUint64_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abdlb.c

@@ -119,15 +119,13 @@ svuint64_t test_svabdlb_u64(svuint32_t op1, svuint32_t op2)
 
 // CHECK-LABEL: @test_svabdlb_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabdlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabdlb_n_s16u10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabdlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint16_t test_svabdlb_n_s16(svint8_t op1, int8_t op2)
 
 // CHECK-LABEL: @test_svabdlb_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabdlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabdlb_n_s32u11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabdlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint32_t test_svabdlb_n_s32(svint16_t op1, int16_t op2)
 
 // CHECK-LABEL: @test_svabdlb_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabdlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabdlb_n_s64u11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabdlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint64_t test_svabdlb_n_s64(svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svabdlb_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabdlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabdlb_n_u16u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabdlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint16_t test_svabdlb_n_u16(svuint8_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svabdlb_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabdlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabdlb_n_u32u12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabdlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint32_t test_svabdlb_n_u32(svuint16_t op1, uint16_t op2)
 
 // CHECK-LABEL: @test_svabdlb_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabdlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabdlb_n_u64u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabdlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_abdlt.c

@@ -119,15 +119,13 @@ svuint64_t test_svabdlt_u64(svuint32_t op1, svuint32_t op2)
 
 // CHECK-LABEL: @test_svabdlt_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabdlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabdlt_n_s16u10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sabdlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint16_t test_svabdlt_n_s16(svint8_t op1, int8_t op2)
 
 // CHECK-LABEL: @test_svabdlt_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabdlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabdlt_n_s32u11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sabdlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint32_t test_svabdlt_n_s32(svint16_t op1, int16_t op2)
 
 // CHECK-LABEL: @test_svabdlt_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabdlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabdlt_n_s64u11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sabdlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint64_t test_svabdlt_n_s64(svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svabdlt_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabdlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabdlt_n_u16u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uabdlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint16_t test_svabdlt_n_u16(svuint8_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svabdlt_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabdlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabdlt_n_u32u12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uabdlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint32_t test_svabdlt_n_u32(svuint16_t op1, uint16_t op2)
 
 // CHECK-LABEL: @test_svabdlt_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabdlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svabdlt_n_u64u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uabdlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_adclb.c

@@ -51,15 +51,13 @@ svuint64_t test_svadclb_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
 
 // CHECK-LABEL: @test_svadclb_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.adclb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadclb_n_u32u12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.adclb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -72,15 +70,13 @@ svuint32_t test_svadclb_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
 
 // CHECK-LABEL: @test_svadclb_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.adclb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadclb_n_u64u12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.adclb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_adclt.c

@@ -51,15 +51,13 @@ svuint64_t test_svadclt_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
 
 // CHECK-LABEL: @test_svadclt_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.adclt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadclt_n_u32u12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.adclt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -72,15 +70,13 @@ svuint32_t test_svadclt_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
 
 // CHECK-LABEL: @test_svadclt_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.adclt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svadclt_n_u64u12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.adclt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addhnb.c

@@ -119,15 +119,13 @@ svuint32_t test_svaddhnb_u64(svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svaddhnb_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_s16u11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint8_t test_svaddhnb_n_s16(svint16_t op1, int16_t op2)
 
 // CHECK-LABEL: @test_svaddhnb_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_s32u11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint16_t test_svaddhnb_n_s32(svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svaddhnb_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_s64u11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint32_t test_svaddhnb_n_s64(svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svaddhnb_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_u16u12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint8_t test_svaddhnb_n_u16(svuint16_t op1, uint16_t op2)
 
 // CHECK-LABEL: @test_svaddhnb_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_u32u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint16_t test_svaddhnb_n_u32(svuint32_t op1, uint32_t op2)
 
 // CHECK-LABEL: @test_svaddhnb_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svaddhnb_n_u64u12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addhnt.c

@@ -119,15 +119,13 @@ svuint32_t test_svaddhnt_u64(svuint32_t op1, svuint64_t op2, svuint64_t op3)
 
 // CHECK-LABEL: @test_svaddhnt_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_s16u10__SVInt8_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint8_t test_svaddhnt_n_s16(svint8_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svaddhnt_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_s32u11__SVInt16_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint16_t test_svaddhnt_n_s32(svint16_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_svaddhnt_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_s64u11__SVInt32_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint32_t test_svaddhnt_n_s64(svint32_t op1, svint64_t op2, int64_t op3)
 
 // CHECK-LABEL: @test_svaddhnt_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_u16u11__SVUint8_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.addhnt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint8_t test_svaddhnt_n_u16(svuint8_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_svaddhnt_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_u32u12__SVUint16_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.addhnt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint16_t test_svaddhnt_n_u32(svuint16_t op1, svuint32_t op2, uint32_t op3)
 
 // CHECK-LABEL: @test_svaddhnt_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svaddhnt_n_u64u12__SVUint32_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.addhnt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlb.c

@@ -119,15 +119,13 @@ svuint64_t test_svaddlb_u64(svuint32_t op1, svuint32_t op2)
 
 // CHECK-LABEL: @test_svaddlb_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddlb_n_s16u10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint16_t test_svaddlb_n_s16(svint8_t op1, int8_t op2)
 
 // CHECK-LABEL: @test_svaddlb_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddlb_n_s32u11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint32_t test_svaddlb_n_s32(svint16_t op1, int16_t op2)
 
 // CHECK-LABEL: @test_svaddlb_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddlb_n_s64u11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint64_t test_svaddlb_n_s64(svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svaddlb_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddlb_n_u16u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddlb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint16_t test_svaddlb_n_u16(svuint8_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svaddlb_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddlb_n_u32u12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddlb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint32_t test_svaddlb_n_u32(svuint16_t op1, uint16_t op2)
 
 // CHECK-LABEL: @test_svaddlb_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddlb_n_u64u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddlb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlbt.c

@@ -68,15 +68,13 @@ svint64_t test_svaddlbt_s64(svint32_t op1, svint32_t op2)
 
 // CHECK-LABEL: @test_svaddlbt_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlbt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svaddlbt_n_s16u10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlbt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -89,15 +87,13 @@ svint16_t test_svaddlbt_n_s16(svint8_t op1, int8_t op2)
 
 // CHECK-LABEL: @test_svaddlbt_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlbt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svaddlbt_n_s32u11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlbt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -110,15 +106,13 @@ svint32_t test_svaddlbt_n_s32(svint16_t op1, int16_t op2)
 
 // CHECK-LABEL: @test_svaddlbt_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlbt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svaddlbt_n_s64u11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlbt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addlt.c

@@ -119,15 +119,13 @@ svuint64_t test_svaddlt_u64(svuint32_t op1, svuint32_t op2)
 
