[RISCV][VP] Lower FP VP ISD nodes to RVV instructions

With the exception of `frem`, this patch supports the current set of VP
floating-point binary intrinsics by lowering them to to RVV instructions. It
does so by using the existing `RISCVISD *_VL` custom nodes as an intermediate
layer. Both scalable and fixed-length vectors are supported by using this
method.

The `frem` node is unsupported due to a lack of available instructions. For
fixed-length vectors we could scalarize but that option is not (currently)
available for scalable-vector types. The support is intentionally left out so
it equivalent for both vector types.

The matching of vector/scalar forms is currently lacking, as scalable vector
types do not lower to the custom `VFMV_V_F_VL` node. We could either make
floating-point scalable vector splats lower to this node, or support the
matching of multiple kinds of splat via a `ComplexPattern`, much like we do for
integer types.

Reviewed By: rogfer01

Differential Revision: https://reviews.llvm.org/D104237

llvm/lib/Target/RISCV/RISCVISelLowering.cpp

@@ -420,6 +420,9 @@ RISCVTargetLowering::RISCVTargetLowering(const TargetMachine &TM,
         ISD::VP_SREM, ISD::VP_UREM, ISD::VP_AND, ISD::VP_OR,   ISD::VP_XOR,
         ISD::VP_ASHR, ISD::VP_LSHR, ISD::VP_SHL};
 
+    static unsigned FloatingPointVPOps[] = {ISD::VP_FADD, ISD::VP_FSUB,
+                                            ISD::VP_FMUL, ISD::VP_FDIV};
+
     if (!Subtarget.is64Bit()) {
       // We must custom-lower certain vXi64 operations on RV32 due to the vector
       // element type being illegal.
@@ -603,6 +606,9 @@ RISCVTargetLowering::RISCVTargetLowering(const TargetMachine &TM,
       setOperationAction(ISD::EXTRACT_SUBVECTOR, VT, Custom);
 
       setOperationAction(ISD::VECTOR_REVERSE, VT, Custom);
+
+      for (unsigned VPOpc : FloatingPointVPOps)
+        setOperationAction(VPOpc, VT, Custom);
     };
 
     // Sets common extload/truncstore actions on RVV floating-point vector
@@ -797,6 +803,9 @@ RISCVTargetLowering::RISCVTargetLowering(const TargetMachine &TM,
         setOperationAction(ISD::VECREDUCE_SEQ_FADD, VT, Custom);
         setOperationAction(ISD::VECREDUCE_FMIN, VT, Custom);
         setOperationAction(ISD::VECREDUCE_FMAX, VT, Custom);
+
+        for (unsigned VPOpc : FloatingPointVPOps)
+          setOperationAction(VPOpc, VT, Custom);
       }
 
       // Custom-legalize bitcasts from fixed-length vectors to scalar types.
@@ -2494,6 +2503,14 @@ SDValue RISCVTargetLowering::LowerOperation(SDValue Op,
     return lowerVPOp(Op, DAG, RISCVISD::SRL_VL);
   case ISD::VP_SHL:
     return lowerVPOp(Op, DAG, RISCVISD::SHL_VL);
+  case ISD::VP_FADD:
+    return lowerVPOp(Op, DAG, RISCVISD::FADD_VL);
+  case ISD::VP_FSUB:
+    return lowerVPOp(Op, DAG, RISCVISD::FSUB_VL);
+  case ISD::VP_FMUL:
+    return lowerVPOp(Op, DAG, RISCVISD::FMUL_VL);
+  case ISD::VP_FDIV:
+    return lowerVPOp(Op, DAG, RISCVISD::FDIV_VL);
   }
 }
 

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-vfadd-vp.ll

@@ -0,0 +1,629 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=ilp32d -riscv-v-vector-bits-min=128 \
+; RUN:   -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=lp64d -riscv-v-vector-bits-min=128 \
+; RUN:   -verify-machineinstrs < %s | FileCheck %s
+
+declare <2 x half> @llvm.vp.fadd.v2f16(<2 x half>, <2 x half>, <2 x i1>, i32)
+
+define <2 x half> @vfadd_vv_v2f16(<2 x half> %va, <2 x half> %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <2 x half> @llvm.vp.fadd.v2f16(<2 x half> %va, <2 x half> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+define <2 x half> @vfadd_vv_v2f16_unmasked(<2 x half> %va, <2 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fadd.v2f16(<2 x half> %va, <2 x half> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+define <2 x half> @vfadd_vf_v2f16(<2 x half> %va, half %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x half> undef, half %b, i32 0
+  %vb = shufflevector <2 x half> %elt.head, <2 x half> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fadd.v2f16(<2 x half> %va, <2 x half> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+define <2 x half> @vfadd_vf_v2f16_unmasked(<2 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x half> undef, half %b, i32 0
+  %vb = shufflevector <2 x half> %elt.head, <2 x half> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fadd.v2f16(<2 x half> %va, <2 x half> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+declare <4 x half> @llvm.vp.fadd.v4f16(<4 x half>, <4 x half>, <4 x i1>, i32)
+
+define <4 x half> @vfadd_vv_v4f16(<4 x half> %va, <4 x half> %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <4 x half> @llvm.vp.fadd.v4f16(<4 x half> %va, <4 x half> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+define <4 x half> @vfadd_vv_v4f16_unmasked(<4 x half> %va, <4 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fadd.v4f16(<4 x half> %va, <4 x half> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+define <4 x half> @vfadd_vf_v4f16(<4 x half> %va, half %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x half> undef, half %b, i32 0
+  %vb = shufflevector <4 x half> %elt.head, <4 x half> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fadd.v4f16(<4 x half> %va, <4 x half> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+define <4 x half> @vfadd_vf_v4f16_unmasked(<4 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x half> undef, half %b, i32 0
+  %vb = shufflevector <4 x half> %elt.head, <4 x half> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fadd.v4f16(<4 x half> %va, <4 x half> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+declare <8 x half> @llvm.vp.fadd.v8f16(<8 x half>, <8 x half>, <8 x i1>, i32)
+
+define <8 x half> @vfadd_vv_v8f16(<8 x half> %va, <8 x half> %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x half> @llvm.vp.fadd.v8f16(<8 x half> %va, <8 x half> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+define <8 x half> @vfadd_vv_v8f16_unmasked(<8 x half> %va, <8 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fadd.v8f16(<8 x half> %va, <8 x half> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+define <8 x half> @vfadd_vf_v8f16(<8 x half> %va, half %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x half> undef, half %b, i32 0
+  %vb = shufflevector <8 x half> %elt.head, <8 x half> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fadd.v8f16(<8 x half> %va, <8 x half> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+define <8 x half> @vfadd_vf_v8f16_unmasked(<8 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x half> undef, half %b, i32 0
+  %vb = shufflevector <8 x half> %elt.head, <8 x half> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fadd.v8f16(<8 x half> %va, <8 x half> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+declare <16 x half> @llvm.vp.fadd.v16f16(<16 x half>, <16 x half>, <16 x i1>, i32)
+
+define <16 x half> @vfadd_vv_v16f16(<16 x half> %va, <16 x half> %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <16 x half> @llvm.vp.fadd.v16f16(<16 x half> %va, <16 x half> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+define <16 x half> @vfadd_vv_v16f16_unmasked(<16 x half> %va, <16 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fadd.v16f16(<16 x half> %va, <16 x half> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+define <16 x half> @vfadd_vf_v16f16(<16 x half> %va, half %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x half> undef, half %b, i32 0
+  %vb = shufflevector <16 x half> %elt.head, <16 x half> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fadd.v16f16(<16 x half> %va, <16 x half> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+define <16 x half> @vfadd_vf_v16f16_unmasked(<16 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x half> undef, half %b, i32 0
+  %vb = shufflevector <16 x half> %elt.head, <16 x half> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fadd.v16f16(<16 x half> %va, <16 x half> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+declare <2 x float> @llvm.vp.fadd.v2f32(<2 x float>, <2 x float>, <2 x i1>, i32)
+
+define <2 x float> @vfadd_vv_v2f32(<2 x float> %va, <2 x float> %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <2 x float> @llvm.vp.fadd.v2f32(<2 x float> %va, <2 x float> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+define <2 x float> @vfadd_vv_v2f32_unmasked(<2 x float> %va, <2 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fadd.v2f32(<2 x float> %va, <2 x float> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+define <2 x float> @vfadd_vf_v2f32(<2 x float> %va, float %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x float> undef, float %b, i32 0
+  %vb = shufflevector <2 x float> %elt.head, <2 x float> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fadd.v2f32(<2 x float> %va, <2 x float> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+define <2 x float> @vfadd_vf_v2f32_unmasked(<2 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x float> undef, float %b, i32 0
+  %vb = shufflevector <2 x float> %elt.head, <2 x float> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fadd.v2f32(<2 x float> %va, <2 x float> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+declare <4 x float> @llvm.vp.fadd.v4f32(<4 x float>, <4 x float>, <4 x i1>, i32)
+
+define <4 x float> @vfadd_vv_v4f32(<4 x float> %va, <4 x float> %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <4 x float> @llvm.vp.fadd.v4f32(<4 x float> %va, <4 x float> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+define <4 x float> @vfadd_vv_v4f32_unmasked(<4 x float> %va, <4 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fadd.v4f32(<4 x float> %va, <4 x float> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+define <4 x float> @vfadd_vf_v4f32(<4 x float> %va, float %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x float> undef, float %b, i32 0
+  %vb = shufflevector <4 x float> %elt.head, <4 x float> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fadd.v4f32(<4 x float> %va, <4 x float> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+define <4 x float> @vfadd_vf_v4f32_unmasked(<4 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x float> undef, float %b, i32 0
+  %vb = shufflevector <4 x float> %elt.head, <4 x float> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fadd.v4f32(<4 x float> %va, <4 x float> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+declare <8 x float> @llvm.vp.fadd.v8f32(<8 x float>, <8 x float>, <8 x i1>, i32)
+
+define <8 x float> @vfadd_vv_v8f32(<8 x float> %va, <8 x float> %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x float> @llvm.vp.fadd.v8f32(<8 x float> %va, <8 x float> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+define <8 x float> @vfadd_vv_v8f32_unmasked(<8 x float> %va, <8 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fadd.v8f32(<8 x float> %va, <8 x float> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+define <8 x float> @vfadd_vf_v8f32(<8 x float> %va, float %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x float> undef, float %b, i32 0
+  %vb = shufflevector <8 x float> %elt.head, <8 x float> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fadd.v8f32(<8 x float> %va, <8 x float> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+define <8 x float> @vfadd_vf_v8f32_unmasked(<8 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x float> undef, float %b, i32 0
+  %vb = shufflevector <8 x float> %elt.head, <8 x float> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fadd.v8f32(<8 x float> %va, <8 x float> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+declare <16 x float> @llvm.vp.fadd.v16f32(<16 x float>, <16 x float>, <16 x i1>, i32)
+
+define <16 x float> @vfadd_vv_v16f32(<16 x float> %va, <16 x float> %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <16 x float> @llvm.vp.fadd.v16f32(<16 x float> %va, <16 x float> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+define <16 x float> @vfadd_vv_v16f32_unmasked(<16 x float> %va, <16 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fadd.v16f32(<16 x float> %va, <16 x float> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+define <16 x float> @vfadd_vf_v16f32(<16 x float> %va, float %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x float> undef, float %b, i32 0
+  %vb = shufflevector <16 x float> %elt.head, <16 x float> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fadd.v16f32(<16 x float> %va, <16 x float> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+define <16 x float> @vfadd_vf_v16f32_unmasked(<16 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x float> undef, float %b, i32 0
+  %vb = shufflevector <16 x float> %elt.head, <16 x float> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fadd.v16f32(<16 x float> %va, <16 x float> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+declare <2 x double> @llvm.vp.fadd.v2f64(<2 x double>, <2 x double>, <2 x i1>, i32)
+
+define <2 x double> @vfadd_vv_v2f64(<2 x double> %va, <2 x double> %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <2 x double> @llvm.vp.fadd.v2f64(<2 x double> %va, <2 x double> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+define <2 x double> @vfadd_vv_v2f64_unmasked(<2 x double> %va, <2 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fadd.v2f64(<2 x double> %va, <2 x double> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+define <2 x double> @vfadd_vf_v2f64(<2 x double> %va, double %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x double> undef, double %b, i32 0
+  %vb = shufflevector <2 x double> %elt.head, <2 x double> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fadd.v2f64(<2 x double> %va, <2 x double> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+define <2 x double> @vfadd_vf_v2f64_unmasked(<2 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x double> undef, double %b, i32 0
+  %vb = shufflevector <2 x double> %elt.head, <2 x double> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fadd.v2f64(<2 x double> %va, <2 x double> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+declare <4 x double> @llvm.vp.fadd.v4f64(<4 x double>, <4 x double>, <4 x i1>, i32)
+
+define <4 x double> @vfadd_vv_v4f64(<4 x double> %va, <4 x double> %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <4 x double> @llvm.vp.fadd.v4f64(<4 x double> %va, <4 x double> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+define <4 x double> @vfadd_vv_v4f64_unmasked(<4 x double> %va, <4 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fadd.v4f64(<4 x double> %va, <4 x double> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+define <4 x double> @vfadd_vf_v4f64(<4 x double> %va, double %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x double> undef, double %b, i32 0
+  %vb = shufflevector <4 x double> %elt.head, <4 x double> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fadd.v4f64(<4 x double> %va, <4 x double> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+define <4 x double> @vfadd_vf_v4f64_unmasked(<4 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x double> undef, double %b, i32 0
+  %vb = shufflevector <4 x double> %elt.head, <4 x double> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fadd.v4f64(<4 x double> %va, <4 x double> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+declare <8 x double> @llvm.vp.fadd.v8f64(<8 x double>, <8 x double>, <8 x i1>, i32)
+
+define <8 x double> @vfadd_vv_v8f64(<8 x double> %va, <8 x double> %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x double> @llvm.vp.fadd.v8f64(<8 x double> %va, <8 x double> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+define <8 x double> @vfadd_vv_v8f64_unmasked(<8 x double> %va, <8 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fadd.v8f64(<8 x double> %va, <8 x double> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+define <8 x double> @vfadd_vf_v8f64(<8 x double> %va, double %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x double> undef, double %b, i32 0
+  %vb = shufflevector <8 x double> %elt.head, <8 x double> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fadd.v8f64(<8 x double> %va, <8 x double> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+define <8 x double> @vfadd_vf_v8f64_unmasked(<8 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x double> undef, double %b, i32 0
+  %vb = shufflevector <8 x double> %elt.head, <8 x double> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fadd.v8f64(<8 x double> %va, <8 x double> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+declare <16 x double> @llvm.vp.fadd.v16f64(<16 x double>, <16 x double>, <16 x i1>, i32)
+
+define <16 x double> @vfadd_vv_v16f64(<16 x double> %va, <16 x double> %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v16f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <16 x double> @llvm.vp.fadd.v16f64(<16 x double> %va, <16 x double> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}
+
+define <16 x double> @vfadd_vv_v16f64_unmasked(<16 x double> %va, <16 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_v16f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fadd.v16f64(<16 x double> %va, <16 x double> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}
+
+define <16 x double> @vfadd_vf_v16f64(<16 x double> %va, double %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v16f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x double> undef, double %b, i32 0
+  %vb = shufflevector <16 x double> %elt.head, <16 x double> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fadd.v16f64(<16 x double> %va, <16 x double> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}
+
+define <16 x double> @vfadd_vf_v16f64_unmasked(<16 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_v16f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfadd.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x double> undef, double %b, i32 0
+  %vb = shufflevector <16 x double> %elt.head, <16 x double> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fadd.v16f64(<16 x double> %va, <16 x double> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-vfdiv-vp.ll