 // CHECK-LABEL: @test_svaddlt_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddlt_n_s16u10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint16_t test_svaddlt_n_s16(svint8_t op1, int8_t op2)
 
 // CHECK-LABEL: @test_svaddlt_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddlt_n_s32u11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint32_t test_svaddlt_n_s32(svint16_t op1, int16_t op2)
 
 // CHECK-LABEL: @test_svaddlt_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddlt_n_s64u11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint64_t test_svaddlt_n_s64(svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svaddlt_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddlt_n_u16u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddlt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint16_t test_svaddlt_n_u16(svuint8_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svaddlt_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddlt_n_u32u12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddlt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint32_t test_svaddlt_n_u32(svuint16_t op1, uint16_t op2)
 
 // CHECK-LABEL: @test_svaddlt_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddlt_n_u64u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddlt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addwb.c

@@ -119,15 +119,13 @@ svuint64_t test_svaddwb_u64(svuint64_t op1, svuint32_t op2)
 
 // CHECK-LABEL: @test_svaddwb_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddwb_n_s16u11__SVInt16_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint16_t test_svaddwb_n_s16(svint16_t op1, int8_t op2)
 
 // CHECK-LABEL: @test_svaddwb_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddwb_n_s32u11__SVInt32_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint32_t test_svaddwb_n_s32(svint32_t op1, int16_t op2)
 
 // CHECK-LABEL: @test_svaddwb_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddwb_n_s64u11__SVInt64_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint64_t test_svaddwb_n_s64(svint64_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svaddwb_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddwb_n_u16u12__SVUint16_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddwb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint16_t test_svaddwb_n_u16(svuint16_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svaddwb_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddwb_n_u32u12__SVUint32_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddwb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint32_t test_svaddwb_n_u32(svuint32_t op1, uint16_t op2)
 
 // CHECK-LABEL: @test_svaddwb_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddwb_n_u64u12__SVUint64_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddwb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_addwt.c

@@ -119,15 +119,13 @@ svuint64_t test_svaddwt_u64(svuint64_t op1, svuint32_t op2)
 
 // CHECK-LABEL: @test_svaddwt_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddwt_n_s16u11__SVInt16_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.saddwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint16_t test_svaddwt_n_s16(svint16_t op1, int8_t op2)
 
 // CHECK-LABEL: @test_svaddwt_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddwt_n_s32u11__SVInt32_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.saddwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint32_t test_svaddwt_n_s32(svint32_t op1, int16_t op2)
 
 // CHECK-LABEL: @test_svaddwt_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddwt_n_s64u11__SVInt64_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.saddwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint64_t test_svaddwt_n_s64(svint64_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svaddwt_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddwt_n_u16u12__SVUint16_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uaddwt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint16_t test_svaddwt_n_u16(svuint16_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svaddwt_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddwt_n_u32u12__SVUint32_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uaddwt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint32_t test_svaddwt_n_u32(svuint32_t op1, uint16_t op2)
 
 // CHECK-LABEL: @test_svaddwt_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svaddwt_n_u64u12__SVUint64_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uaddwt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bcax.c

@@ -153,15 +153,13 @@ svuint64_t test_svbcax_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
 
 // CHECK-LABEL: @test_svbcax_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bcax.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svbcax_n_s8u10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bcax.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -174,15 +172,13 @@ svint8_t test_svbcax_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svbcax_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bcax.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbcax_n_s16u11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bcax.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -195,15 +191,13 @@ svint16_t test_svbcax_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svbcax_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bcax.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbcax_n_s32u11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bcax.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -216,15 +210,13 @@ svint32_t test_svbcax_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_svbcax_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bcax.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbcax_n_s64u11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bcax.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -237,15 +229,13 @@ svint64_t test_svbcax_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
 
 // CHECK-LABEL: @test_svbcax_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bcax.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svbcax_n_u8u11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bcax.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -258,15 +248,13 @@ svuint8_t test_svbcax_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_svbcax_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bcax.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbcax_n_u16u12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bcax.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -279,15 +267,13 @@ svuint16_t test_svbcax_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_svbcax_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bcax.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbcax_n_u32u12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bcax.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -300,15 +286,13 @@ svuint32_t test_svbcax_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
 
 // CHECK-LABEL: @test_svbcax_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bcax.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbcax_n_u64u12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bcax.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bdep.c

@@ -85,15 +85,13 @@ svuint64_t test_svbdep_u64(svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svbdep_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bdep.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svbdep_n_u8u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bdep.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -106,15 +104,13 @@ svuint8_t test_svbdep_n_u8(svuint8_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svbdep_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bdep.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbdep_n_u16u12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bdep.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -127,15 +123,13 @@ svuint16_t test_svbdep_n_u16(svuint16_t op1, uint16_t op2)
 
 // CHECK-LABEL: @test_svbdep_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bdep.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbdep_n_u32u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bdep.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -148,15 +142,13 @@ svuint32_t test_svbdep_n_u32(svuint32_t op1, uint32_t op2)
 
 // CHECK-LABEL: @test_svbdep_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bdep.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbdep_n_u64u12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bdep.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bext.c

@@ -85,15 +85,13 @@ svuint64_t test_svbext_u64(svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svbext_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bext.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svbext_n_u8u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bext.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -106,15 +104,13 @@ svuint8_t test_svbext_n_u8(svuint8_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svbext_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bext.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbext_n_u16u12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bext.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -127,15 +123,13 @@ svuint16_t test_svbext_n_u16(svuint16_t op1, uint16_t op2)
 
 // CHECK-LABEL: @test_svbext_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bext.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbext_n_u32u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bext.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -148,15 +142,13 @@ svuint32_t test_svbext_n_u32(svuint32_t op1, uint32_t op2)
 
 // CHECK-LABEL: @test_svbext_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bext.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbext_n_u64u12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bext.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bgrp.c

@@ -85,15 +85,13 @@ svuint64_t test_svbgrp_u64(svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svbgrp_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bgrp.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svbgrp_n_u8u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bgrp.x.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -106,15 +104,13 @@ svuint8_t test_svbgrp_n_u8(svuint8_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svbgrp_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bgrp.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbgrp_n_u16u12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bgrp.x.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -127,15 +123,13 @@ svuint16_t test_svbgrp_n_u16(svuint16_t op1, uint16_t op2)
 
 // CHECK-LABEL: @test_svbgrp_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bgrp.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbgrp_n_u32u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bgrp.x.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -148,15 +142,13 @@ svuint32_t test_svbgrp_n_u32(svuint32_t op1, uint32_t op2)
 
 // CHECK-LABEL: @test_svbgrp_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bgrp.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbgrp_n_u64u12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bgrp.x.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl.c

@@ -153,15 +153,13 @@ svuint64_t test_svbsl_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
 