@@ -0,0 +1,629 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=ilp32d -riscv-v-vector-bits-min=128 \
+; RUN:   -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=lp64d -riscv-v-vector-bits-min=128 \
+; RUN:   -verify-machineinstrs < %s | FileCheck %s
+
+declare <2 x half> @llvm.vp.fdiv.v2f16(<2 x half>, <2 x half>, <2 x i1>, i32)
+
+define <2 x half> @vfdiv_vv_v2f16(<2 x half> %va, <2 x half> %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <2 x half> @llvm.vp.fdiv.v2f16(<2 x half> %va, <2 x half> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+define <2 x half> @vfdiv_vv_v2f16_unmasked(<2 x half> %va, <2 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fdiv.v2f16(<2 x half> %va, <2 x half> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+define <2 x half> @vfdiv_vf_v2f16(<2 x half> %va, half %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x half> undef, half %b, i32 0
+  %vb = shufflevector <2 x half> %elt.head, <2 x half> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fdiv.v2f16(<2 x half> %va, <2 x half> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+define <2 x half> @vfdiv_vf_v2f16_unmasked(<2 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x half> undef, half %b, i32 0
+  %vb = shufflevector <2 x half> %elt.head, <2 x half> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fdiv.v2f16(<2 x half> %va, <2 x half> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+declare <4 x half> @llvm.vp.fdiv.v4f16(<4 x half>, <4 x half>, <4 x i1>, i32)
+
+define <4 x half> @vfdiv_vv_v4f16(<4 x half> %va, <4 x half> %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <4 x half> @llvm.vp.fdiv.v4f16(<4 x half> %va, <4 x half> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+define <4 x half> @vfdiv_vv_v4f16_unmasked(<4 x half> %va, <4 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fdiv.v4f16(<4 x half> %va, <4 x half> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+define <4 x half> @vfdiv_vf_v4f16(<4 x half> %va, half %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x half> undef, half %b, i32 0
+  %vb = shufflevector <4 x half> %elt.head, <4 x half> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fdiv.v4f16(<4 x half> %va, <4 x half> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+define <4 x half> @vfdiv_vf_v4f16_unmasked(<4 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x half> undef, half %b, i32 0
+  %vb = shufflevector <4 x half> %elt.head, <4 x half> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fdiv.v4f16(<4 x half> %va, <4 x half> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+declare <8 x half> @llvm.vp.fdiv.v8f16(<8 x half>, <8 x half>, <8 x i1>, i32)
+
+define <8 x half> @vfdiv_vv_v8f16(<8 x half> %va, <8 x half> %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x half> @llvm.vp.fdiv.v8f16(<8 x half> %va, <8 x half> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+define <8 x half> @vfdiv_vv_v8f16_unmasked(<8 x half> %va, <8 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fdiv.v8f16(<8 x half> %va, <8 x half> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+define <8 x half> @vfdiv_vf_v8f16(<8 x half> %va, half %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x half> undef, half %b, i32 0
+  %vb = shufflevector <8 x half> %elt.head, <8 x half> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fdiv.v8f16(<8 x half> %va, <8 x half> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+define <8 x half> @vfdiv_vf_v8f16_unmasked(<8 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x half> undef, half %b, i32 0
+  %vb = shufflevector <8 x half> %elt.head, <8 x half> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fdiv.v8f16(<8 x half> %va, <8 x half> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+declare <16 x half> @llvm.vp.fdiv.v16f16(<16 x half>, <16 x half>, <16 x i1>, i32)
+
+define <16 x half> @vfdiv_vv_v16f16(<16 x half> %va, <16 x half> %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <16 x half> @llvm.vp.fdiv.v16f16(<16 x half> %va, <16 x half> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+define <16 x half> @vfdiv_vv_v16f16_unmasked(<16 x half> %va, <16 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fdiv.v16f16(<16 x half> %va, <16 x half> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+define <16 x half> @vfdiv_vf_v16f16(<16 x half> %va, half %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x half> undef, half %b, i32 0
+  %vb = shufflevector <16 x half> %elt.head, <16 x half> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fdiv.v16f16(<16 x half> %va, <16 x half> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+define <16 x half> @vfdiv_vf_v16f16_unmasked(<16 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x half> undef, half %b, i32 0
+  %vb = shufflevector <16 x half> %elt.head, <16 x half> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fdiv.v16f16(<16 x half> %va, <16 x half> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+declare <2 x float> @llvm.vp.fdiv.v2f32(<2 x float>, <2 x float>, <2 x i1>, i32)
+
+define <2 x float> @vfdiv_vv_v2f32(<2 x float> %va, <2 x float> %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <2 x float> @llvm.vp.fdiv.v2f32(<2 x float> %va, <2 x float> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+define <2 x float> @vfdiv_vv_v2f32_unmasked(<2 x float> %va, <2 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fdiv.v2f32(<2 x float> %va, <2 x float> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+define <2 x float> @vfdiv_vf_v2f32(<2 x float> %va, float %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x float> undef, float %b, i32 0
+  %vb = shufflevector <2 x float> %elt.head, <2 x float> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fdiv.v2f32(<2 x float> %va, <2 x float> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+define <2 x float> @vfdiv_vf_v2f32_unmasked(<2 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x float> undef, float %b, i32 0
+  %vb = shufflevector <2 x float> %elt.head, <2 x float> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fdiv.v2f32(<2 x float> %va, <2 x float> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+declare <4 x float> @llvm.vp.fdiv.v4f32(<4 x float>, <4 x float>, <4 x i1>, i32)
+
+define <4 x float> @vfdiv_vv_v4f32(<4 x float> %va, <4 x float> %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <4 x float> @llvm.vp.fdiv.v4f32(<4 x float> %va, <4 x float> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+define <4 x float> @vfdiv_vv_v4f32_unmasked(<4 x float> %va, <4 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fdiv.v4f32(<4 x float> %va, <4 x float> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+define <4 x float> @vfdiv_vf_v4f32(<4 x float> %va, float %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x float> undef, float %b, i32 0
+  %vb = shufflevector <4 x float> %elt.head, <4 x float> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fdiv.v4f32(<4 x float> %va, <4 x float> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+define <4 x float> @vfdiv_vf_v4f32_unmasked(<4 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x float> undef, float %b, i32 0
+  %vb = shufflevector <4 x float> %elt.head, <4 x float> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fdiv.v4f32(<4 x float> %va, <4 x float> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+declare <8 x float> @llvm.vp.fdiv.v8f32(<8 x float>, <8 x float>, <8 x i1>, i32)
+
+define <8 x float> @vfdiv_vv_v8f32(<8 x float> %va, <8 x float> %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x float> @llvm.vp.fdiv.v8f32(<8 x float> %va, <8 x float> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+define <8 x float> @vfdiv_vv_v8f32_unmasked(<8 x float> %va, <8 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fdiv.v8f32(<8 x float> %va, <8 x float> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+define <8 x float> @vfdiv_vf_v8f32(<8 x float> %va, float %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x float> undef, float %b, i32 0
+  %vb = shufflevector <8 x float> %elt.head, <8 x float> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fdiv.v8f32(<8 x float> %va, <8 x float> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+define <8 x float> @vfdiv_vf_v8f32_unmasked(<8 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x float> undef, float %b, i32 0
+  %vb = shufflevector <8 x float> %elt.head, <8 x float> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fdiv.v8f32(<8 x float> %va, <8 x float> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+declare <16 x float> @llvm.vp.fdiv.v16f32(<16 x float>, <16 x float>, <16 x i1>, i32)
+
+define <16 x float> @vfdiv_vv_v16f32(<16 x float> %va, <16 x float> %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <16 x float> @llvm.vp.fdiv.v16f32(<16 x float> %va, <16 x float> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+define <16 x float> @vfdiv_vv_v16f32_unmasked(<16 x float> %va, <16 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fdiv.v16f32(<16 x float> %va, <16 x float> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+define <16 x float> @vfdiv_vf_v16f32(<16 x float> %va, float %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x float> undef, float %b, i32 0
+  %vb = shufflevector <16 x float> %elt.head, <16 x float> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fdiv.v16f32(<16 x float> %va, <16 x float> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+define <16 x float> @vfdiv_vf_v16f32_unmasked(<16 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x float> undef, float %b, i32 0
+  %vb = shufflevector <16 x float> %elt.head, <16 x float> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fdiv.v16f32(<16 x float> %va, <16 x float> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+declare <2 x double> @llvm.vp.fdiv.v2f64(<2 x double>, <2 x double>, <2 x i1>, i32)
+
+define <2 x double> @vfdiv_vv_v2f64(<2 x double> %va, <2 x double> %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <2 x double> @llvm.vp.fdiv.v2f64(<2 x double> %va, <2 x double> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+define <2 x double> @vfdiv_vv_v2f64_unmasked(<2 x double> %va, <2 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fdiv.v2f64(<2 x double> %va, <2 x double> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+define <2 x double> @vfdiv_vf_v2f64(<2 x double> %va, double %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x double> undef, double %b, i32 0
+  %vb = shufflevector <2 x double> %elt.head, <2 x double> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fdiv.v2f64(<2 x double> %va, <2 x double> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+define <2 x double> @vfdiv_vf_v2f64_unmasked(<2 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x double> undef, double %b, i32 0
+  %vb = shufflevector <2 x double> %elt.head, <2 x double> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fdiv.v2f64(<2 x double> %va, <2 x double> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+declare <4 x double> @llvm.vp.fdiv.v4f64(<4 x double>, <4 x double>, <4 x i1>, i32)
+
+define <4 x double> @vfdiv_vv_v4f64(<4 x double> %va, <4 x double> %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <4 x double> @llvm.vp.fdiv.v4f64(<4 x double> %va, <4 x double> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+define <4 x double> @vfdiv_vv_v4f64_unmasked(<4 x double> %va, <4 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fdiv.v4f64(<4 x double> %va, <4 x double> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+define <4 x double> @vfdiv_vf_v4f64(<4 x double> %va, double %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x double> undef, double %b, i32 0
+  %vb = shufflevector <4 x double> %elt.head, <4 x double> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fdiv.v4f64(<4 x double> %va, <4 x double> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+define <4 x double> @vfdiv_vf_v4f64_unmasked(<4 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x double> undef, double %b, i32 0
+  %vb = shufflevector <4 x double> %elt.head, <4 x double> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fdiv.v4f64(<4 x double> %va, <4 x double> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+declare <8 x double> @llvm.vp.fdiv.v8f64(<8 x double>, <8 x double>, <8 x i1>, i32)
+
+define <8 x double> @vfdiv_vv_v8f64(<8 x double> %va, <8 x double> %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x double> @llvm.vp.fdiv.v8f64(<8 x double> %va, <8 x double> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+define <8 x double> @vfdiv_vv_v8f64_unmasked(<8 x double> %va, <8 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fdiv.v8f64(<8 x double> %va, <8 x double> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+define <8 x double> @vfdiv_vf_v8f64(<8 x double> %va, double %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x double> undef, double %b, i32 0
+  %vb = shufflevector <8 x double> %elt.head, <8 x double> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fdiv.v8f64(<8 x double> %va, <8 x double> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+define <8 x double> @vfdiv_vf_v8f64_unmasked(<8 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x double> undef, double %b, i32 0
+  %vb = shufflevector <8 x double> %elt.head, <8 x double> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fdiv.v8f64(<8 x double> %va, <8 x double> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+declare <16 x double> @llvm.vp.fdiv.v16f64(<16 x double>, <16 x double>, <16 x i1>, i32)
+
+define <16 x double> @vfdiv_vv_v16f64(<16 x double> %va, <16 x double> %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v16f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <16 x double> @llvm.vp.fdiv.v16f64(<16 x double> %va, <16 x double> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}
+
+define <16 x double> @vfdiv_vv_v16f64_unmasked(<16 x double> %va, <16 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_v16f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fdiv.v16f64(<16 x double> %va, <16 x double> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}
+
+define <16 x double> @vfdiv_vf_v16f64(<16 x double> %va, double %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v16f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x double> undef, double %b, i32 0
+  %vb = shufflevector <16 x double> %elt.head, <16 x double> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fdiv.v16f64(<16 x double> %va, <16 x double> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}
+
+define <16 x double> @vfdiv_vf_v16f64_unmasked(<16 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_v16f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x double> undef, double %b, i32 0
+  %vb = shufflevector <16 x double> %elt.head, <16 x double> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fdiv.v16f64(<16 x double> %va, <16 x double> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-vfmul-vp.ll

@@ -0,0 +1,629 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=ilp32d -riscv-v-vector-bits-min=128 \
+; RUN:   -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=lp64d -riscv-v-vector-bits-min=128 \
+; RUN:   -verify-machineinstrs < %s | FileCheck %s
+
+declare <2 x half> @llvm.vp.fmul.v2f16(<2 x half>, <2 x half>, <2 x i1>, i32)
+
+define <2 x half> @vfmul_vv_v2f16(<2 x half> %va, <2 x half> %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <2 x half> @llvm.vp.fmul.v2f16(<2 x half> %va, <2 x half> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+define <2 x half> @vfmul_vv_v2f16_unmasked(<2 x half> %va, <2 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fmul.v2f16(<2 x half> %va, <2 x half> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+define <2 x half> @vfmul_vf_v2f16(<2 x half> %va, half %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x half> undef, half %b, i32 0
+  %vb = shufflevector <2 x half> %elt.head, <2 x half> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fmul.v2f16(<2 x half> %va, <2 x half> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+define <2 x half> @vfmul_vf_v2f16_unmasked(<2 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x half> undef, half %b, i32 0
+  %vb = shufflevector <2 x half> %elt.head, <2 x half> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fmul.v2f16(<2 x half> %va, <2 x half> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+declare <4 x half> @llvm.vp.fmul.v4f16(<4 x half>, <4 x half>, <4 x i1>, i32)
+
+define <4 x half> @vfmul_vv_v4f16(<4 x half> %va, <4 x half> %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <4 x half> @llvm.vp.fmul.v4f16(<4 x half> %va, <4 x half> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+define <4 x half> @vfmul_vv_v4f16_unmasked(<4 x half> %va, <4 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fmul.v4f16(<4 x half> %va, <4 x half> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+define <4 x half> @vfmul_vf_v4f16(<4 x half> %va, half %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x half> undef, half %b, i32 0
+  %vb = shufflevector <4 x half> %elt.head, <4 x half> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fmul.v4f16(<4 x half> %va, <4 x half> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+define <4 x half> @vfmul_vf_v4f16_unmasked(<4 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x half> undef, half %b, i32 0
+  %vb = shufflevector <4 x half> %elt.head, <4 x half> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fmul.v4f16(<4 x half> %va, <4 x half> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+declare <8 x half> @llvm.vp.fmul.v8f16(<8 x half>, <8 x half>, <8 x i1>, i32)
+
+define <8 x half> @vfmul_vv_v8f16(<8 x half> %va, <8 x half> %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x half> @llvm.vp.fmul.v8f16(<8 x half> %va, <8 x half> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+define <8 x half> @vfmul_vv_v8f16_unmasked(<8 x half> %va, <8 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fmul.v8f16(<8 x half> %va, <8 x half> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+define <8 x half> @vfmul_vf_v8f16(<8 x half> %va, half %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x half> undef, half %b, i32 0
+  %vb = shufflevector <8 x half> %elt.head, <8 x half> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fmul.v8f16(<8 x half> %va, <8 x half> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+define <8 x half> @vfmul_vf_v8f16_unmasked(<8 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x half> undef, half %b, i32 0
+  %vb = shufflevector <8 x half> %elt.head, <8 x half> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fmul.v8f16(<8 x half> %va, <8 x half> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+declare <16 x half> @llvm.vp.fmul.v16f16(<16 x half>, <16 x half>, <16 x i1>, i32)
+
+define <16 x half> @vfmul_vv_v16f16(<16 x half> %va, <16 x half> %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <16 x half> @llvm.vp.fmul.v16f16(<16 x half> %va, <16 x half> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+define <16 x half> @vfmul_vv_v16f16_unmasked(<16 x half> %va, <16 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fmul.v16f16(<16 x half> %va, <16 x half> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+define <16 x half> @vfmul_vf_v16f16(<16 x half> %va, half %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x half> undef, half %b, i32 0
+  %vb = shufflevector <16 x half> %elt.head, <16 x half> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fmul.v16f16(<16 x half> %va, <16 x half> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+define <16 x half> @vfmul_vf_v16f16_unmasked(<16 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x half> undef, half %b, i32 0
+  %vb = shufflevector <16 x half> %elt.head, <16 x half> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fmul.v16f16(<16 x half> %va, <16 x half> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+declare <2 x float> @llvm.vp.fmul.v2f32(<2 x float>, <2 x float>, <2 x i1>, i32)
+
+define <2 x float> @vfmul_vv_v2f32(<2 x float> %va, <2 x float> %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <2 x float> @llvm.vp.fmul.v2f32(<2 x float> %va, <2 x float> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+define <2 x float> @vfmul_vv_v2f32_unmasked(<2 x float> %va, <2 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fmul.v2f32(<2 x float> %va, <2 x float> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+define <2 x float> @vfmul_vf_v2f32(<2 x float> %va, float %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x float> undef, float %b, i32 0
+  %vb = shufflevector <2 x float> %elt.head, <2 x float> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fmul.v2f32(<2 x float> %va, <2 x float> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+define <2 x float> @vfmul_vf_v2f32_unmasked(<2 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x float> undef, float %b, i32 0
+  %vb = shufflevector <2 x float> %elt.head, <2 x float> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fmul.v2f32(<2 x float> %va, <2 x float> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+declare <4 x float> @llvm.vp.fmul.v4f32(<4 x float>, <4 x float>, <4 x i1>, i32)
+
+define <4 x float> @vfmul_vv_v4f32(<4 x float> %va, <4 x float> %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <4 x float> @llvm.vp.fmul.v4f32(<4 x float> %va, <4 x float> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+define <4 x float> @vfmul_vv_v4f32_unmasked(<4 x float> %va, <4 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fmul.v4f32(<4 x float> %va, <4 x float> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+define <4 x float> @vfmul_vf_v4f32(<4 x float> %va, float %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x float> undef, float %b, i32 0
+  %vb = shufflevector <4 x float> %elt.head, <4 x float> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fmul.v4f32(<4 x float> %va, <4 x float> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+define <4 x float> @vfmul_vf_v4f32_unmasked(<4 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x float> undef, float %b, i32 0
+  %vb = shufflevector <4 x float> %elt.head, <4 x float> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fmul.v4f32(<4 x float> %va, <4 x float> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+declare <8 x float> @llvm.vp.fmul.v8f32(<8 x float>, <8 x float>, <8 x i1>, i32)
+
+define <8 x float> @vfmul_vv_v8f32(<8 x float> %va, <8 x float> %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x float> @llvm.vp.fmul.v8f32(<8 x float> %va, <8 x float> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+define <8 x float> @vfmul_vv_v8f32_unmasked(<8 x float> %va, <8 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fmul.v8f32(<8 x float> %va, <8 x float> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+define <8 x float> @vfmul_vf_v8f32(<8 x float> %va, float %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x float> undef, float %b, i32 0
+  %vb = shufflevector <8 x float> %elt.head, <8 x float> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fmul.v8f32(<8 x float> %va, <8 x float> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+define <8 x float> @vfmul_vf_v8f32_unmasked(<8 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x float> undef, float %b, i32 0
+  %vb = shufflevector <8 x float> %elt.head, <8 x float> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fmul.v8f32(<8 x float> %va, <8 x float> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+declare <16 x float> @llvm.vp.fmul.v16f32(<16 x float>, <16 x float>, <16 x i1>, i32)
+
+define <16 x float> @vfmul_vv_v16f32(<16 x float> %va, <16 x float> %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <16 x float> @llvm.vp.fmul.v16f32(<16 x float> %va, <16 x float> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+define <16 x float> @vfmul_vv_v16f32_unmasked(<16 x float> %va, <16 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fmul.v16f32(<16 x float> %va, <16 x float> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+define <16 x float> @vfmul_vf_v16f32(<16 x float> %va, float %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x float> undef, float %b, i32 0
+  %vb = shufflevector <16 x float> %elt.head, <16 x float> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fmul.v16f32(<16 x float> %va, <16 x float> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+define <16 x float> @vfmul_vf_v16f32_unmasked(<16 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x float> undef, float %b, i32 0
+  %vb = shufflevector <16 x float> %elt.head, <16 x float> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fmul.v16f32(<16 x float> %va, <16 x float> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+declare <2 x double> @llvm.vp.fmul.v2f64(<2 x double>, <2 x double>, <2 x i1>, i32)
+
+define <2 x double> @vfmul_vv_v2f64(<2 x double> %va, <2 x double> %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <2 x double> @llvm.vp.fmul.v2f64(<2 x double> %va, <2 x double> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+define <2 x double> @vfmul_vv_v2f64_unmasked(<2 x double> %va, <2 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fmul.v2f64(<2 x double> %va, <2 x double> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+define <2 x double> @vfmul_vf_v2f64(<2 x double> %va, double %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x double> undef, double %b, i32 0
+  %vb = shufflevector <2 x double> %elt.head, <2 x double> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fmul.v2f64(<2 x double> %va, <2 x double> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+define <2 x double> @vfmul_vf_v2f64_unmasked(<2 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x double> undef, double %b, i32 0
+  %vb = shufflevector <2 x double> %elt.head, <2 x double> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fmul.v2f64(<2 x double> %va, <2 x double> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+declare <4 x double> @llvm.vp.fmul.v4f64(<4 x double>, <4 x double>, <4 x i1>, i32)
+
+define <4 x double> @vfmul_vv_v4f64(<4 x double> %va, <4 x double> %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <4 x double> @llvm.vp.fmul.v4f64(<4 x double> %va, <4 x double> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+define <4 x double> @vfmul_vv_v4f64_unmasked(<4 x double> %va, <4 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fmul.v4f64(<4 x double> %va, <4 x double> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+define <4 x double> @vfmul_vf_v4f64(<4 x double> %va, double %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x double> undef, double %b, i32 0
+  %vb = shufflevector <4 x double> %elt.head, <4 x double> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fmul.v4f64(<4 x double> %va, <4 x double> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+define <4 x double> @vfmul_vf_v4f64_unmasked(<4 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x double> undef, double %b, i32 0
+  %vb = shufflevector <4 x double> %elt.head, <4 x double> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fmul.v4f64(<4 x double> %va, <4 x double> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+declare <8 x double> @llvm.vp.fmul.v8f64(<8 x double>, <8 x double>, <8 x i1>, i32)
+
+define <8 x double> @vfmul_vv_v8f64(<8 x double> %va, <8 x double> %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x double> @llvm.vp.fmul.v8f64(<8 x double> %va, <8 x double> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+define <8 x double> @vfmul_vv_v8f64_unmasked(<8 x double> %va, <8 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fmul.v8f64(<8 x double> %va, <8 x double> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+define <8 x double> @vfmul_vf_v8f64(<8 x double> %va, double %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x double> undef, double %b, i32 0
+  %vb = shufflevector <8 x double> %elt.head, <8 x double> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fmul.v8f64(<8 x double> %va, <8 x double> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+define <8 x double> @vfmul_vf_v8f64_unmasked(<8 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x double> undef, double %b, i32 0
+  %vb = shufflevector <8 x double> %elt.head, <8 x double> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fmul.v8f64(<8 x double> %va, <8 x double> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+declare <16 x double> @llvm.vp.fmul.v16f64(<16 x double>, <16 x double>, <16 x i1>, i32)
+
+define <16 x double> @vfmul_vv_v16f64(<16 x double> %va, <16 x double> %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v16f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <16 x double> @llvm.vp.fmul.v16f64(<16 x double> %va, <16 x double> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}
+
+define <16 x double> @vfmul_vv_v16f64_unmasked(<16 x double> %va, <16 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_v16f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fmul.v16f64(<16 x double> %va, <16 x double> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}
+
+define <16 x double> @vfmul_vf_v16f64(<16 x double> %va, double %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v16f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x double> undef, double %b, i32 0
+  %vb = shufflevector <16 x double> %elt.head, <16 x double> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fmul.v16f64(<16 x double> %va, <16 x double> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}
+
+define <16 x double> @vfmul_vf_v16f64_unmasked(<16 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_v16f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfmul.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x double> undef, double %b, i32 0
+  %vb = shufflevector <16 x double> %elt.head, <16 x double> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fmul.v16f64(<16 x double> %va, <16 x double> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-vfrdiv-vp.ll

@@ -0,0 +1,365 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=ilp32d -riscv-v-vector-bits-min=128 \
+; RUN:   -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=lp64d -riscv-v-vector-bits-min=128 \
+; RUN:   -verify-machineinstrs < %s | FileCheck %s
+
+declare <2 x half> @llvm.vp.fdiv.v2f16(<2 x half>, <2 x half>, <2 x i1>, i32)
+
+define <2 x half> @vfrdiv_vf_v2f16(<2 x half> %va, half %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x half> undef, half %b, i32 0
+  %vb = shufflevector <2 x half> %elt.head, <2 x half> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fdiv.v2f16(<2 x half> %vb, <2 x half> %va, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+define <2 x half> @vfrdiv_vf_v2f16_unmasked(<2 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfrdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x half> undef, half %b, i32 0
+  %vb = shufflevector <2 x half> %elt.head, <2 x half> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fdiv.v2f16(<2 x half> %vb, <2 x half> %va, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+declare <4 x half> @llvm.vp.fdiv.v4f16(<4 x half>, <4 x half>, <4 x i1>, i32)
+
+define <4 x half> @vfrdiv_vf_v4f16(<4 x half> %va, half %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x half> undef, half %b, i32 0
+  %vb = shufflevector <4 x half> %elt.head, <4 x half> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fdiv.v4f16(<4 x half> %vb, <4 x half> %va, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+define <4 x half> @vfrdiv_vf_v4f16_unmasked(<4 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfrdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x half> undef, half %b, i32 0
+  %vb = shufflevector <4 x half> %elt.head, <4 x half> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fdiv.v4f16(<4 x half> %vb, <4 x half> %va, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+declare <8 x half> @llvm.vp.fdiv.v8f16(<8 x half>, <8 x half>, <8 x i1>, i32)
+
+define <8 x half> @vfrdiv_vf_v8f16(<8 x half> %va, half %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x half> undef, half %b, i32 0
+  %vb = shufflevector <8 x half> %elt.head, <8 x half> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fdiv.v8f16(<8 x half> %vb, <8 x half> %va, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+define <8 x half> @vfrdiv_vf_v8f16_unmasked(<8 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfrdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x half> undef, half %b, i32 0
+  %vb = shufflevector <8 x half> %elt.head, <8 x half> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fdiv.v8f16(<8 x half> %vb, <8 x half> %va, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+declare <16 x half> @llvm.vp.fdiv.v16f16(<16 x half>, <16 x half>, <16 x i1>, i32)
+
+define <16 x half> @vfrdiv_vf_v16f16(<16 x half> %va, half %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v26, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x half> undef, half %b, i32 0
+  %vb = shufflevector <16 x half> %elt.head, <16 x half> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fdiv.v16f16(<16 x half> %vb, <16 x half> %va, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+define <16 x half> @vfrdiv_vf_v16f16_unmasked(<16 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfrdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x half> undef, half %b, i32 0
+  %vb = shufflevector <16 x half> %elt.head, <16 x half> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fdiv.v16f16(<16 x half> %vb, <16 x half> %va, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+declare <2 x float> @llvm.vp.fdiv.v2f32(<2 x float>, <2 x float>, <2 x i1>, i32)
+
+define <2 x float> @vfrdiv_vf_v2f32(<2 x float> %va, float %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x float> undef, float %b, i32 0
+  %vb = shufflevector <2 x float> %elt.head, <2 x float> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fdiv.v2f32(<2 x float> %vb, <2 x float> %va, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+define <2 x float> @vfrdiv_vf_v2f32_unmasked(<2 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfrdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x float> undef, float %b, i32 0
+  %vb = shufflevector <2 x float> %elt.head, <2 x float> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fdiv.v2f32(<2 x float> %vb, <2 x float> %va, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+declare <4 x float> @llvm.vp.fdiv.v4f32(<4 x float>, <4 x float>, <4 x i1>, i32)
+
+define <4 x float> @vfrdiv_vf_v4f32(<4 x float> %va, float %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x float> undef, float %b, i32 0
+  %vb = shufflevector <4 x float> %elt.head, <4 x float> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fdiv.v4f32(<4 x float> %vb, <4 x float> %va, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+define <4 x float> @vfrdiv_vf_v4f32_unmasked(<4 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfrdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x float> undef, float %b, i32 0
+  %vb = shufflevector <4 x float> %elt.head, <4 x float> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fdiv.v4f32(<4 x float> %vb, <4 x float> %va, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+declare <8 x float> @llvm.vp.fdiv.v8f32(<8 x float>, <8 x float>, <8 x i1>, i32)
+
+define <8 x float> @vfrdiv_vf_v8f32(<8 x float> %va, float %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v26, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x float> undef, float %b, i32 0
+  %vb = shufflevector <8 x float> %elt.head, <8 x float> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fdiv.v8f32(<8 x float> %vb, <8 x float> %va, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+define <8 x float> @vfrdiv_vf_v8f32_unmasked(<8 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfrdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x float> undef, float %b, i32 0
+  %vb = shufflevector <8 x float> %elt.head, <8 x float> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fdiv.v8f32(<8 x float> %vb, <8 x float> %va, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+declare <16 x float> @llvm.vp.fdiv.v16f32(<16 x float>, <16 x float>, <16 x i1>, i32)
+
+define <16 x float> @vfrdiv_vf_v16f32(<16 x float> %va, float %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v28, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x float> undef, float %b, i32 0
+  %vb = shufflevector <16 x float> %elt.head, <16 x float> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fdiv.v16f32(<16 x float> %vb, <16 x float> %va, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+define <16 x float> @vfrdiv_vf_v16f32_unmasked(<16 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfrdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x float> undef, float %b, i32 0
+  %vb = shufflevector <16 x float> %elt.head, <16 x float> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fdiv.v16f32(<16 x float> %vb, <16 x float> %va, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+declare <2 x double> @llvm.vp.fdiv.v2f64(<2 x double>, <2 x double>, <2 x i1>, i32)
+
+define <2 x double> @vfrdiv_vf_v2f64(<2 x double> %va, double %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x double> undef, double %b, i32 0
+  %vb = shufflevector <2 x double> %elt.head, <2 x double> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fdiv.v2f64(<2 x double> %vb, <2 x double> %va, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+define <2 x double> @vfrdiv_vf_v2f64_unmasked(<2 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfrdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x double> undef, double %b, i32 0
+  %vb = shufflevector <2 x double> %elt.head, <2 x double> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fdiv.v2f64(<2 x double> %vb, <2 x double> %va, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+declare <4 x double> @llvm.vp.fdiv.v4f64(<4 x double>, <4 x double>, <4 x i1>, i32)
+
+define <4 x double> @vfrdiv_vf_v4f64(<4 x double> %va, double %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v26, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x double> undef, double %b, i32 0
+  %vb = shufflevector <4 x double> %elt.head, <4 x double> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fdiv.v4f64(<4 x double> %vb, <4 x double> %va, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+define <4 x double> @vfrdiv_vf_v4f64_unmasked(<4 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfrdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x double> undef, double %b, i32 0
+  %vb = shufflevector <4 x double> %elt.head, <4 x double> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fdiv.v4f64(<4 x double> %vb, <4 x double> %va, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+declare <8 x double> @llvm.vp.fdiv.v8f64(<8 x double>, <8 x double>, <8 x i1>, i32)
+
+define <8 x double> @vfrdiv_vf_v8f64(<8 x double> %va, double %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v28, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x double> undef, double %b, i32 0
+  %vb = shufflevector <8 x double> %elt.head, <8 x double> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fdiv.v8f64(<8 x double> %vb, <8 x double> %va, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+define <8 x double> @vfrdiv_vf_v8f64_unmasked(<8 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfrdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x double> undef, double %b, i32 0
+  %vb = shufflevector <8 x double> %elt.head, <8 x double> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fdiv.v8f64(<8 x double> %vb, <8 x double> %va, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+declare <16 x double> @llvm.vp.fdiv.v16f64(<16 x double>, <16 x double>, <16 x i1>, i32)
+
+define <16 x double> @vfrdiv_vf_v16f64(<16 x double> %va, double %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v16f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v16, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x double> undef, double %b, i32 0
+  %vb = shufflevector <16 x double> %elt.head, <16 x double> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fdiv.v16f64(<16 x double> %vb, <16 x double> %va, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}
+
+define <16 x double> @vfrdiv_vf_v16f64_unmasked(<16 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_v16f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfrdiv.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x double> undef, double %b, i32 0
+  %vb = shufflevector <16 x double> %elt.head, <16 x double> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fdiv.v16f64(<16 x double> %vb, <16 x double> %va, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-vfrsub-vp.ll