 // CHECK-LABEL: @test_svbsl_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z15test_svbsl_n_s8u10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -174,15 +172,13 @@ svint8_t test_svbsl_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svbsl_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svbsl_n_s16u11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -195,15 +191,13 @@ svint16_t test_svbsl_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svbsl_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svbsl_n_s32u11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -216,15 +210,13 @@ svint32_t test_svbsl_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_svbsl_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svbsl_n_s64u11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -237,15 +229,13 @@ svint64_t test_svbsl_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
 
 // CHECK-LABEL: @test_svbsl_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z15test_svbsl_n_u8u11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -258,15 +248,13 @@ svuint8_t test_svbsl_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_svbsl_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svbsl_n_u16u12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -279,15 +267,13 @@ svuint16_t test_svbsl_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_svbsl_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svbsl_n_u32u12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -300,15 +286,13 @@ svuint32_t test_svbsl_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
 
 // CHECK-LABEL: @test_svbsl_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svbsl_n_u64u12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl1n.c

@@ -153,15 +153,13 @@ svuint64_t test_svbsl1n_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
 
 // CHECK-LABEL: @test_svbsl1n_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl1n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbsl1n_n_s8u10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl1n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -174,15 +172,13 @@ svint8_t test_svbsl1n_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svbsl1n_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl1n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_s16u11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl1n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -195,15 +191,13 @@ svint16_t test_svbsl1n_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svbsl1n_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl1n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_s32u11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl1n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -216,15 +210,13 @@ svint32_t test_svbsl1n_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_svbsl1n_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl1n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_s64u11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl1n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -237,15 +229,13 @@ svint64_t test_svbsl1n_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
 
 // CHECK-LABEL: @test_svbsl1n_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl1n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbsl1n_n_u8u11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl1n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -258,15 +248,13 @@ svuint8_t test_svbsl1n_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_svbsl1n_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl1n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_u16u12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl1n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -279,15 +267,13 @@ svuint16_t test_svbsl1n_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_svbsl1n_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl1n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_u32u12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl1n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -300,15 +286,13 @@ svuint32_t test_svbsl1n_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
 
 // CHECK-LABEL: @test_svbsl1n_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl1n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbsl1n_n_u64u12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl1n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_bsl2n.c

@@ -153,15 +153,13 @@ svuint64_t test_svbsl2n_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
 
 // CHECK-LABEL: @test_svbsl2n_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl2n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbsl2n_n_s8u10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl2n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -174,15 +172,13 @@ svint8_t test_svbsl2n_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svbsl2n_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl2n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_s16u11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl2n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -195,15 +191,13 @@ svint16_t test_svbsl2n_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svbsl2n_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl2n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_s32u11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl2n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -216,15 +210,13 @@ svint32_t test_svbsl2n_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_svbsl2n_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl2n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_s64u11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl2n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -237,15 +229,13 @@ svint64_t test_svbsl2n_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
 
 // CHECK-LABEL: @test_svbsl2n_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl2n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svbsl2n_n_u8u11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.bsl2n.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -258,15 +248,13 @@ svuint8_t test_svbsl2n_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_svbsl2n_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl2n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_u16u12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.bsl2n.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -279,15 +267,13 @@ svuint16_t test_svbsl2n_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_svbsl2n_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl2n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_u32u12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.bsl2n.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -300,15 +286,13 @@ svuint32_t test_svbsl2n_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
 
 // CHECK-LABEL: @test_svbsl2n_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl2n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svbsl2n_n_u64u12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.bsl2n.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eor3.c

@@ -153,15 +153,13 @@ svuint64_t test_sveor3_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
 
 // CHECK-LABEL: @test_sveor3_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor3.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_sveor3_n_s8u10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor3.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -174,15 +172,13 @@ svint8_t test_sveor3_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_sveor3_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor3.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveor3_n_s16u11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor3.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -195,15 +191,13 @@ svint16_t test_sveor3_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_sveor3_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor3.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveor3_n_s32u11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor3.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -216,15 +210,13 @@ svint32_t test_sveor3_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_sveor3_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor3.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveor3_n_s64u11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor3.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -237,15 +229,13 @@ svint64_t test_sveor3_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
 
 // CHECK-LABEL: @test_sveor3_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor3.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_sveor3_n_u8u11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eor3.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -258,15 +248,13 @@ svuint8_t test_sveor3_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_sveor3_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor3.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveor3_n_u16u12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eor3.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -279,15 +267,13 @@ svuint16_t test_sveor3_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_sveor3_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor3.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveor3_n_u32u12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eor3.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -300,15 +286,13 @@ svuint32_t test_sveor3_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
 
 // CHECK-LABEL: @test_sveor3_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor3.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveor3_n_u64u12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eor3.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eorbt.c

@@ -153,15 +153,13 @@ svuint64_t test_sveorbt_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
 
 // CHECK-LABEL: @test_sveorbt_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eorbt.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveorbt_n_s8u10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eorbt.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -174,15 +172,13 @@ svint8_t test_sveorbt_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_sveorbt_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eorbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveorbt_n_s16u11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eorbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -195,15 +191,13 @@ svint16_t test_sveorbt_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_sveorbt_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eorbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveorbt_n_s32u11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eorbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -216,15 +210,13 @@ svint32_t test_sveorbt_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_sveorbt_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eorbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveorbt_n_s64u11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eorbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -237,15 +229,13 @@ svint64_t test_sveorbt_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
 
 // CHECK-LABEL: @test_sveorbt_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eorbt.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveorbt_n_u8u11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eorbt.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -258,15 +248,13 @@ svuint8_t test_sveorbt_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_sveorbt_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eorbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveorbt_n_u16u12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eorbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -279,15 +267,13 @@ svuint16_t test_sveorbt_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_sveorbt_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eorbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveorbt_n_u32u12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eorbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -300,15 +286,13 @@ svuint32_t test_sveorbt_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
 
 // CHECK-LABEL: @test_sveorbt_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eorbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveorbt_n_u64u12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eorbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_eortb.c

@@ -153,15 +153,13 @@ svuint64_t test_sveortb_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
 
 // CHECK-LABEL: @test_sveortb_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eortb.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveortb_n_s8u10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eortb.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -174,15 +172,13 @@ svint8_t test_sveortb_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_sveortb_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eortb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveortb_n_s16u11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eortb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -195,15 +191,13 @@ svint16_t test_sveortb_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_sveortb_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eortb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveortb_n_s32u11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eortb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -216,15 +210,13 @@ svint32_t test_sveortb_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_sveortb_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eortb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveortb_n_s64u11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eortb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -237,15 +229,13 @@ svint64_t test_sveortb_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
 