@@ -0,0 +1,365 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=ilp32d -riscv-v-vector-bits-min=128 \
+; RUN:   -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=lp64d -riscv-v-vector-bits-min=128 \
+; RUN:   -verify-machineinstrs < %s | FileCheck %s
+
+declare <2 x half> @llvm.vp.fsub.v2f16(<2 x half>, <2 x half>, <2 x i1>, i32)
+
+define <2 x half> @vfrsub_vf_v2f16(<2 x half> %va, half %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x half> undef, half %b, i32 0
+  %vb = shufflevector <2 x half> %elt.head, <2 x half> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fsub.v2f16(<2 x half> %vb, <2 x half> %va, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+define <2 x half> @vfrsub_vf_v2f16_unmasked(<2 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x half> undef, half %b, i32 0
+  %vb = shufflevector <2 x half> %elt.head, <2 x half> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fsub.v2f16(<2 x half> %vb, <2 x half> %va, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+declare <4 x half> @llvm.vp.fsub.v4f16(<4 x half>, <4 x half>, <4 x i1>, i32)
+
+define <4 x half> @vfrsub_vf_v4f16(<4 x half> %va, half %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x half> undef, half %b, i32 0
+  %vb = shufflevector <4 x half> %elt.head, <4 x half> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fsub.v4f16(<4 x half> %vb, <4 x half> %va, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+define <4 x half> @vfrsub_vf_v4f16_unmasked(<4 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x half> undef, half %b, i32 0
+  %vb = shufflevector <4 x half> %elt.head, <4 x half> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fsub.v4f16(<4 x half> %vb, <4 x half> %va, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+declare <8 x half> @llvm.vp.fsub.v8f16(<8 x half>, <8 x half>, <8 x i1>, i32)
+
+define <8 x half> @vfrsub_vf_v8f16(<8 x half> %va, half %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x half> undef, half %b, i32 0
+  %vb = shufflevector <8 x half> %elt.head, <8 x half> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fsub.v8f16(<8 x half> %vb, <8 x half> %va, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+define <8 x half> @vfrsub_vf_v8f16_unmasked(<8 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x half> undef, half %b, i32 0
+  %vb = shufflevector <8 x half> %elt.head, <8 x half> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fsub.v8f16(<8 x half> %vb, <8 x half> %va, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+declare <16 x half> @llvm.vp.fsub.v16f16(<16 x half>, <16 x half>, <16 x i1>, i32)
+
+define <16 x half> @vfrsub_vf_v16f16(<16 x half> %va, half %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v26, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x half> undef, half %b, i32 0
+  %vb = shufflevector <16 x half> %elt.head, <16 x half> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fsub.v16f16(<16 x half> %vb, <16 x half> %va, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+define <16 x half> @vfrsub_vf_v16f16_unmasked(<16 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x half> undef, half %b, i32 0
+  %vb = shufflevector <16 x half> %elt.head, <16 x half> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fsub.v16f16(<16 x half> %vb, <16 x half> %va, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+declare <2 x float> @llvm.vp.fsub.v2f32(<2 x float>, <2 x float>, <2 x i1>, i32)
+
+define <2 x float> @vfrsub_vf_v2f32(<2 x float> %va, float %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x float> undef, float %b, i32 0
+  %vb = shufflevector <2 x float> %elt.head, <2 x float> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fsub.v2f32(<2 x float> %vb, <2 x float> %va, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+define <2 x float> @vfrsub_vf_v2f32_unmasked(<2 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x float> undef, float %b, i32 0
+  %vb = shufflevector <2 x float> %elt.head, <2 x float> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fsub.v2f32(<2 x float> %vb, <2 x float> %va, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+declare <4 x float> @llvm.vp.fsub.v4f32(<4 x float>, <4 x float>, <4 x i1>, i32)
+
+define <4 x float> @vfrsub_vf_v4f32(<4 x float> %va, float %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x float> undef, float %b, i32 0
+  %vb = shufflevector <4 x float> %elt.head, <4 x float> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fsub.v4f32(<4 x float> %vb, <4 x float> %va, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+define <4 x float> @vfrsub_vf_v4f32_unmasked(<4 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x float> undef, float %b, i32 0
+  %vb = shufflevector <4 x float> %elt.head, <4 x float> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fsub.v4f32(<4 x float> %vb, <4 x float> %va, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+declare <8 x float> @llvm.vp.fsub.v8f32(<8 x float>, <8 x float>, <8 x i1>, i32)
+
+define <8 x float> @vfrsub_vf_v8f32(<8 x float> %va, float %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v26, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x float> undef, float %b, i32 0
+  %vb = shufflevector <8 x float> %elt.head, <8 x float> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fsub.v8f32(<8 x float> %vb, <8 x float> %va, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+define <8 x float> @vfrsub_vf_v8f32_unmasked(<8 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x float> undef, float %b, i32 0
+  %vb = shufflevector <8 x float> %elt.head, <8 x float> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fsub.v8f32(<8 x float> %vb, <8 x float> %va, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+declare <16 x float> @llvm.vp.fsub.v16f32(<16 x float>, <16 x float>, <16 x i1>, i32)
+
+define <16 x float> @vfrsub_vf_v16f32(<16 x float> %va, float %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v28, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x float> undef, float %b, i32 0
+  %vb = shufflevector <16 x float> %elt.head, <16 x float> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fsub.v16f32(<16 x float> %vb, <16 x float> %va, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+define <16 x float> @vfrsub_vf_v16f32_unmasked(<16 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x float> undef, float %b, i32 0
+  %vb = shufflevector <16 x float> %elt.head, <16 x float> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fsub.v16f32(<16 x float> %vb, <16 x float> %va, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+declare <2 x double> @llvm.vp.fsub.v2f64(<2 x double>, <2 x double>, <2 x i1>, i32)
+
+define <2 x double> @vfrsub_vf_v2f64(<2 x double> %va, double %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x double> undef, double %b, i32 0
+  %vb = shufflevector <2 x double> %elt.head, <2 x double> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fsub.v2f64(<2 x double> %vb, <2 x double> %va, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+define <2 x double> @vfrsub_vf_v2f64_unmasked(<2 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x double> undef, double %b, i32 0
+  %vb = shufflevector <2 x double> %elt.head, <2 x double> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fsub.v2f64(<2 x double> %vb, <2 x double> %va, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+declare <4 x double> @llvm.vp.fsub.v4f64(<4 x double>, <4 x double>, <4 x i1>, i32)
+
+define <4 x double> @vfrsub_vf_v4f64(<4 x double> %va, double %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v26, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x double> undef, double %b, i32 0
+  %vb = shufflevector <4 x double> %elt.head, <4 x double> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fsub.v4f64(<4 x double> %vb, <4 x double> %va, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+define <4 x double> @vfrsub_vf_v4f64_unmasked(<4 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x double> undef, double %b, i32 0
+  %vb = shufflevector <4 x double> %elt.head, <4 x double> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fsub.v4f64(<4 x double> %vb, <4 x double> %va, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+declare <8 x double> @llvm.vp.fsub.v8f64(<8 x double>, <8 x double>, <8 x i1>, i32)
+
+define <8 x double> @vfrsub_vf_v8f64(<8 x double> %va, double %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v28, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x double> undef, double %b, i32 0
+  %vb = shufflevector <8 x double> %elt.head, <8 x double> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fsub.v8f64(<8 x double> %vb, <8 x double> %va, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+define <8 x double> @vfrsub_vf_v8f64_unmasked(<8 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x double> undef, double %b, i32 0
+  %vb = shufflevector <8 x double> %elt.head, <8 x double> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fsub.v8f64(<8 x double> %vb, <8 x double> %va, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+declare <16 x double> @llvm.vp.fsub.v16f64(<16 x double>, <16 x double>, <16 x i1>, i32)
+
+define <16 x double> @vfrsub_vf_v16f64(<16 x double> %va, double %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v16f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v16, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x double> undef, double %b, i32 0
+  %vb = shufflevector <16 x double> %elt.head, <16 x double> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fsub.v16f64(<16 x double> %vb, <16 x double> %va, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}
+
+define <16 x double> @vfrsub_vf_v16f64_unmasked(<16 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_v16f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfrsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x double> undef, double %b, i32 0
+  %vb = shufflevector <16 x double> %elt.head, <16 x double> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fsub.v16f64(<16 x double> %vb, <16 x double> %va, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}

llvm/test/CodeGen/RISCV/rvv/fixed-vectors-vfsub-vp.ll

@@ -0,0 +1,629 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=ilp32d -riscv-v-vector-bits-min=128 \
+; RUN:   -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=lp64d -riscv-v-vector-bits-min=128 \
+; RUN:   -verify-machineinstrs < %s | FileCheck %s
+
+declare <2 x half> @llvm.vp.fsub.v2f16(<2 x half>, <2 x half>, <2 x i1>, i32)
+
+define <2 x half> @vfsub_vv_v2f16(<2 x half> %va, <2 x half> %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <2 x half> @llvm.vp.fsub.v2f16(<2 x half> %va, <2 x half> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+define <2 x half> @vfsub_vv_v2f16_unmasked(<2 x half> %va, <2 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fsub.v2f16(<2 x half> %va, <2 x half> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+define <2 x half> @vfsub_vf_v2f16(<2 x half> %va, half %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x half> undef, half %b, i32 0
+  %vb = shufflevector <2 x half> %elt.head, <2 x half> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fsub.v2f16(<2 x half> %va, <2 x half> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+define <2 x half> @vfsub_vf_v2f16_unmasked(<2 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x half> undef, half %b, i32 0
+  %vb = shufflevector <2 x half> %elt.head, <2 x half> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x half> @llvm.vp.fsub.v2f16(<2 x half> %va, <2 x half> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x half> %v
+}
+
+declare <4 x half> @llvm.vp.fsub.v4f16(<4 x half>, <4 x half>, <4 x i1>, i32)
+
+define <4 x half> @vfsub_vv_v4f16(<4 x half> %va, <4 x half> %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <4 x half> @llvm.vp.fsub.v4f16(<4 x half> %va, <4 x half> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+define <4 x half> @vfsub_vv_v4f16_unmasked(<4 x half> %va, <4 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fsub.v4f16(<4 x half> %va, <4 x half> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+define <4 x half> @vfsub_vf_v4f16(<4 x half> %va, half %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x half> undef, half %b, i32 0
+  %vb = shufflevector <4 x half> %elt.head, <4 x half> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fsub.v4f16(<4 x half> %va, <4 x half> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+define <4 x half> @vfsub_vf_v4f16_unmasked(<4 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x half> undef, half %b, i32 0
+  %vb = shufflevector <4 x half> %elt.head, <4 x half> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x half> @llvm.vp.fsub.v4f16(<4 x half> %va, <4 x half> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x half> %v
+}
+
+declare <8 x half> @llvm.vp.fsub.v8f16(<8 x half>, <8 x half>, <8 x i1>, i32)
+
+define <8 x half> @vfsub_vv_v8f16(<8 x half> %va, <8 x half> %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x half> @llvm.vp.fsub.v8f16(<8 x half> %va, <8 x half> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+define <8 x half> @vfsub_vv_v8f16_unmasked(<8 x half> %va, <8 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fsub.v8f16(<8 x half> %va, <8 x half> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+define <8 x half> @vfsub_vf_v8f16(<8 x half> %va, half %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x half> undef, half %b, i32 0
+  %vb = shufflevector <8 x half> %elt.head, <8 x half> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fsub.v8f16(<8 x half> %va, <8 x half> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+define <8 x half> @vfsub_vf_v8f16_unmasked(<8 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x half> undef, half %b, i32 0
+  %vb = shufflevector <8 x half> %elt.head, <8 x half> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x half> @llvm.vp.fsub.v8f16(<8 x half> %va, <8 x half> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x half> %v
+}
+
+declare <16 x half> @llvm.vp.fsub.v16f16(<16 x half>, <16 x half>, <16 x i1>, i32)
+
+define <16 x half> @vfsub_vv_v16f16(<16 x half> %va, <16 x half> %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <16 x half> @llvm.vp.fsub.v16f16(<16 x half> %va, <16 x half> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+define <16 x half> @vfsub_vv_v16f16_unmasked(<16 x half> %va, <16 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fsub.v16f16(<16 x half> %va, <16 x half> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+define <16 x half> @vfsub_vf_v16f16(<16 x half> %va, half %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x half> undef, half %b, i32 0
+  %vb = shufflevector <16 x half> %elt.head, <16 x half> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fsub.v16f16(<16 x half> %va, <16 x half> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+define <16 x half> @vfsub_vf_v16f16_unmasked(<16 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x half> undef, half %b, i32 0
+  %vb = shufflevector <16 x half> %elt.head, <16 x half> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x half> @llvm.vp.fsub.v16f16(<16 x half> %va, <16 x half> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x half> %v
+}
+
+declare <2 x float> @llvm.vp.fsub.v2f32(<2 x float>, <2 x float>, <2 x i1>, i32)
+
+define <2 x float> @vfsub_vv_v2f32(<2 x float> %va, <2 x float> %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <2 x float> @llvm.vp.fsub.v2f32(<2 x float> %va, <2 x float> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+define <2 x float> @vfsub_vv_v2f32_unmasked(<2 x float> %va, <2 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fsub.v2f32(<2 x float> %va, <2 x float> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+define <2 x float> @vfsub_vf_v2f32(<2 x float> %va, float %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x float> undef, float %b, i32 0
+  %vb = shufflevector <2 x float> %elt.head, <2 x float> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fsub.v2f32(<2 x float> %va, <2 x float> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+define <2 x float> @vfsub_vf_v2f32_unmasked(<2 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x float> undef, float %b, i32 0
+  %vb = shufflevector <2 x float> %elt.head, <2 x float> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x float> @llvm.vp.fsub.v2f32(<2 x float> %va, <2 x float> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x float> %v
+}
+
+declare <4 x float> @llvm.vp.fsub.v4f32(<4 x float>, <4 x float>, <4 x i1>, i32)
+
+define <4 x float> @vfsub_vv_v4f32(<4 x float> %va, <4 x float> %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <4 x float> @llvm.vp.fsub.v4f32(<4 x float> %va, <4 x float> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+define <4 x float> @vfsub_vv_v4f32_unmasked(<4 x float> %va, <4 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fsub.v4f32(<4 x float> %va, <4 x float> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+define <4 x float> @vfsub_vf_v4f32(<4 x float> %va, float %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x float> undef, float %b, i32 0
+  %vb = shufflevector <4 x float> %elt.head, <4 x float> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fsub.v4f32(<4 x float> %va, <4 x float> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+define <4 x float> @vfsub_vf_v4f32_unmasked(<4 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x float> undef, float %b, i32 0
+  %vb = shufflevector <4 x float> %elt.head, <4 x float> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x float> @llvm.vp.fsub.v4f32(<4 x float> %va, <4 x float> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x float> %v
+}
+
+declare <8 x float> @llvm.vp.fsub.v8f32(<8 x float>, <8 x float>, <8 x i1>, i32)
+
+define <8 x float> @vfsub_vv_v8f32(<8 x float> %va, <8 x float> %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x float> @llvm.vp.fsub.v8f32(<8 x float> %va, <8 x float> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+define <8 x float> @vfsub_vv_v8f32_unmasked(<8 x float> %va, <8 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fsub.v8f32(<8 x float> %va, <8 x float> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+define <8 x float> @vfsub_vf_v8f32(<8 x float> %va, float %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x float> undef, float %b, i32 0
+  %vb = shufflevector <8 x float> %elt.head, <8 x float> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fsub.v8f32(<8 x float> %va, <8 x float> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+define <8 x float> @vfsub_vf_v8f32_unmasked(<8 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x float> undef, float %b, i32 0
+  %vb = shufflevector <8 x float> %elt.head, <8 x float> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x float> @llvm.vp.fsub.v8f32(<8 x float> %va, <8 x float> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x float> %v
+}
+
+declare <16 x float> @llvm.vp.fsub.v16f32(<16 x float>, <16 x float>, <16 x i1>, i32)
+
+define <16 x float> @vfsub_vv_v16f32(<16 x float> %va, <16 x float> %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <16 x float> @llvm.vp.fsub.v16f32(<16 x float> %va, <16 x float> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+define <16 x float> @vfsub_vv_v16f32_unmasked(<16 x float> %va, <16 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fsub.v16f32(<16 x float> %va, <16 x float> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+define <16 x float> @vfsub_vf_v16f32(<16 x float> %va, float %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x float> undef, float %b, i32 0
+  %vb = shufflevector <16 x float> %elt.head, <16 x float> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fsub.v16f32(<16 x float> %va, <16 x float> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+define <16 x float> @vfsub_vf_v16f32_unmasked(<16 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x float> undef, float %b, i32 0
+  %vb = shufflevector <16 x float> %elt.head, <16 x float> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x float> @llvm.vp.fsub.v16f32(<16 x float> %va, <16 x float> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x float> %v
+}
+
+declare <2 x double> @llvm.vp.fsub.v2f64(<2 x double>, <2 x double>, <2 x i1>, i32)
+
+define <2 x double> @vfsub_vv_v2f64(<2 x double> %va, <2 x double> %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <2 x double> @llvm.vp.fsub.v2f64(<2 x double> %va, <2 x double> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+define <2 x double> @vfsub_vv_v2f64_unmasked(<2 x double> %va, <2 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fsub.v2f64(<2 x double> %va, <2 x double> %b, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+define <2 x double> @vfsub_vf_v2f64(<2 x double> %va, double %b, <2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 2, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x double> undef, double %b, i32 0
+  %vb = shufflevector <2 x double> %elt.head, <2 x double> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fsub.v2f64(<2 x double> %va, <2 x double> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+define <2 x double> @vfsub_vf_v2f64_unmasked(<2 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <2 x double> undef, double %b, i32 0
+  %vb = shufflevector <2 x double> %elt.head, <2 x double> undef, <2 x i32> zeroinitializer
+  %head = insertelement <2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <2 x i1> %head, <2 x i1> undef, <2 x i32> zeroinitializer
+  %v = call <2 x double> @llvm.vp.fsub.v2f64(<2 x double> %va, <2 x double> %vb, <2 x i1> %m, i32 %evl)
+  ret <2 x double> %v
+}
+
+declare <4 x double> @llvm.vp.fsub.v4f64(<4 x double>, <4 x double>, <4 x i1>, i32)
+
+define <4 x double> @vfsub_vv_v4f64(<4 x double> %va, <4 x double> %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <4 x double> @llvm.vp.fsub.v4f64(<4 x double> %va, <4 x double> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+define <4 x double> @vfsub_vv_v4f64_unmasked(<4 x double> %va, <4 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fsub.v4f64(<4 x double> %va, <4 x double> %b, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+define <4 x double> @vfsub_vf_v4f64(<4 x double> %va, double %b, <4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 4, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x double> undef, double %b, i32 0
+  %vb = shufflevector <4 x double> %elt.head, <4 x double> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fsub.v4f64(<4 x double> %va, <4 x double> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+define <4 x double> @vfsub_vf_v4f64_unmasked(<4 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <4 x double> undef, double %b, i32 0
+  %vb = shufflevector <4 x double> %elt.head, <4 x double> undef, <4 x i32> zeroinitializer
+  %head = insertelement <4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <4 x i1> %head, <4 x i1> undef, <4 x i32> zeroinitializer
+  %v = call <4 x double> @llvm.vp.fsub.v4f64(<4 x double> %va, <4 x double> %vb, <4 x i1> %m, i32 %evl)
+  ret <4 x double> %v
+}
+
+declare <8 x double> @llvm.vp.fsub.v8f64(<8 x double>, <8 x double>, <8 x i1>, i32)
+
+define <8 x double> @vfsub_vv_v8f64(<8 x double> %va, <8 x double> %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <8 x double> @llvm.vp.fsub.v8f64(<8 x double> %va, <8 x double> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+define <8 x double> @vfsub_vv_v8f64_unmasked(<8 x double> %va, <8 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fsub.v8f64(<8 x double> %va, <8 x double> %b, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+define <8 x double> @vfsub_vf_v8f64(<8 x double> %va, double %b, <8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 8, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x double> undef, double %b, i32 0
+  %vb = shufflevector <8 x double> %elt.head, <8 x double> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fsub.v8f64(<8 x double> %va, <8 x double> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+define <8 x double> @vfsub_vf_v8f64_unmasked(<8 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <8 x double> undef, double %b, i32 0
+  %vb = shufflevector <8 x double> %elt.head, <8 x double> undef, <8 x i32> zeroinitializer
+  %head = insertelement <8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <8 x i1> %head, <8 x i1> undef, <8 x i32> zeroinitializer
+  %v = call <8 x double> @llvm.vp.fsub.v8f64(<8 x double> %va, <8 x double> %vb, <8 x i1> %m, i32 %evl)
+  ret <8 x double> %v
+}
+
+declare <16 x double> @llvm.vp.fsub.v16f64(<16 x double>, <16 x double>, <16 x i1>, i32)
+
+define <16 x double> @vfsub_vv_v16f64(<16 x double> %va, <16 x double> %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v16f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <16 x double> @llvm.vp.fsub.v16f64(<16 x double> %va, <16 x double> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}
+
+define <16 x double> @vfsub_vv_v16f64_unmasked(<16 x double> %va, <16 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_v16f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fsub.v16f64(<16 x double> %va, <16 x double> %b, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}
+
+define <16 x double> @vfsub_vf_v16f64(<16 x double> %va, double %b, <16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v16f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetivli zero, 16, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x double> undef, double %b, i32 0
+  %vb = shufflevector <16 x double> %elt.head, <16 x double> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fsub.v16f64(<16 x double> %va, <16 x double> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}
+
+define <16 x double> @vfsub_vf_v16f64_unmasked(<16 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_v16f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfsub.vf v8, v8, fa0
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <16 x double> undef, double %b, i32 0
+  %vb = shufflevector <16 x double> %elt.head, <16 x double> undef, <16 x i32> zeroinitializer
+  %head = insertelement <16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <16 x i1> %head, <16 x i1> undef, <16 x i32> zeroinitializer
+  %v = call <16 x double> @llvm.vp.fsub.v16f64(<16 x double> %va, <16 x double> %vb, <16 x i1> %m, i32 %evl)
+  ret <16 x double> %v
+}

llvm/test/CodeGen/RISCV/rvv/vfadd-vp.ll

@@ -0,0 +1,815 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x half> @llvm.vp.fadd.nxv1f16(<vscale x 1 x half>, <vscale x 1 x half>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x half> @vfadd_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x half> @llvm.vp.fadd.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+define <vscale x 1 x half> @vfadd_vv_nxv1f16_unmasked(<vscale x 1 x half> %va, <vscale x 1 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv1f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fadd.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+define <vscale x 1 x half> @vfadd_vf_nxv1f16(<vscale x 1 x half> %va, half %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fadd.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+define <vscale x 1 x half> @vfadd_vf_nxv1f16_unmasked(<vscale x 1 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv1f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fadd.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+declare <vscale x 2 x half> @llvm.vp.fadd.nxv2f16(<vscale x 2 x half>, <vscale x 2 x half>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x half> @vfadd_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x half> @llvm.vp.fadd.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+define <vscale x 2 x half> @vfadd_vv_nxv2f16_unmasked(<vscale x 2 x half> %va, <vscale x 2 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fadd.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+define <vscale x 2 x half> @vfadd_vf_nxv2f16(<vscale x 2 x half> %va, half %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fadd.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+define <vscale x 2 x half> @vfadd_vf_nxv2f16_unmasked(<vscale x 2 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fadd.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+declare <vscale x 4 x half> @llvm.vp.fadd.nxv4f16(<vscale x 4 x half>, <vscale x 4 x half>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x half> @vfadd_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 4 x half> @llvm.vp.fadd.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+define <vscale x 4 x half> @vfadd_vv_nxv4f16_unmasked(<vscale x 4 x half> %va, <vscale x 4 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fadd.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+define <vscale x 4 x half> @vfadd_vf_nxv4f16(<vscale x 4 x half> %va, half %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fadd.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+define <vscale x 4 x half> @vfadd_vf_nxv4f16_unmasked(<vscale x 4 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fadd.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+declare <vscale x 8 x half> @llvm.vp.fadd.nxv8f16(<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x half> @vfadd_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x half> @llvm.vp.fadd.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+define <vscale x 8 x half> @vfadd_vv_nxv8f16_unmasked(<vscale x 8 x half> %va, <vscale x 8 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fadd.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+define <vscale x 8 x half> @vfadd_vf_nxv8f16(<vscale x 8 x half> %va, half %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fadd.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+define <vscale x 8 x half> @vfadd_vf_nxv8f16_unmasked(<vscale x 8 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v26
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fadd.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+declare <vscale x 16 x half> @llvm.vp.fadd.nxv16f16(<vscale x 16 x half>, <vscale x 16 x half>, <vscale x 16 x i1>, i32)
+
+define <vscale x 16 x half> @vfadd_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 16 x half> @llvm.vp.fadd.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+define <vscale x 16 x half> @vfadd_vv_nxv16f16_unmasked(<vscale x 16 x half> %va, <vscale x 16 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fadd.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+define <vscale x 16 x half> @vfadd_vf_nxv16f16(<vscale x 16 x half> %va, half %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fadd.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+define <vscale x 16 x half> @vfadd_vf_nxv16f16_unmasked(<vscale x 16 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v28
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> undef, <vscale x 16 x i32> zeroinitializer
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fadd.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+declare <vscale x 32 x half> @llvm.vp.fadd.nxv32f16(<vscale x 32 x half>, <vscale x 32 x half>, <vscale x 32 x i1>, i32)
+
+define <vscale x 32 x half> @vfadd_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 32 x half> @llvm.vp.fadd.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+define <vscale x 32 x half> @vfadd_vv_nxv32f16_unmasked(<vscale x 32 x half> %va, <vscale x 32 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv32f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 32 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 32 x i1> %head, <vscale x 32 x i1> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fadd.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+define <vscale x 32 x half> @vfadd_vf_nxv32f16(<vscale x 32 x half> %va, half %b, <vscale x 32 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 32 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fadd.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+define <vscale x 32 x half> @vfadd_vf_nxv32f16_unmasked(<vscale x 32 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv32f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 32 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> undef, <vscale x 32 x i32> zeroinitializer
+  %head = insertelement <vscale x 32 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 32 x i1> %head, <vscale x 32 x i1> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fadd.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+declare <vscale x 1 x float> @llvm.vp.fadd.nxv1f32(<vscale x 1 x float>, <vscale x 1 x float>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x float> @vfadd_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x float> @llvm.vp.fadd.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+define <vscale x 1 x float> @vfadd_vv_nxv1f32_unmasked(<vscale x 1 x float> %va, <vscale x 1 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv1f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fadd.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+define <vscale x 1 x float> @vfadd_vf_nxv1f32(<vscale x 1 x float> %va, float %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fadd.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+define <vscale x 1 x float> @vfadd_vf_nxv1f32_unmasked(<vscale x 1 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv1f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fadd.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+declare <vscale x 2 x float> @llvm.vp.fadd.nxv2f32(<vscale x 2 x float>, <vscale x 2 x float>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x float> @vfadd_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x float> @llvm.vp.fadd.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+define <vscale x 2 x float> @vfadd_vv_nxv2f32_unmasked(<vscale x 2 x float> %va, <vscale x 2 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fadd.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+define <vscale x 2 x float> @vfadd_vf_nxv2f32(<vscale x 2 x float> %va, float %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fadd.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+define <vscale x 2 x float> @vfadd_vf_nxv2f32_unmasked(<vscale x 2 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fadd.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+declare <vscale x 4 x float> @llvm.vp.fadd.nxv4f32(<vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x float> @vfadd_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 4 x float> @llvm.vp.fadd.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+define <vscale x 4 x float> @vfadd_vv_nxv4f32_unmasked(<vscale x 4 x float> %va, <vscale x 4 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fadd.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+define <vscale x 4 x float> @vfadd_vf_nxv4f32(<vscale x 4 x float> %va, float %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fadd.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+define <vscale x 4 x float> @vfadd_vf_nxv4f32_unmasked(<vscale x 4 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v26
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fadd.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+declare <vscale x 8 x float> @llvm.vp.fadd.nxv8f32(<vscale x 8 x float>, <vscale x 8 x float>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x float> @vfadd_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x float> @llvm.vp.fadd.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+define <vscale x 8 x float> @vfadd_vv_nxv8f32_unmasked(<vscale x 8 x float> %va, <vscale x 8 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fadd.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+define <vscale x 8 x float> @vfadd_vf_nxv8f32(<vscale x 8 x float> %va, float %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fadd.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+define <vscale x 8 x float> @vfadd_vf_nxv8f32_unmasked(<vscale x 8 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v28
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fadd.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+declare <vscale x 16 x float> @llvm.vp.fadd.nxv16f32(<vscale x 16 x float>, <vscale x 16 x float>, <vscale x 16 x i1>, i32)
+
+define <vscale x 16 x float> @vfadd_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 16 x float> @llvm.vp.fadd.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+define <vscale x 16 x float> @vfadd_vv_nxv16f32_unmasked(<vscale x 16 x float> %va, <vscale x 16 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fadd.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+define <vscale x 16 x float> @vfadd_vf_nxv16f32(<vscale x 16 x float> %va, float %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fadd.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+define <vscale x 16 x float> @vfadd_vf_nxv16f32_unmasked(<vscale x 16 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> undef, <vscale x 16 x i32> zeroinitializer
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fadd.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+declare <vscale x 1 x double> @llvm.vp.fadd.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x double> @vfadd_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x double> @llvm.vp.fadd.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+define <vscale x 1 x double> @vfadd_vv_nxv1f64_unmasked(<vscale x 1 x double> %va, <vscale x 1 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv1f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fadd.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+define <vscale x 1 x double> @vfadd_vf_nxv1f64(<vscale x 1 x double> %va, double %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fadd.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+define <vscale x 1 x double> @vfadd_vf_nxv1f64_unmasked(<vscale x 1 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv1f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fadd.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+declare <vscale x 2 x double> @llvm.vp.fadd.nxv2f64(<vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x double> @vfadd_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x double> @llvm.vp.fadd.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+define <vscale x 2 x double> @vfadd_vv_nxv2f64_unmasked(<vscale x 2 x double> %va, <vscale x 2 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fadd.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+define <vscale x 2 x double> @vfadd_vf_nxv2f64(<vscale x 2 x double> %va, double %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fadd.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+define <vscale x 2 x double> @vfadd_vf_nxv2f64_unmasked(<vscale x 2 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v26
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fadd.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+declare <vscale x 4 x double> @llvm.vp.fadd.nxv4f64(<vscale x 4 x double>, <vscale x 4 x double>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x double> @vfadd_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 4 x double> @llvm.vp.fadd.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+define <vscale x 4 x double> @vfadd_vv_nxv4f64_unmasked(<vscale x 4 x double> %va, <vscale x 4 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fadd.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+define <vscale x 4 x double> @vfadd_vf_nxv4f64(<vscale x 4 x double> %va, double %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fadd.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+define <vscale x 4 x double> @vfadd_vf_nxv4f64_unmasked(<vscale x 4 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v28
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fadd.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+declare <vscale x 8 x double> @llvm.vp.fadd.nxv8f64(<vscale x 8 x double>, <vscale x 8 x double>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x double> @vfadd_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x double> @llvm.vp.fadd.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}
+
+define <vscale x 8 x double> @vfadd_vv_nxv8f64_unmasked(<vscale x 8 x double> %va, <vscale x 8 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vv_nxv8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fadd.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}
+
+define <vscale x 8 x double> @vfadd_vf_nxv8f64(<vscale x 8 x double> %va, double %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fadd.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}
+
+define <vscale x 8 x double> @vfadd_vf_nxv8f64_unmasked(<vscale x 8 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfadd_vf_nxv8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfadd.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fadd.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}