 // CHECK-LABEL: @test_sveortb_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eortb.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_sveortb_n_u8u11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.eortb.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -258,15 +248,13 @@ svuint8_t test_sveortb_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_sveortb_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eortb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveortb_n_u16u12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.eortb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -279,15 +267,13 @@ svuint16_t test_sveortb_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_sveortb_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eortb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveortb_n_u32u12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.eortb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -300,15 +286,13 @@ svuint32_t test_sveortb_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
 
 // CHECK-LABEL: @test_sveortb_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eortb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_sveortb_n_u64u12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.eortb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hadd.c

@@ -166,15 +166,13 @@ svuint64_t test_svhadd_u64_m(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svhadd_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svhadd_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -188,16 +186,14 @@ svint8_t test_svhadd_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svhadd_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -211,16 +207,14 @@ svint16_t test_svhadd_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svhadd_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -234,16 +228,14 @@ svint32_t test_svhadd_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svhadd_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -256,15 +248,13 @@ svint64_t test_svhadd_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svhadd_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svhadd_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -278,16 +268,14 @@ svuint8_t test_svhadd_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svhadd_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -301,16 +289,14 @@ svuint16_t test_svhadd_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svhadd_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -324,16 +310,14 @@ svuint32_t test_svhadd_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svhadd_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -511,16 +495,14 @@ svuint64_t test_svhadd_u64_z(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svhadd_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svhadd_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -535,8 +517,7 @@ svint8_t test_svhadd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svhadd_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -544,8 +525,7 @@ svint8_t test_svhadd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -560,8 +540,7 @@ svint16_t test_svhadd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svhadd_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -569,8 +548,7 @@ svint16_t test_svhadd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -585,8 +563,7 @@ svint32_t test_svhadd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svhadd_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -594,8 +571,7 @@ svint32_t test_svhadd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -609,16 +585,14 @@ svint64_t test_svhadd_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svhadd_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svhadd_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -633,8 +607,7 @@ svuint8_t test_svhadd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svhadd_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -642,8 +615,7 @@ svuint8_t test_svhadd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -658,8 +630,7 @@ svuint16_t test_svhadd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svhadd_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -667,8 +638,7 @@ svuint16_t test_svhadd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -683,8 +653,7 @@ svuint32_t test_svhadd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svhadd_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -692,8 +661,7 @@ svuint32_t test_svhadd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -856,15 +824,13 @@ svuint64_t test_svhadd_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svhadd_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svhadd_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -878,16 +844,14 @@ svint8_t test_svhadd_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svhadd_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -901,16 +865,14 @@ svint16_t test_svhadd_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svhadd_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -924,16 +886,14 @@ svint32_t test_svhadd_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svhadd_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -946,15 +906,13 @@ svint64_t test_svhadd_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svhadd_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svhadd_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -968,16 +926,14 @@ svuint8_t test_svhadd_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svhadd_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -991,16 +947,14 @@ svuint16_t test_svhadd_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svhadd_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -1014,16 +968,14 @@ svuint32_t test_svhadd_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svhadd_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhadd_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hsub.c

@@ -477,16 +477,14 @@ svuint64_t test_svhsub_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svhsub_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svhsub_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -501,8 +499,7 @@ svint8_t test_svhsub_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svhsub_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -510,8 +507,7 @@ svint8_t test_svhsub_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -526,8 +522,7 @@ svint16_t test_svhsub_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svhsub_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -535,8 +530,7 @@ svint16_t test_svhsub_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -551,8 +545,7 @@ svint32_t test_svhsub_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svhsub_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -560,8 +553,7 @@ svint32_t test_svhsub_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -575,16 +567,14 @@ svint64_t test_svhsub_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svhsub_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svhsub_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -599,8 +589,7 @@ svuint8_t test_svhsub_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svhsub_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -608,8 +597,7 @@ svuint8_t test_svhsub_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -624,8 +612,7 @@ svuint16_t test_svhsub_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svhsub_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -633,8 +620,7 @@ svuint16_t test_svhsub_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -649,8 +635,7 @@ svuint32_t test_svhsub_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svhsub_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -658,8 +643,7 @@ svuint32_t test_svhsub_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -673,15 +657,13 @@ svuint64_t test_svhsub_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svhsub_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svhsub_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -695,16 +677,14 @@ svint8_t test_svhsub_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svhsub_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -718,16 +698,14 @@ svint16_t test_svhsub_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svhsub_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -741,16 +719,14 @@ svint32_t test_svhsub_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svhsub_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -763,15 +739,13 @@ svint64_t test_svhsub_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svhsub_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svhsub_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -785,16 +759,14 @@ svuint8_t test_svhsub_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svhsub_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -808,16 +780,14 @@ svuint16_t test_svhsub_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svhsub_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -831,16 +801,14 @@ svuint32_t test_svhsub_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svhsub_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -853,15 +821,13 @@ svuint64_t test_svhsub_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svhsub_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svhsub_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -875,16 +841,14 @@ svint8_t test_svhsub_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svhsub_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -898,16 +862,14 @@ svint16_t test_svhsub_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svhsub_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -921,16 +883,14 @@ svint32_t test_svhsub_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svhsub_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -943,15 +903,13 @@ svint64_t test_svhsub_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svhsub_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svhsub_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsub.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -965,16 +923,14 @@ svuint8_t test_svhsub_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svhsub_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsub.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -988,16 +944,14 @@ svuint16_t test_svhsub_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svhsub_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsub.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -1011,16 +965,14 @@ svuint32_t test_svhsub_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svhsub_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsub_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsub.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_hsubr.c

@@ -477,16 +477,14 @@ svuint64_t test_svhsubr_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svhsubr_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsubr_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -501,8 +499,7 @@ svint8_t test_svhsubr_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svhsubr_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -510,8 +507,7 @@ svint8_t test_svhsubr_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -526,8 +522,7 @@ svint16_t test_svhsubr_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svhsubr_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -535,8 +530,7 @@ svint16_t test_svhsubr_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -551,8 +545,7 @@ svint32_t test_svhsubr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svhsubr_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -560,8 +553,7 @@ svint32_t test_svhsubr_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -575,16 +567,14 @@ svint64_t test_svhsubr_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svhsubr_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsubr_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -599,8 +589,7 @@ svuint8_t test_svhsubr_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svhsubr_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -608,8 +597,7 @@ svuint8_t test_svhsubr_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -624,8 +612,7 @@ svuint16_t test_svhsubr_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svhsubr_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -633,8 +620,7 @@ svuint16_t test_svhsubr_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -649,8 +635,7 @@ svuint32_t test_svhsubr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svhsubr_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -658,8 +643,7 @@ svuint32_t test_svhsubr_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -673,15 +657,13 @@ svuint64_t test_svhsubr_n_u64_z(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svhsubr_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsubr_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -695,16 +677,14 @@ svint8_t test_svhsubr_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svhsubr_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -718,16 +698,14 @@ svint16_t test_svhsubr_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svhsubr_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -741,16 +719,14 @@ svint32_t test_svhsubr_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svhsubr_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -763,15 +739,13 @@ svint64_t test_svhsubr_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svhsubr_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsubr_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -785,16 +759,14 @@ svuint8_t test_svhsubr_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svhsubr_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -808,16 +780,14 @@ svuint16_t test_svhsubr_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svhsubr_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -831,16 +801,14 @@ svuint32_t test_svhsubr_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svhsubr_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -853,15 +821,13 @@ svuint64_t test_svhsubr_n_u64_m(svbool_t pg, svuint64_t op1, uint64_t op2)
 
 // CHECK-LABEL: @test_svhsubr_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsubr_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.shsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -875,16 +841,14 @@ svint8_t test_svhsubr_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svhsubr_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.shsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -898,16 +862,14 @@ svint16_t test_svhsubr_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svhsubr_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.shsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -921,16 +883,14 @@ svint32_t test_svhsubr_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svhsubr_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.shsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -943,15 +903,13 @@ svint64_t test_svhsubr_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svhsubr_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svhsubr_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uhsubr.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -965,16 +923,14 @@ svuint8_t test_svhsubr_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svhsubr_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uhsubr.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -988,16 +944,14 @@ svuint16_t test_svhsubr_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svhsubr_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uhsubr.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -1011,16 +965,14 @@ svuint32_t test_svhsubr_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svhsubr_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svhsubr_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uhsubr.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlalb.c

@@ -119,15 +119,13 @@ svuint64_t test_svmlalb_u64(svuint64_t op1, svuint32_t op2, svuint32_t op3)
 
 // CHECK-LABEL: @test_svmlalb_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlalb_n_s16u11__SVInt16_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint16_t test_svmlalb_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svmlalb_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlalb_n_s32u11__SVInt32_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint32_t test_svmlalb_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svmlalb_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlalb_n_s64u11__SVInt64_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint64_t test_svmlalb_n_s64(svint64_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_svmlalb_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlalb_n_u16u12__SVUint16_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint16_t test_svmlalb_n_u16(svuint16_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_svmlalb_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlalb_n_u32u12__SVUint32_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint32_t test_svmlalb_n_u32(svuint32_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_svmlalb_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlalb_n_u64u12__SVUint64_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -398,15 +386,13 @@ svfloat32_t test_svmlalb_f32(svfloat32_t op1, svfloat16_t op2, svfloat16_t op3)
 
 // CHECK-LABEL: @test_svmlalb_n_f32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlalb.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlalb_n_f32u13__SVFloat32_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlalb.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlalt.c

@@ -119,15 +119,13 @@ svuint64_t test_svmlalt_u64(svuint64_t op1, svuint32_t op2, svuint32_t op3)
 
 // CHECK-LABEL: @test_svmlalt_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlalt_n_s16u11__SVInt16_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint16_t test_svmlalt_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svmlalt_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlalt_n_s32u11__SVInt32_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint32_t test_svmlalt_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svmlalt_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlalt_n_s64u11__SVInt64_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint64_t test_svmlalt_n_s64(svint64_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_svmlalt_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlalt_n_u16u12__SVUint16_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint16_t test_svmlalt_n_u16(svuint16_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_svmlalt_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlalt_n_u32u12__SVUint32_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint32_t test_svmlalt_n_u32(svuint32_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_svmlalt_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlalt_n_u64u12__SVUint64_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -398,15 +386,13 @@ svfloat32_t test_svmlalt_f32(svfloat32_t op1, svfloat16_t op2, svfloat16_t op3)
 
 // CHECK-LABEL: @test_svmlalt_n_f32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlalt.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlalt_n_f32u13__SVFloat32_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlalt.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlslb.c

@@ -119,15 +119,13 @@ svuint64_t test_svmlslb_u64(svuint64_t op1, svuint32_t op2, svuint32_t op3)
 
 // CHECK-LABEL: @test_svmlslb_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlslb_n_s16u11__SVInt16_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint16_t test_svmlslb_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svmlslb_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlslb_n_s32u11__SVInt32_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint32_t test_svmlslb_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svmlslb_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlslb_n_s64u11__SVInt64_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint64_t test_svmlslb_n_s64(svint64_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_svmlslb_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlslb_n_u16u12__SVUint16_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint16_t test_svmlslb_n_u16(svuint16_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_svmlslb_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlslb_n_u32u12__SVUint32_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint32_t test_svmlslb_n_u32(svuint32_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_svmlslb_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlslb_n_u64u12__SVUint64_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -398,15 +386,13 @@ svfloat32_t test_svmlslb_f32(svfloat32_t op1, svfloat16_t op2, svfloat16_t op3)
 
 // CHECK-LABEL: @test_svmlslb_n_f32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlslb.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlslb_n_f32u13__SVFloat32_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlslb.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mlslt.c

@@ -119,15 +119,13 @@ svuint64_t test_svmlslt_u64(svuint64_t op1, svuint32_t op2, svuint32_t op3)
 
 // CHECK-LABEL: @test_svmlslt_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlslt_n_s16u11__SVInt16_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint16_t test_svmlslt_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svmlslt_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlslt_n_s32u11__SVInt32_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint32_t test_svmlslt_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svmlslt_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlslt_n_s64u11__SVInt64_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint64_t test_svmlslt_n_s64(svint64_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_svmlslt_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlslt_n_u16u12__SVUint16_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint16_t test_svmlslt_n_u16(svuint16_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_svmlslt_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlslt_n_u32u12__SVUint32_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint32_t test_svmlslt_n_u32(svuint32_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_svmlslt_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlslt_n_u64u12__SVUint64_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -398,15 +386,13 @@ svfloat32_t test_svmlslt_f32(svfloat32_t op1, svfloat16_t op2, svfloat16_t op3)
 
 // CHECK-LABEL: @test_svmlslt_n_f32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlslt.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmlslt_n_f32u13__SVFloat32_tu13__SVFloat16_tDh(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x half> @llvm.aarch64.sve.dup.x.nxv8f16(half [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x float> @llvm.aarch64.sve.fmlslt.nxv4f32(<vscale x 4 x float> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]], <vscale x 8 x half> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x float> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mullb.c

@@ -119,15 +119,13 @@ svuint64_t test_svmullb_u64(svuint32_t op1, svuint32_t op2)
 
 // CHECK-LABEL: @test_svmullb_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmullb_n_s16u10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint16_t test_svmullb_n_s16(svint8_t op1, int8_t op2)
 
 // CHECK-LABEL: @test_svmullb_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmullb_n_s32u11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint32_t test_svmullb_n_s32(svint16_t op1, int16_t op2)
 