llvm/test/CodeGen/RISCV/rvv/vfdiv-vp.ll

@@ -0,0 +1,815 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x half> @llvm.vp.fdiv.nxv1f16(<vscale x 1 x half>, <vscale x 1 x half>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x half> @vfdiv_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x half> @llvm.vp.fdiv.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+define <vscale x 1 x half> @vfdiv_vv_nxv1f16_unmasked(<vscale x 1 x half> %va, <vscale x 1 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv1f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fdiv.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+define <vscale x 1 x half> @vfdiv_vf_nxv1f16(<vscale x 1 x half> %va, half %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fdiv.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+define <vscale x 1 x half> @vfdiv_vf_nxv1f16_unmasked(<vscale x 1 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv1f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fdiv.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+declare <vscale x 2 x half> @llvm.vp.fdiv.nxv2f16(<vscale x 2 x half>, <vscale x 2 x half>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x half> @vfdiv_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x half> @llvm.vp.fdiv.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+define <vscale x 2 x half> @vfdiv_vv_nxv2f16_unmasked(<vscale x 2 x half> %va, <vscale x 2 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fdiv.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+define <vscale x 2 x half> @vfdiv_vf_nxv2f16(<vscale x 2 x half> %va, half %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fdiv.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+define <vscale x 2 x half> @vfdiv_vf_nxv2f16_unmasked(<vscale x 2 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fdiv.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+declare <vscale x 4 x half> @llvm.vp.fdiv.nxv4f16(<vscale x 4 x half>, <vscale x 4 x half>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x half> @vfdiv_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 4 x half> @llvm.vp.fdiv.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+define <vscale x 4 x half> @vfdiv_vv_nxv4f16_unmasked(<vscale x 4 x half> %va, <vscale x 4 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fdiv.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+define <vscale x 4 x half> @vfdiv_vf_nxv4f16(<vscale x 4 x half> %va, half %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fdiv.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+define <vscale x 4 x half> @vfdiv_vf_nxv4f16_unmasked(<vscale x 4 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fdiv.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+declare <vscale x 8 x half> @llvm.vp.fdiv.nxv8f16(<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x half> @vfdiv_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x half> @llvm.vp.fdiv.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+define <vscale x 8 x half> @vfdiv_vv_nxv8f16_unmasked(<vscale x 8 x half> %va, <vscale x 8 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fdiv.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+define <vscale x 8 x half> @vfdiv_vf_nxv8f16(<vscale x 8 x half> %va, half %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fdiv.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+define <vscale x 8 x half> @vfdiv_vf_nxv8f16_unmasked(<vscale x 8 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v26
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fdiv.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+declare <vscale x 16 x half> @llvm.vp.fdiv.nxv16f16(<vscale x 16 x half>, <vscale x 16 x half>, <vscale x 16 x i1>, i32)
+
+define <vscale x 16 x half> @vfdiv_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 16 x half> @llvm.vp.fdiv.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+define <vscale x 16 x half> @vfdiv_vv_nxv16f16_unmasked(<vscale x 16 x half> %va, <vscale x 16 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fdiv.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+define <vscale x 16 x half> @vfdiv_vf_nxv16f16(<vscale x 16 x half> %va, half %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fdiv.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+define <vscale x 16 x half> @vfdiv_vf_nxv16f16_unmasked(<vscale x 16 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v28
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> undef, <vscale x 16 x i32> zeroinitializer
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fdiv.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+declare <vscale x 32 x half> @llvm.vp.fdiv.nxv32f16(<vscale x 32 x half>, <vscale x 32 x half>, <vscale x 32 x i1>, i32)
+
+define <vscale x 32 x half> @vfdiv_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 32 x half> @llvm.vp.fdiv.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+define <vscale x 32 x half> @vfdiv_vv_nxv32f16_unmasked(<vscale x 32 x half> %va, <vscale x 32 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv32f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 32 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 32 x i1> %head, <vscale x 32 x i1> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fdiv.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+define <vscale x 32 x half> @vfdiv_vf_nxv32f16(<vscale x 32 x half> %va, half %b, <vscale x 32 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 32 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fdiv.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+define <vscale x 32 x half> @vfdiv_vf_nxv32f16_unmasked(<vscale x 32 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv32f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 32 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> undef, <vscale x 32 x i32> zeroinitializer
+  %head = insertelement <vscale x 32 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 32 x i1> %head, <vscale x 32 x i1> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fdiv.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+declare <vscale x 1 x float> @llvm.vp.fdiv.nxv1f32(<vscale x 1 x float>, <vscale x 1 x float>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x float> @vfdiv_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x float> @llvm.vp.fdiv.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+define <vscale x 1 x float> @vfdiv_vv_nxv1f32_unmasked(<vscale x 1 x float> %va, <vscale x 1 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv1f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fdiv.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+define <vscale x 1 x float> @vfdiv_vf_nxv1f32(<vscale x 1 x float> %va, float %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fdiv.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+define <vscale x 1 x float> @vfdiv_vf_nxv1f32_unmasked(<vscale x 1 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv1f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fdiv.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+declare <vscale x 2 x float> @llvm.vp.fdiv.nxv2f32(<vscale x 2 x float>, <vscale x 2 x float>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x float> @vfdiv_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x float> @llvm.vp.fdiv.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+define <vscale x 2 x float> @vfdiv_vv_nxv2f32_unmasked(<vscale x 2 x float> %va, <vscale x 2 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fdiv.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+define <vscale x 2 x float> @vfdiv_vf_nxv2f32(<vscale x 2 x float> %va, float %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fdiv.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+define <vscale x 2 x float> @vfdiv_vf_nxv2f32_unmasked(<vscale x 2 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fdiv.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+declare <vscale x 4 x float> @llvm.vp.fdiv.nxv4f32(<vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x float> @vfdiv_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 4 x float> @llvm.vp.fdiv.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+define <vscale x 4 x float> @vfdiv_vv_nxv4f32_unmasked(<vscale x 4 x float> %va, <vscale x 4 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fdiv.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+define <vscale x 4 x float> @vfdiv_vf_nxv4f32(<vscale x 4 x float> %va, float %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fdiv.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+define <vscale x 4 x float> @vfdiv_vf_nxv4f32_unmasked(<vscale x 4 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v26
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fdiv.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+declare <vscale x 8 x float> @llvm.vp.fdiv.nxv8f32(<vscale x 8 x float>, <vscale x 8 x float>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x float> @vfdiv_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x float> @llvm.vp.fdiv.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+define <vscale x 8 x float> @vfdiv_vv_nxv8f32_unmasked(<vscale x 8 x float> %va, <vscale x 8 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fdiv.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+define <vscale x 8 x float> @vfdiv_vf_nxv8f32(<vscale x 8 x float> %va, float %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fdiv.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+define <vscale x 8 x float> @vfdiv_vf_nxv8f32_unmasked(<vscale x 8 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v28
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fdiv.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+declare <vscale x 16 x float> @llvm.vp.fdiv.nxv16f32(<vscale x 16 x float>, <vscale x 16 x float>, <vscale x 16 x i1>, i32)
+
+define <vscale x 16 x float> @vfdiv_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 16 x float> @llvm.vp.fdiv.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+define <vscale x 16 x float> @vfdiv_vv_nxv16f32_unmasked(<vscale x 16 x float> %va, <vscale x 16 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fdiv.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+define <vscale x 16 x float> @vfdiv_vf_nxv16f32(<vscale x 16 x float> %va, float %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fdiv.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+define <vscale x 16 x float> @vfdiv_vf_nxv16f32_unmasked(<vscale x 16 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> undef, <vscale x 16 x i32> zeroinitializer
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fdiv.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+declare <vscale x 1 x double> @llvm.vp.fdiv.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x double> @vfdiv_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x double> @llvm.vp.fdiv.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+define <vscale x 1 x double> @vfdiv_vv_nxv1f64_unmasked(<vscale x 1 x double> %va, <vscale x 1 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv1f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fdiv.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+define <vscale x 1 x double> @vfdiv_vf_nxv1f64(<vscale x 1 x double> %va, double %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fdiv.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+define <vscale x 1 x double> @vfdiv_vf_nxv1f64_unmasked(<vscale x 1 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv1f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fdiv.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+declare <vscale x 2 x double> @llvm.vp.fdiv.nxv2f64(<vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x double> @vfdiv_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x double> @llvm.vp.fdiv.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+define <vscale x 2 x double> @vfdiv_vv_nxv2f64_unmasked(<vscale x 2 x double> %va, <vscale x 2 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fdiv.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+define <vscale x 2 x double> @vfdiv_vf_nxv2f64(<vscale x 2 x double> %va, double %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fdiv.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+define <vscale x 2 x double> @vfdiv_vf_nxv2f64_unmasked(<vscale x 2 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v26
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fdiv.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+declare <vscale x 4 x double> @llvm.vp.fdiv.nxv4f64(<vscale x 4 x double>, <vscale x 4 x double>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x double> @vfdiv_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 4 x double> @llvm.vp.fdiv.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+define <vscale x 4 x double> @vfdiv_vv_nxv4f64_unmasked(<vscale x 4 x double> %va, <vscale x 4 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fdiv.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+define <vscale x 4 x double> @vfdiv_vf_nxv4f64(<vscale x 4 x double> %va, double %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fdiv.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+define <vscale x 4 x double> @vfdiv_vf_nxv4f64_unmasked(<vscale x 4 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v28
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fdiv.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+declare <vscale x 8 x double> @llvm.vp.fdiv.nxv8f64(<vscale x 8 x double>, <vscale x 8 x double>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x double> @vfdiv_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x double> @llvm.vp.fdiv.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}
+
+define <vscale x 8 x double> @vfdiv_vv_nxv8f64_unmasked(<vscale x 8 x double> %va, <vscale x 8 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vv_nxv8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fdiv.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}
+
+define <vscale x 8 x double> @vfdiv_vf_nxv8f64(<vscale x 8 x double> %va, double %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fdiv.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}
+
+define <vscale x 8 x double> @vfdiv_vf_nxv8f64_unmasked(<vscale x 8 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfdiv_vf_nxv8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fdiv.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}