 // CHECK-LABEL: @test_svmullb_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmullb_n_s64u11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint64_t test_svmullb_n_s64(svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svmullb_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmullb_n_u16u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint16_t test_svmullb_n_u16(svuint8_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svmullb_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmullb_n_u32u12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint32_t test_svmullb_n_u32(svuint16_t op1, uint16_t op2)
 
 // CHECK-LABEL: @test_svmullb_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmullb_n_u64u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_mullt.c

@@ -119,15 +119,13 @@ svuint64_t test_svmullt_u64(svuint32_t op1, svuint32_t op2)
 
 // CHECK-LABEL: @test_svmullt_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmullt_n_s16u10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.smullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -140,15 +138,13 @@ svint16_t test_svmullt_n_s16(svint8_t op1, int8_t op2)
 
 // CHECK-LABEL: @test_svmullt_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmullt_n_s32u11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.smullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -161,15 +157,13 @@ svint32_t test_svmullt_n_s32(svint16_t op1, int16_t op2)
 
 // CHECK-LABEL: @test_svmullt_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmullt_n_s64u11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.smullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -182,15 +176,13 @@ svint64_t test_svmullt_n_s64(svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svmullt_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmullt_n_u16u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.umullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -203,15 +195,13 @@ svuint16_t test_svmullt_n_u16(svuint8_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svmullt_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmullt_n_u32u12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.umullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -224,15 +214,13 @@ svuint32_t test_svmullt_n_u32(svuint16_t op1, uint16_t op2)
 
 // CHECK-LABEL: @test_svmullt_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svmullt_n_u64u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.umullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_nbsl.c

@@ -153,15 +153,13 @@ svuint64_t test_svnbsl_u64(svuint64_t op1, svuint64_t op2, svuint64_t op3)
 
 // CHECK-LABEL: @test_svnbsl_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.nbsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svnbsl_n_s8u10__SVInt8_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.nbsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -174,15 +172,13 @@ svint8_t test_svnbsl_n_s8(svint8_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svnbsl_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.nbsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svnbsl_n_s16u11__SVInt16_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.nbsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -195,15 +191,13 @@ svint16_t test_svnbsl_n_s16(svint16_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svnbsl_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.nbsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svnbsl_n_s32u11__SVInt32_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.nbsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -216,15 +210,13 @@ svint32_t test_svnbsl_n_s32(svint32_t op1, svint32_t op2, int32_t op3)
 
 // CHECK-LABEL: @test_svnbsl_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.nbsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svnbsl_n_s64u11__SVInt64_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.nbsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
@@ -237,15 +229,13 @@ svint64_t test_svnbsl_n_s64(svint64_t op1, svint64_t op2, int64_t op3)
 
 // CHECK-LABEL: @test_svnbsl_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.nbsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svnbsl_n_u8u11__SVUint8_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.nbsl.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -258,15 +248,13 @@ svuint8_t test_svnbsl_n_u8(svuint8_t op1, svuint8_t op2, uint8_t op3)
 
 // CHECK-LABEL: @test_svnbsl_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.nbsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svnbsl_n_u16u12__SVUint16_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.nbsl.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -279,15 +267,13 @@ svuint16_t test_svnbsl_n_u16(svuint16_t op1, svuint16_t op2, uint16_t op3)
 
 // CHECK-LABEL: @test_svnbsl_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.nbsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svnbsl_n_u32u12__SVUint32_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.nbsl.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -300,15 +286,13 @@ svuint32_t test_svnbsl_n_u32(svuint32_t op1, svuint32_t op2, uint32_t op3)
 
 // CHECK-LABEL: @test_svnbsl_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.nbsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z17test_svnbsl_n_u64u12__SVUint64_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.nbsl.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmul.c

@@ -34,15 +34,13 @@ svuint8_t test_svpmul_u8(svuint8_t op1, svuint8_t op2)
 
 // CHECK-LABEL: @test_svpmul_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmul.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z16test_svpmul_n_u8u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmul.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullb.c

@@ -51,15 +51,13 @@ svuint32_t test_svpmullb_pair_u32(svuint32_t op1, svuint32_t op2)
 
 // CHECK-LABEL: @test_svpmullb_pair_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullb.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z23test_svpmullb_pair_n_u8u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullb.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -72,15 +70,13 @@ svuint8_t test_svpmullb_pair_n_u8(svuint8_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svpmullb_pair_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullb.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z24test_svpmullb_pair_n_u32u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullb.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -131,16 +127,14 @@ svuint64_t test_svpmullb_u64(svuint32_t op1, svuint32_t op2)
 
 // CHECK-LABEL: @test_svpmullb_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullb.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    [[TMP2:%.*]] = bitcast <vscale x 16 x i8> [[TMP1]] to <vscale x 8 x i16>
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svpmullb_n_u16u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullb.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = bitcast <vscale x 16 x i8> [[TMP1]] to <vscale x 8 x i16>
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
@@ -154,16 +148,14 @@ svuint16_t test_svpmullb_n_u16(svuint8_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svpmullb_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullb.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    [[TMP2:%.*]] = bitcast <vscale x 4 x i32> [[TMP1]] to <vscale x 2 x i64>
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svpmullb_n_u64u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullb.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = bitcast <vscale x 4 x i32> [[TMP1]] to <vscale x 2 x i64>
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullb_128.c

@@ -34,15 +34,13 @@ svuint64_t test_svpmullb_pair_u64(svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svpmullb_pair_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.pmullb.pair.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z24test_svpmullb_pair_n_u64u12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.pmullb.pair.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullt.c

@@ -51,15 +51,13 @@ svuint32_t test_svpmullt_pair_u32(svuint32_t op1, svuint32_t op2)
 
 // CHECK-LABEL: @test_svpmullt_pair_n_u8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z23test_svpmullt_pair_n_u8u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -72,15 +70,13 @@ svuint8_t test_svpmullt_pair_n_u8(svuint8_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svpmullt_pair_n_u32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z24test_svpmullt_pair_n_u32u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -131,16 +127,14 @@ svuint64_t test_svpmullt_u64(svuint32_t op1, svuint32_t op2)
 
 // CHECK-LABEL: @test_svpmullt_n_u16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    [[TMP2:%.*]] = bitcast <vscale x 16 x i8> [[TMP1]] to <vscale x 8 x i16>
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svpmullt_n_u16u11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.pmullt.pair.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = bitcast <vscale x 16 x i8> [[TMP1]] to <vscale x 8 x i16>
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
@@ -154,16 +148,14 @@ svuint16_t test_svpmullt_n_u16(svuint8_t op1, uint8_t op2)
 
 // CHECK-LABEL: @test_svpmullt_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    [[TMP2:%.*]] = bitcast <vscale x 4 x i32> [[TMP1]] to <vscale x 2 x i64>
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svpmullt_n_u64u12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.pmullt.pair.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = bitcast <vscale x 4 x i32> [[TMP1]] to <vscale x 2 x i64>
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_pmullt_128.c

@@ -34,15 +34,13 @@ svuint64_t test_svpmullt_pair_u64(svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svpmullt_pair_n_u64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.pmullt.pair.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z24test_svpmullt_pair_n_u64u12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.pmullt.pair.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qadd.c

@@ -166,15 +166,13 @@ svuint64_t test_svqadd_u64_m(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svqadd_n_s8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svqadd_n_s8_mu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -188,16 +186,14 @@ svint8_t test_svqadd_n_s8_m(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svqadd_n_s16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_s16_mu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -211,16 +207,14 @@ svint16_t test_svqadd_n_s16_m(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svqadd_n_s32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_s32_mu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -234,16 +228,14 @@ svint32_t test_svqadd_n_s32_m(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svqadd_n_s64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_s64_mu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -256,15 +248,13 @@ svint64_t test_svqadd_n_s64_m(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svqadd_n_u8_m(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svqadd_n_u8_mu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -278,16 +268,14 @@ svuint8_t test_svqadd_n_u8_m(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svqadd_n_u16_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_u16_mu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -301,16 +289,14 @@ svuint16_t test_svqadd_n_u16_m(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svqadd_n_u32_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_u32_mu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -324,16 +310,14 @@ svuint32_t test_svqadd_n_u32_m(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svqadd_n_u64_m(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_u64_mu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -511,16 +495,14 @@ svuint64_t test_svqadd_u64_z(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svqadd_n_s8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svqadd_n_s8_zu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -535,8 +517,7 @@ svint8_t test_svqadd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svqadd_n_s16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -544,8 +525,7 @@ svint8_t test_svqadd_n_s8_z(svbool_t pg, svint8_t op1, int8_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_s16_zu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -560,8 +540,7 @@ svint16_t test_svqadd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svqadd_n_s32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -569,8 +548,7 @@ svint16_t test_svqadd_n_s16_z(svbool_t pg, svint16_t op1, int16_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_s32_zu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -585,8 +563,7 @@ svint32_t test_svqadd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svqadd_n_s64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -594,8 +571,7 @@ svint32_t test_svqadd_n_s32_z(svbool_t pg, svint32_t op1, int32_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_s64_zu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -609,16 +585,14 @@ svint64_t test_svqadd_n_s64_z(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svqadd_n_u8_z(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svqadd_n_u8_zu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sel.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG]], <vscale x 16 x i8> [[TMP1]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP2]]
@@ -633,8 +607,7 @@ svuint8_t test_svqadd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svqadd_n_u16_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -642,8 +615,7 @@ svuint8_t test_svqadd_n_u8_z(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_u16_zu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sel.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[TMP2]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP3]]
@@ -658,8 +630,7 @@ svuint16_t test_svqadd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svqadd_n_u32_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -667,8 +638,7 @@ svuint16_t test_svqadd_n_u16_z(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_u32_zu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sel.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[TMP2]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP3]]
@@ -683,8 +653,7 @@ svuint32_t test_svqadd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svqadd_n_u64_z(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -692,8 +661,7 @@ svuint32_t test_svqadd_n_u32_z(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_u64_zu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sel.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> zeroinitializer)
 // CPP-CHECK-NEXT:    [[TMP3:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[TMP2]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP3]]
@@ -856,15 +824,13 @@ svuint64_t test_svqadd_u64_x(svbool_t pg, svuint64_t op1, svuint64_t op2)
 
 // CHECK-LABEL: @test_svqadd_n_s8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svqadd_n_s8_xu10__SVBool_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -878,16 +844,14 @@ svint8_t test_svqadd_n_s8_x(svbool_t pg, svint8_t op1, int8_t op2)
 // CHECK-LABEL: @test_svqadd_n_s16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_s16_xu10__SVBool_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -901,16 +865,14 @@ svint16_t test_svqadd_n_s16_x(svbool_t pg, svint16_t op1, int16_t op2)
 // CHECK-LABEL: @test_svqadd_n_s32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_s32_xu10__SVBool_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -924,16 +886,14 @@ svint32_t test_svqadd_n_s32_x(svbool_t pg, svint32_t op1, int32_t op2)
 // CHECK-LABEL: @test_svqadd_n_s64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_s64_xu10__SVBool_tu11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
@@ -946,15 +906,13 @@ svint64_t test_svqadd_n_s64_x(svbool_t pg, svint64_t op1, int64_t op2)
 
 // CHECK-LABEL: @test_svqadd_n_u8_x(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svqadd_n_u8_xu10__SVBool_tu11__SVUint8_th(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.uqadd.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -968,16 +926,14 @@ svuint8_t test_svqadd_n_u8_x(svbool_t pg, svuint8_t op1, uint8_t op2)
 // CHECK-LABEL: @test_svqadd_n_u16_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_u16_xu10__SVBool_tu12__SVUint16_tt(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.uqadd.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP2]]
 //
@@ -991,16 +947,14 @@ svuint16_t test_svqadd_n_u16_x(svbool_t pg, svuint16_t op1, uint16_t op2)
 // CHECK-LABEL: @test_svqadd_n_u32_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_u32_xu10__SVBool_tu12__SVUint32_tj(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.uqadd.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP2]]
 //
@@ -1014,16 +968,14 @@ svuint32_t test_svqadd_n_u32_x(svbool_t pg, svuint32_t op1, uint32_t op2)
 // CHECK-LABEL: @test_svqadd_n_u64_x(
 // CHECK-NEXT:  entry:
 // CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svqadd_n_u64_xu10__SVBool_tu12__SVUint64_tm(
 // CPP-CHECK-NEXT:  entry:
 // CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP2:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.uqadd.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP2]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalb.c

@@ -68,15 +68,13 @@ svint64_t test_svqdmlalb_s64(svint64_t op1, svint32_t op2, svint32_t op3)
 
 // CHECK-LABEL: @test_svqdmlalb_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmlalb_n_s16u11__SVInt16_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -89,15 +87,13 @@ svint16_t test_svqdmlalb_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svqdmlalb_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmlalb_n_s32u11__SVInt32_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -110,15 +106,13 @@ svint32_t test_svqdmlalb_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svqdmlalb_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmlalb_n_s64u11__SVInt64_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalbt.c

@@ -68,15 +68,13 @@ svint64_t test_svqdmlalbt_s64(svint64_t op1, svint32_t op2, svint32_t op3)
 
 // CHECK-LABEL: @test_svqdmlalbt_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z21test_svqdmlalbt_n_s16u11__SVInt16_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -89,15 +87,13 @@ svint16_t test_svqdmlalbt_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svqdmlalbt_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z21test_svqdmlalbt_n_s32u11__SVInt32_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -110,15 +106,13 @@ svint32_t test_svqdmlalbt_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svqdmlalbt_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z21test_svqdmlalbt_n_s64u11__SVInt64_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlalt.c