llvm/test/CodeGen/RISCV/rvv/vfmul-vp.ll

@@ -0,0 +1,815 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x half> @llvm.vp.fmul.nxv1f16(<vscale x 1 x half>, <vscale x 1 x half>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x half> @vfmul_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x half> @llvm.vp.fmul.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+define <vscale x 1 x half> @vfmul_vv_nxv1f16_unmasked(<vscale x 1 x half> %va, <vscale x 1 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv1f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fmul.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+define <vscale x 1 x half> @vfmul_vf_nxv1f16(<vscale x 1 x half> %va, half %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fmul.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+define <vscale x 1 x half> @vfmul_vf_nxv1f16_unmasked(<vscale x 1 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv1f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fmul.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+declare <vscale x 2 x half> @llvm.vp.fmul.nxv2f16(<vscale x 2 x half>, <vscale x 2 x half>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x half> @vfmul_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x half> @llvm.vp.fmul.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+define <vscale x 2 x half> @vfmul_vv_nxv2f16_unmasked(<vscale x 2 x half> %va, <vscale x 2 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fmul.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+define <vscale x 2 x half> @vfmul_vf_nxv2f16(<vscale x 2 x half> %va, half %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fmul.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+define <vscale x 2 x half> @vfmul_vf_nxv2f16_unmasked(<vscale x 2 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fmul.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+declare <vscale x 4 x half> @llvm.vp.fmul.nxv4f16(<vscale x 4 x half>, <vscale x 4 x half>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x half> @vfmul_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 4 x half> @llvm.vp.fmul.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+define <vscale x 4 x half> @vfmul_vv_nxv4f16_unmasked(<vscale x 4 x half> %va, <vscale x 4 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fmul.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+define <vscale x 4 x half> @vfmul_vf_nxv4f16(<vscale x 4 x half> %va, half %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fmul.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+define <vscale x 4 x half> @vfmul_vf_nxv4f16_unmasked(<vscale x 4 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fmul.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+declare <vscale x 8 x half> @llvm.vp.fmul.nxv8f16(<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x half> @vfmul_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x half> @llvm.vp.fmul.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+define <vscale x 8 x half> @vfmul_vv_nxv8f16_unmasked(<vscale x 8 x half> %va, <vscale x 8 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fmul.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+define <vscale x 8 x half> @vfmul_vf_nxv8f16(<vscale x 8 x half> %va, half %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fmul.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+define <vscale x 8 x half> @vfmul_vf_nxv8f16_unmasked(<vscale x 8 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v26
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fmul.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+declare <vscale x 16 x half> @llvm.vp.fmul.nxv16f16(<vscale x 16 x half>, <vscale x 16 x half>, <vscale x 16 x i1>, i32)
+
+define <vscale x 16 x half> @vfmul_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 16 x half> @llvm.vp.fmul.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+define <vscale x 16 x half> @vfmul_vv_nxv16f16_unmasked(<vscale x 16 x half> %va, <vscale x 16 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fmul.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+define <vscale x 16 x half> @vfmul_vf_nxv16f16(<vscale x 16 x half> %va, half %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fmul.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+define <vscale x 16 x half> @vfmul_vf_nxv16f16_unmasked(<vscale x 16 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v28
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> undef, <vscale x 16 x i32> zeroinitializer
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fmul.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+declare <vscale x 32 x half> @llvm.vp.fmul.nxv32f16(<vscale x 32 x half>, <vscale x 32 x half>, <vscale x 32 x i1>, i32)
+
+define <vscale x 32 x half> @vfmul_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 32 x half> @llvm.vp.fmul.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+define <vscale x 32 x half> @vfmul_vv_nxv32f16_unmasked(<vscale x 32 x half> %va, <vscale x 32 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv32f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 32 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 32 x i1> %head, <vscale x 32 x i1> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fmul.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+define <vscale x 32 x half> @vfmul_vf_nxv32f16(<vscale x 32 x half> %va, half %b, <vscale x 32 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 32 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fmul.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+define <vscale x 32 x half> @vfmul_vf_nxv32f16_unmasked(<vscale x 32 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv32f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 32 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> undef, <vscale x 32 x i32> zeroinitializer
+  %head = insertelement <vscale x 32 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 32 x i1> %head, <vscale x 32 x i1> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fmul.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+declare <vscale x 1 x float> @llvm.vp.fmul.nxv1f32(<vscale x 1 x float>, <vscale x 1 x float>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x float> @vfmul_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x float> @llvm.vp.fmul.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+define <vscale x 1 x float> @vfmul_vv_nxv1f32_unmasked(<vscale x 1 x float> %va, <vscale x 1 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv1f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fmul.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+define <vscale x 1 x float> @vfmul_vf_nxv1f32(<vscale x 1 x float> %va, float %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fmul.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+define <vscale x 1 x float> @vfmul_vf_nxv1f32_unmasked(<vscale x 1 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv1f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fmul.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+declare <vscale x 2 x float> @llvm.vp.fmul.nxv2f32(<vscale x 2 x float>, <vscale x 2 x float>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x float> @vfmul_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x float> @llvm.vp.fmul.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+define <vscale x 2 x float> @vfmul_vv_nxv2f32_unmasked(<vscale x 2 x float> %va, <vscale x 2 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fmul.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+define <vscale x 2 x float> @vfmul_vf_nxv2f32(<vscale x 2 x float> %va, float %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fmul.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+define <vscale x 2 x float> @vfmul_vf_nxv2f32_unmasked(<vscale x 2 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fmul.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+declare <vscale x 4 x float> @llvm.vp.fmul.nxv4f32(<vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x float> @vfmul_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 4 x float> @llvm.vp.fmul.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+define <vscale x 4 x float> @vfmul_vv_nxv4f32_unmasked(<vscale x 4 x float> %va, <vscale x 4 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fmul.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+define <vscale x 4 x float> @vfmul_vf_nxv4f32(<vscale x 4 x float> %va, float %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fmul.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+define <vscale x 4 x float> @vfmul_vf_nxv4f32_unmasked(<vscale x 4 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v26
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fmul.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+declare <vscale x 8 x float> @llvm.vp.fmul.nxv8f32(<vscale x 8 x float>, <vscale x 8 x float>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x float> @vfmul_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x float> @llvm.vp.fmul.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+define <vscale x 8 x float> @vfmul_vv_nxv8f32_unmasked(<vscale x 8 x float> %va, <vscale x 8 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fmul.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+define <vscale x 8 x float> @vfmul_vf_nxv8f32(<vscale x 8 x float> %va, float %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fmul.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+define <vscale x 8 x float> @vfmul_vf_nxv8f32_unmasked(<vscale x 8 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v28
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fmul.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+declare <vscale x 16 x float> @llvm.vp.fmul.nxv16f32(<vscale x 16 x float>, <vscale x 16 x float>, <vscale x 16 x i1>, i32)
+
+define <vscale x 16 x float> @vfmul_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 16 x float> @llvm.vp.fmul.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+define <vscale x 16 x float> @vfmul_vv_nxv16f32_unmasked(<vscale x 16 x float> %va, <vscale x 16 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fmul.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+define <vscale x 16 x float> @vfmul_vf_nxv16f32(<vscale x 16 x float> %va, float %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fmul.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+define <vscale x 16 x float> @vfmul_vf_nxv16f32_unmasked(<vscale x 16 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> undef, <vscale x 16 x i32> zeroinitializer
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fmul.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+declare <vscale x 1 x double> @llvm.vp.fmul.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x double> @vfmul_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x double> @llvm.vp.fmul.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+define <vscale x 1 x double> @vfmul_vv_nxv1f64_unmasked(<vscale x 1 x double> %va, <vscale x 1 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv1f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fmul.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+define <vscale x 1 x double> @vfmul_vf_nxv1f64(<vscale x 1 x double> %va, double %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fmul.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+define <vscale x 1 x double> @vfmul_vf_nxv1f64_unmasked(<vscale x 1 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv1f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fmul.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+declare <vscale x 2 x double> @llvm.vp.fmul.nxv2f64(<vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x double> @vfmul_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x double> @llvm.vp.fmul.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+define <vscale x 2 x double> @vfmul_vv_nxv2f64_unmasked(<vscale x 2 x double> %va, <vscale x 2 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fmul.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+define <vscale x 2 x double> @vfmul_vf_nxv2f64(<vscale x 2 x double> %va, double %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fmul.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+define <vscale x 2 x double> @vfmul_vf_nxv2f64_unmasked(<vscale x 2 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v26
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fmul.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+declare <vscale x 4 x double> @llvm.vp.fmul.nxv4f64(<vscale x 4 x double>, <vscale x 4 x double>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x double> @vfmul_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 4 x double> @llvm.vp.fmul.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+define <vscale x 4 x double> @vfmul_vv_nxv4f64_unmasked(<vscale x 4 x double> %va, <vscale x 4 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fmul.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+define <vscale x 4 x double> @vfmul_vf_nxv4f64(<vscale x 4 x double> %va, double %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fmul.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+define <vscale x 4 x double> @vfmul_vf_nxv4f64_unmasked(<vscale x 4 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v28
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fmul.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+declare <vscale x 8 x double> @llvm.vp.fmul.nxv8f64(<vscale x 8 x double>, <vscale x 8 x double>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x double> @vfmul_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x double> @llvm.vp.fmul.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}
+
+define <vscale x 8 x double> @vfmul_vv_nxv8f64_unmasked(<vscale x 8 x double> %va, <vscale x 8 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vv_nxv8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fmul.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}
+
+define <vscale x 8 x double> @vfmul_vf_nxv8f64(<vscale x 8 x double> %va, double %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fmul.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}
+
+define <vscale x 8 x double> @vfmul_vf_nxv8f64_unmasked(<vscale x 8 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfmul_vf_nxv8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfmul.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fmul.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}

llvm/test/CodeGen/RISCV/rvv/vfrdiv-vp.ll

@@ -0,0 +1,485 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x half> @llvm.vp.fdiv.nxv1f16(<vscale x 1 x half>, <vscale x 1 x half>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x half> @vfrdiv_vf_nxv1f16(<vscale x 1 x half> %va, half %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fdiv.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x half> %va, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+define <vscale x 1 x half> @vfrdiv_vf_nxv1f16_unmasked(<vscale x 1 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv1f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fdiv.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x half> %va, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+declare <vscale x 2 x half> @llvm.vp.fdiv.nxv2f16(<vscale x 2 x half>, <vscale x 2 x half>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x half> @vfrdiv_vf_nxv2f16(<vscale x 2 x half> %va, half %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fdiv.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x half> %va, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+define <vscale x 2 x half> @vfrdiv_vf_nxv2f16_unmasked(<vscale x 2 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fdiv.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x half> %va, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+declare <vscale x 4 x half> @llvm.vp.fdiv.nxv4f16(<vscale x 4 x half>, <vscale x 4 x half>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x half> @vfrdiv_vf_nxv4f16(<vscale x 4 x half> %va, half %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fdiv.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x half> %va, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+define <vscale x 4 x half> @vfrdiv_vf_nxv4f16_unmasked(<vscale x 4 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fdiv.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x half> %va, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+declare <vscale x 8 x half> @llvm.vp.fdiv.nxv8f16(<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x half> @vfrdiv_vf_nxv8f16(<vscale x 8 x half> %va, half %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v26, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fdiv.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %va, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+define <vscale x 8 x half> @vfrdiv_vf_nxv8f16_unmasked(<vscale x 8 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v26, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fdiv.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %va, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+declare <vscale x 16 x half> @llvm.vp.fdiv.nxv16f16(<vscale x 16 x half>, <vscale x 16 x half>, <vscale x 16 x i1>, i32)
+
+define <vscale x 16 x half> @vfrdiv_vf_nxv16f16(<vscale x 16 x half> %va, half %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v28, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fdiv.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x half> %va, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+define <vscale x 16 x half> @vfrdiv_vf_nxv16f16_unmasked(<vscale x 16 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v28, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> undef, <vscale x 16 x i32> zeroinitializer
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fdiv.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x half> %va, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+declare <vscale x 32 x half> @llvm.vp.fdiv.nxv32f16(<vscale x 32 x half>, <vscale x 32 x half>, <vscale x 32 x i1>, i32)
+
+define <vscale x 32 x half> @vfrdiv_vf_nxv32f16(<vscale x 32 x half> %va, half %b, <vscale x 32 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v16, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 32 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fdiv.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x half> %va, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+define <vscale x 32 x half> @vfrdiv_vf_nxv32f16_unmasked(<vscale x 32 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv32f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v16, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 32 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> undef, <vscale x 32 x i32> zeroinitializer
+  %head = insertelement <vscale x 32 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 32 x i1> %head, <vscale x 32 x i1> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fdiv.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x half> %va, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+declare <vscale x 1 x float> @llvm.vp.fdiv.nxv1f32(<vscale x 1 x float>, <vscale x 1 x float>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x float> @vfrdiv_vf_nxv1f32(<vscale x 1 x float> %va, float %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fdiv.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x float> %va, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+define <vscale x 1 x float> @vfrdiv_vf_nxv1f32_unmasked(<vscale x 1 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv1f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fdiv.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x float> %va, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+declare <vscale x 2 x float> @llvm.vp.fdiv.nxv2f32(<vscale x 2 x float>, <vscale x 2 x float>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x float> @vfrdiv_vf_nxv2f32(<vscale x 2 x float> %va, float %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fdiv.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x float> %va, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+define <vscale x 2 x float> @vfrdiv_vf_nxv2f32_unmasked(<vscale x 2 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fdiv.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x float> %va, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+declare <vscale x 4 x float> @llvm.vp.fdiv.nxv4f32(<vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x float> @vfrdiv_vf_nxv4f32(<vscale x 4 x float> %va, float %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v26, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fdiv.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %va, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+define <vscale x 4 x float> @vfrdiv_vf_nxv4f32_unmasked(<vscale x 4 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v26, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fdiv.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %va, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+declare <vscale x 8 x float> @llvm.vp.fdiv.nxv8f32(<vscale x 8 x float>, <vscale x 8 x float>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x float> @vfrdiv_vf_nxv8f32(<vscale x 8 x float> %va, float %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v28, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fdiv.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %va, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+define <vscale x 8 x float> @vfrdiv_vf_nxv8f32_unmasked(<vscale x 8 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v28, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fdiv.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %va, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+declare <vscale x 16 x float> @llvm.vp.fdiv.nxv16f32(<vscale x 16 x float>, <vscale x 16 x float>, <vscale x 16 x i1>, i32)
+
+define <vscale x 16 x float> @vfrdiv_vf_nxv16f32(<vscale x 16 x float> %va, float %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v16, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fdiv.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x float> %va, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+define <vscale x 16 x float> @vfrdiv_vf_nxv16f32_unmasked(<vscale x 16 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v16, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> undef, <vscale x 16 x i32> zeroinitializer
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fdiv.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x float> %va, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+declare <vscale x 1 x double> @llvm.vp.fdiv.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x double> @vfrdiv_vf_nxv1f64(<vscale x 1 x double> %va, double %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fdiv.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x double> %va, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+define <vscale x 1 x double> @vfrdiv_vf_nxv1f64_unmasked(<vscale x 1 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv1f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v25, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fdiv.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x double> %va, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+declare <vscale x 2 x double> @llvm.vp.fdiv.nxv2f64(<vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x double> @vfrdiv_vf_nxv2f64(<vscale x 2 x double> %va, double %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v26, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fdiv.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x double> %va, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+define <vscale x 2 x double> @vfrdiv_vf_nxv2f64_unmasked(<vscale x 2 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v26, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fdiv.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x double> %va, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+declare <vscale x 4 x double> @llvm.vp.fdiv.nxv4f64(<vscale x 4 x double>, <vscale x 4 x double>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x double> @vfrdiv_vf_nxv4f64(<vscale x 4 x double> %va, double %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v28, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fdiv.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x double> %va, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+define <vscale x 4 x double> @vfrdiv_vf_nxv4f64_unmasked(<vscale x 4 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v28, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fdiv.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x double> %va, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+declare <vscale x 8 x double> @llvm.vp.fdiv.nxv8f64(<vscale x 8 x double>, <vscale x 8 x double>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x double> @vfrdiv_vf_nxv8f64(<vscale x 8 x double> %va, double %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v16, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fdiv.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %va, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}
+
+define <vscale x 8 x double> @vfrdiv_vf_nxv8f64_unmasked(<vscale x 8 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrdiv_vf_nxv8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfdiv.vv v8, v16, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fdiv.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %va, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}