@@ -68,15 +68,13 @@ svint64_t test_svqdmlalt_s64(svint64_t op1, svint32_t op2, svint32_t op3)
 
 // CHECK-LABEL: @test_svqdmlalt_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmlalt_n_s16u11__SVInt16_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlalt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -89,15 +87,13 @@ svint16_t test_svqdmlalt_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svqdmlalt_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmlalt_n_s32u11__SVInt32_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlalt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -110,15 +106,13 @@ svint32_t test_svqdmlalt_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svqdmlalt_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmlalt_n_s64u11__SVInt64_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlalt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslb.c

@@ -68,15 +68,13 @@ svint64_t test_svqdmlslb_s64(svint64_t op1, svint32_t op2, svint32_t op3)
 
 // CHECK-LABEL: @test_svqdmlslb_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmlslb_n_s16u11__SVInt16_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslb.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -89,15 +87,13 @@ svint16_t test_svqdmlslb_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svqdmlslb_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmlslb_n_s32u11__SVInt32_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslb.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -110,15 +106,13 @@ svint32_t test_svqdmlslb_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svqdmlslb_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmlslb_n_s64u11__SVInt64_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslb.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslbt.c

@@ -68,15 +68,13 @@ svint64_t test_svqdmlslbt_s64(svint64_t op1, svint32_t op2, svint32_t op3)
 
 // CHECK-LABEL: @test_svqdmlslbt_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z21test_svqdmlslbt_n_s16u11__SVInt16_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslbt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -89,15 +87,13 @@ svint16_t test_svqdmlslbt_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svqdmlslbt_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z21test_svqdmlslbt_n_s32u11__SVInt32_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslbt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -110,15 +106,13 @@ svint32_t test_svqdmlslbt_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svqdmlslbt_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z21test_svqdmlslbt_n_s64u11__SVInt64_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslbt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmlslt.c

@@ -68,15 +68,13 @@ svint64_t test_svqdmlslt_s64(svint64_t op1, svint32_t op2, svint32_t op3)
 
 // CHECK-LABEL: @test_svqdmlslt_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmlslt_n_s16u11__SVInt16_tu10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmlslt.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -89,15 +87,13 @@ svint16_t test_svqdmlslt_n_s16(svint16_t op1, svint8_t op2, int8_t op3)
 
 // CHECK-LABEL: @test_svqdmlslt_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmlslt_n_s32u11__SVInt32_tu11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmlslt.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -110,15 +106,13 @@ svint32_t test_svqdmlslt_n_s32(svint32_t op1, svint16_t op2, int16_t op3)
 
 // CHECK-LABEL: @test_svqdmlslt_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmlslt_n_s64u11__SVInt64_tu11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP3:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP3:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmlslt.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmulh.c

@@ -85,15 +85,13 @@ svint64_t test_svqdmulh_s64(svint64_t op1, svint64_t op2)
 
 // CHECK-LABEL: @test_svqdmulh_n_s8(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqdmulh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z18test_svqdmulh_n_s8u10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.sqdmulh.nxv16i8(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 16 x i8> [[TMP1]]
 //
@@ -106,15 +104,13 @@ svint8_t test_svqdmulh_n_s8(svint8_t op1, int8_t op2)
 
 // CHECK-LABEL: @test_svqdmulh_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmulh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svqdmulh_n_s16u11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmulh.nxv8i16(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -127,15 +123,13 @@ svint16_t test_svqdmulh_n_s16(svint16_t op1, int16_t op2)
 
 // CHECK-LABEL: @test_svqdmulh_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmulh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svqdmulh_n_s32u11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmulh.nxv4i32(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -148,15 +142,13 @@ svint32_t test_svqdmulh_n_s32(svint32_t op1, int32_t op2)
 
 // CHECK-LABEL: @test_svqdmulh_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmulh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z19test_svqdmulh_n_s64u11__SVInt64_tl(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.dup.x.nxv2i64(i64 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmulh.nxv2i64(<vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmullb.c

@@ -68,15 +68,13 @@ svint64_t test_svqdmullb_s64(svint32_t op1, svint32_t op2)
 
 // CHECK-LABEL: @test_svqdmullb_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmullb_n_s16u10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmullb.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -89,15 +87,13 @@ svint16_t test_svqdmullb_n_s16(svint8_t op1, int8_t op2)
 
 // CHECK-LABEL: @test_svqdmullb_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmullb_n_s32u11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmullb.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -110,15 +106,13 @@ svint32_t test_svqdmullb_n_s32(svint16_t op1, int16_t op2)
 
 // CHECK-LABEL: @test_svqdmullb_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmullb_n_s64u11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmullb.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //

clang/test/CodeGen/aarch64-sve2-intrinsics/acle_sve2_qdmullt.c

@@ -68,15 +68,13 @@ svint64_t test_svqdmullt_s64(svint32_t op1, svint32_t op2)
 
 // CHECK-LABEL: @test_svqdmullt_n_s16(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmullt_n_s16u10__SVInt8_ta(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 16 x i8> @llvm.aarch64.sve.dup.x.nxv16i8(i8 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.sqdmullt.nxv8i16(<vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 8 x i16> [[TMP1]]
 //
@@ -89,15 +87,13 @@ svint16_t test_svqdmullt_n_s16(svint8_t op1, int8_t op2)
 
 // CHECK-LABEL: @test_svqdmullt_n_s32(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmullt_n_s32u11__SVInt16_ts(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 8 x i16> @llvm.aarch64.sve.dup.x.nxv8i16(i16 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.sqdmullt.nxv4i32(<vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 4 x i32> [[TMP1]]
 //
@@ -110,15 +106,13 @@ svint32_t test_svqdmullt_n_s32(svint16_t op1, int16_t op2)
 
 // CHECK-LABEL: @test_svqdmullt_n_s64(
 // CHECK-NEXT:  entry:
-// CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //
 // CPP-CHECK-LABEL: @_Z20test_svqdmullt_n_s64u11__SVInt32_ti(
 // CPP-CHECK-NEXT:  entry:
-// CPP-CHECK-NEXT:    [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i32 0
+// CPP-CHECK-NEXT:    [[TMP0:%.*]] = call <vscale x 4 x i32> @llvm.aarch64.sve.dup.x.nxv4i32(i32 [[OP2:%.*]])
-// CPP-CHECK-NEXT:    [[TMP0:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
 // CPP-CHECK-NEXT:    [[TMP1:%.*]] = call <vscale x 2 x i64> @llvm.aarch64.sve.sqdmullt.nxv2i64(<vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP0]])
 // CPP-CHECK-NEXT:    ret <vscale x 2 x i64> [[TMP1]]
 //