llvm/test/CodeGen/RISCV/rvv/vfrsub-vp.ll

@@ -0,0 +1,485 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x half> @llvm.vp.fsub.nxv1f16(<vscale x 1 x half>, <vscale x 1 x half>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x half> @vfrsub_vf_nxv1f16(<vscale x 1 x half> %va, half %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fsub.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x half> %va, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+define <vscale x 1 x half> @vfrsub_vf_nxv1f16_unmasked(<vscale x 1 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv1f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fsub.nxv1f16(<vscale x 1 x half> %vb, <vscale x 1 x half> %va, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+declare <vscale x 2 x half> @llvm.vp.fsub.nxv2f16(<vscale x 2 x half>, <vscale x 2 x half>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x half> @vfrsub_vf_nxv2f16(<vscale x 2 x half> %va, half %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fsub.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x half> %va, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+define <vscale x 2 x half> @vfrsub_vf_nxv2f16_unmasked(<vscale x 2 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fsub.nxv2f16(<vscale x 2 x half> %vb, <vscale x 2 x half> %va, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+declare <vscale x 4 x half> @llvm.vp.fsub.nxv4f16(<vscale x 4 x half>, <vscale x 4 x half>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x half> @vfrsub_vf_nxv4f16(<vscale x 4 x half> %va, half %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fsub.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x half> %va, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+define <vscale x 4 x half> @vfrsub_vf_nxv4f16_unmasked(<vscale x 4 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fsub.nxv4f16(<vscale x 4 x half> %vb, <vscale x 4 x half> %va, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+declare <vscale x 8 x half> @llvm.vp.fsub.nxv8f16(<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x half> @vfrsub_vf_nxv8f16(<vscale x 8 x half> %va, half %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v26, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fsub.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %va, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+define <vscale x 8 x half> @vfrsub_vf_nxv8f16_unmasked(<vscale x 8 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v26, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fsub.nxv8f16(<vscale x 8 x half> %vb, <vscale x 8 x half> %va, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+declare <vscale x 16 x half> @llvm.vp.fsub.nxv16f16(<vscale x 16 x half>, <vscale x 16 x half>, <vscale x 16 x i1>, i32)
+
+define <vscale x 16 x half> @vfrsub_vf_nxv16f16(<vscale x 16 x half> %va, half %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v28, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fsub.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x half> %va, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+define <vscale x 16 x half> @vfrsub_vf_nxv16f16_unmasked(<vscale x 16 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v28, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> undef, <vscale x 16 x i32> zeroinitializer
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fsub.nxv16f16(<vscale x 16 x half> %vb, <vscale x 16 x half> %va, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+declare <vscale x 32 x half> @llvm.vp.fsub.nxv32f16(<vscale x 32 x half>, <vscale x 32 x half>, <vscale x 32 x i1>, i32)
+
+define <vscale x 32 x half> @vfrsub_vf_nxv32f16(<vscale x 32 x half> %va, half %b, <vscale x 32 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v16, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 32 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fsub.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x half> %va, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+define <vscale x 32 x half> @vfrsub_vf_nxv32f16_unmasked(<vscale x 32 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv32f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v16, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 32 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> undef, <vscale x 32 x i32> zeroinitializer
+  %head = insertelement <vscale x 32 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 32 x i1> %head, <vscale x 32 x i1> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fsub.nxv32f16(<vscale x 32 x half> %vb, <vscale x 32 x half> %va, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+declare <vscale x 1 x float> @llvm.vp.fsub.nxv1f32(<vscale x 1 x float>, <vscale x 1 x float>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x float> @vfrsub_vf_nxv1f32(<vscale x 1 x float> %va, float %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fsub.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x float> %va, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+define <vscale x 1 x float> @vfrsub_vf_nxv1f32_unmasked(<vscale x 1 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv1f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fsub.nxv1f32(<vscale x 1 x float> %vb, <vscale x 1 x float> %va, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+declare <vscale x 2 x float> @llvm.vp.fsub.nxv2f32(<vscale x 2 x float>, <vscale x 2 x float>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x float> @vfrsub_vf_nxv2f32(<vscale x 2 x float> %va, float %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fsub.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x float> %va, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+define <vscale x 2 x float> @vfrsub_vf_nxv2f32_unmasked(<vscale x 2 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fsub.nxv2f32(<vscale x 2 x float> %vb, <vscale x 2 x float> %va, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+declare <vscale x 4 x float> @llvm.vp.fsub.nxv4f32(<vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x float> @vfrsub_vf_nxv4f32(<vscale x 4 x float> %va, float %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v26, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fsub.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %va, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+define <vscale x 4 x float> @vfrsub_vf_nxv4f32_unmasked(<vscale x 4 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v26, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fsub.nxv4f32(<vscale x 4 x float> %vb, <vscale x 4 x float> %va, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+declare <vscale x 8 x float> @llvm.vp.fsub.nxv8f32(<vscale x 8 x float>, <vscale x 8 x float>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x float> @vfrsub_vf_nxv8f32(<vscale x 8 x float> %va, float %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v28, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fsub.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %va, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+define <vscale x 8 x float> @vfrsub_vf_nxv8f32_unmasked(<vscale x 8 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v28, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fsub.nxv8f32(<vscale x 8 x float> %vb, <vscale x 8 x float> %va, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+declare <vscale x 16 x float> @llvm.vp.fsub.nxv16f32(<vscale x 16 x float>, <vscale x 16 x float>, <vscale x 16 x i1>, i32)
+
+define <vscale x 16 x float> @vfrsub_vf_nxv16f32(<vscale x 16 x float> %va, float %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v16, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fsub.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x float> %va, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+define <vscale x 16 x float> @vfrsub_vf_nxv16f32_unmasked(<vscale x 16 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v16, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> undef, <vscale x 16 x i32> zeroinitializer
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fsub.nxv16f32(<vscale x 16 x float> %vb, <vscale x 16 x float> %va, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+declare <vscale x 1 x double> @llvm.vp.fsub.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x double> @vfrsub_vf_nxv1f64(<vscale x 1 x double> %va, double %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fsub.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x double> %va, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+define <vscale x 1 x double> @vfrsub_vf_nxv1f64_unmasked(<vscale x 1 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv1f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v25, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fsub.nxv1f64(<vscale x 1 x double> %vb, <vscale x 1 x double> %va, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+declare <vscale x 2 x double> @llvm.vp.fsub.nxv2f64(<vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x double> @vfrsub_vf_nxv2f64(<vscale x 2 x double> %va, double %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v26, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fsub.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x double> %va, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+define <vscale x 2 x double> @vfrsub_vf_nxv2f64_unmasked(<vscale x 2 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v26, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fsub.nxv2f64(<vscale x 2 x double> %vb, <vscale x 2 x double> %va, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+declare <vscale x 4 x double> @llvm.vp.fsub.nxv4f64(<vscale x 4 x double>, <vscale x 4 x double>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x double> @vfrsub_vf_nxv4f64(<vscale x 4 x double> %va, double %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v28, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fsub.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x double> %va, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+define <vscale x 4 x double> @vfrsub_vf_nxv4f64_unmasked(<vscale x 4 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v28, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fsub.nxv4f64(<vscale x 4 x double> %vb, <vscale x 4 x double> %va, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+declare <vscale x 8 x double> @llvm.vp.fsub.nxv8f64(<vscale x 8 x double>, <vscale x 8 x double>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x double> @vfrsub_vf_nxv8f64(<vscale x 8 x double> %va, double %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v16, v8, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fsub.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %va, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}
+
+define <vscale x 8 x double> @vfrsub_vf_nxv8f64_unmasked(<vscale x 8 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfrsub_vf_nxv8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v16, v8
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fsub.nxv8f64(<vscale x 8 x double> %vb, <vscale x 8 x double> %va, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}

llvm/test/CodeGen/RISCV/rvv/vfsub-vp.ll

@@ -0,0 +1,815 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
+; RUN: llc -mtriple=riscv32 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=ilp32d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+; RUN: llc -mtriple=riscv64 -mattr=+d,+experimental-zfh,+experimental-v -target-abi=lp64d \
+; RUN:     -verify-machineinstrs < %s | FileCheck %s
+
+declare <vscale x 1 x half> @llvm.vp.fsub.nxv1f16(<vscale x 1 x half>, <vscale x 1 x half>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x half> @vfsub_vv_nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x half> @llvm.vp.fsub.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+define <vscale x 1 x half> @vfsub_vv_nxv1f16_unmasked(<vscale x 1 x half> %va, <vscale x 1 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv1f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fsub.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+define <vscale x 1 x half> @vfsub_vf_nxv1f16(<vscale x 1 x half> %va, half %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv1f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fsub.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+define <vscale x 1 x half> @vfsub_vf_nxv1f16_unmasked(<vscale x 1 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv1f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 1 x half> %elt.head, <vscale x 1 x half> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x half> @llvm.vp.fsub.nxv1f16(<vscale x 1 x half> %va, <vscale x 1 x half> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x half> %v
+}
+
+declare <vscale x 2 x half> @llvm.vp.fsub.nxv2f16(<vscale x 2 x half>, <vscale x 2 x half>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x half> @vfsub_vv_nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x half> @llvm.vp.fsub.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+define <vscale x 2 x half> @vfsub_vv_nxv2f16_unmasked(<vscale x 2 x half> %va, <vscale x 2 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fsub.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+define <vscale x 2 x half> @vfsub_vf_nxv2f16(<vscale x 2 x half> %va, half %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv2f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fsub.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+define <vscale x 2 x half> @vfsub_vf_nxv2f16_unmasked(<vscale x 2 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv2f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 2 x half> %elt.head, <vscale x 2 x half> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x half> @llvm.vp.fsub.nxv2f16(<vscale x 2 x half> %va, <vscale x 2 x half> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x half> %v
+}
+
+declare <vscale x 4 x half> @llvm.vp.fsub.nxv4f16(<vscale x 4 x half>, <vscale x 4 x half>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x half> @vfsub_vv_nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 4 x half> @llvm.vp.fsub.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+define <vscale x 4 x half> @vfsub_vv_nxv4f16_unmasked(<vscale x 4 x half> %va, <vscale x 4 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fsub.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+define <vscale x 4 x half> @vfsub_vf_nxv4f16(<vscale x 4 x half> %va, half %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv4f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fsub.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+define <vscale x 4 x half> @vfsub_vf_nxv4f16_unmasked(<vscale x 4 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv4f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 4 x half> %elt.head, <vscale x 4 x half> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x half> @llvm.vp.fsub.nxv4f16(<vscale x 4 x half> %va, <vscale x 4 x half> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x half> %v
+}
+
+declare <vscale x 8 x half> @llvm.vp.fsub.nxv8f16(<vscale x 8 x half>, <vscale x 8 x half>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x half> @vfsub_vv_nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x half> @llvm.vp.fsub.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+define <vscale x 8 x half> @vfsub_vv_nxv8f16_unmasked(<vscale x 8 x half> %va, <vscale x 8 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fsub.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+define <vscale x 8 x half> @vfsub_vf_nxv8f16(<vscale x 8 x half> %va, half %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv8f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fsub.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+define <vscale x 8 x half> @vfsub_vf_nxv8f16_unmasked(<vscale x 8 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv8f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v26
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 8 x half> %elt.head, <vscale x 8 x half> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x half> @llvm.vp.fsub.nxv8f16(<vscale x 8 x half> %va, <vscale x 8 x half> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x half> %v
+}
+
+declare <vscale x 16 x half> @llvm.vp.fsub.nxv16f16(<vscale x 16 x half>, <vscale x 16 x half>, <vscale x 16 x i1>, i32)
+
+define <vscale x 16 x half> @vfsub_vv_nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 16 x half> @llvm.vp.fsub.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+define <vscale x 16 x half> @vfsub_vv_nxv16f16_unmasked(<vscale x 16 x half> %va, <vscale x 16 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fsub.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+define <vscale x 16 x half> @vfsub_vf_nxv16f16(<vscale x 16 x half> %va, half %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv16f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fsub.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+define <vscale x 16 x half> @vfsub_vf_nxv16f16_unmasked(<vscale x 16 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv16f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v28
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 16 x half> %elt.head, <vscale x 16 x half> undef, <vscale x 16 x i32> zeroinitializer
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x half> @llvm.vp.fsub.nxv16f16(<vscale x 16 x half> %va, <vscale x 16 x half> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x half> %v
+}
+
+declare <vscale x 32 x half> @llvm.vp.fsub.nxv32f16(<vscale x 32 x half>, <vscale x 32 x half>, <vscale x 32 x i1>, i32)
+
+define <vscale x 32 x half> @vfsub_vv_nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 32 x half> @llvm.vp.fsub.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+define <vscale x 32 x half> @vfsub_vv_nxv32f16_unmasked(<vscale x 32 x half> %va, <vscale x 32 x half> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv32f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 32 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 32 x i1> %head, <vscale x 32 x i1> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fsub.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %b, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+define <vscale x 32 x half> @vfsub_vf_nxv32f16(<vscale x 32 x half> %va, half %b, <vscale x 32 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv32f16:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 32 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fsub.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+define <vscale x 32 x half> @vfsub_vf_nxv32f16_unmasked(<vscale x 32 x half> %va, half %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv32f16_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e16, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e16, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 32 x half> undef, half %b, i32 0
+  %vb = shufflevector <vscale x 32 x half> %elt.head, <vscale x 32 x half> undef, <vscale x 32 x i32> zeroinitializer
+  %head = insertelement <vscale x 32 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 32 x i1> %head, <vscale x 32 x i1> undef, <vscale x 32 x i32> zeroinitializer
+  %v = call <vscale x 32 x half> @llvm.vp.fsub.nxv32f16(<vscale x 32 x half> %va, <vscale x 32 x half> %vb, <vscale x 32 x i1> %m, i32 %evl)
+  ret <vscale x 32 x half> %v
+}
+
+declare <vscale x 1 x float> @llvm.vp.fsub.nxv1f32(<vscale x 1 x float>, <vscale x 1 x float>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x float> @vfsub_vv_nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x float> @llvm.vp.fsub.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+define <vscale x 1 x float> @vfsub_vv_nxv1f32_unmasked(<vscale x 1 x float> %va, <vscale x 1 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv1f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fsub.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+define <vscale x 1 x float> @vfsub_vf_nxv1f32(<vscale x 1 x float> %va, float %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv1f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fsub.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+define <vscale x 1 x float> @vfsub_vf_nxv1f32_unmasked(<vscale x 1 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv1f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, mf2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, mf2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 1 x float> %elt.head, <vscale x 1 x float> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x float> @llvm.vp.fsub.nxv1f32(<vscale x 1 x float> %va, <vscale x 1 x float> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x float> %v
+}
+
+declare <vscale x 2 x float> @llvm.vp.fsub.nxv2f32(<vscale x 2 x float>, <vscale x 2 x float>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x float> @vfsub_vv_nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x float> @llvm.vp.fsub.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+define <vscale x 2 x float> @vfsub_vv_nxv2f32_unmasked(<vscale x 2 x float> %va, <vscale x 2 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fsub.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+define <vscale x 2 x float> @vfsub_vf_nxv2f32(<vscale x 2 x float> %va, float %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv2f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fsub.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+define <vscale x 2 x float> @vfsub_vf_nxv2f32_unmasked(<vscale x 2 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv2f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 2 x float> %elt.head, <vscale x 2 x float> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x float> @llvm.vp.fsub.nxv2f32(<vscale x 2 x float> %va, <vscale x 2 x float> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x float> %v
+}
+
+declare <vscale x 4 x float> @llvm.vp.fsub.nxv4f32(<vscale x 4 x float>, <vscale x 4 x float>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x float> @vfsub_vv_nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 4 x float> @llvm.vp.fsub.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+define <vscale x 4 x float> @vfsub_vv_nxv4f32_unmasked(<vscale x 4 x float> %va, <vscale x 4 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fsub.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+define <vscale x 4 x float> @vfsub_vf_nxv4f32(<vscale x 4 x float> %va, float %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv4f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fsub.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+define <vscale x 4 x float> @vfsub_vf_nxv4f32_unmasked(<vscale x 4 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv4f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v26
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 4 x float> %elt.head, <vscale x 4 x float> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x float> @llvm.vp.fsub.nxv4f32(<vscale x 4 x float> %va, <vscale x 4 x float> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x float> %v
+}
+
+declare <vscale x 8 x float> @llvm.vp.fsub.nxv8f32(<vscale x 8 x float>, <vscale x 8 x float>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x float> @vfsub_vv_nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x float> @llvm.vp.fsub.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+define <vscale x 8 x float> @vfsub_vv_nxv8f32_unmasked(<vscale x 8 x float> %va, <vscale x 8 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fsub.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+define <vscale x 8 x float> @vfsub_vf_nxv8f32(<vscale x 8 x float> %va, float %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv8f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fsub.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+define <vscale x 8 x float> @vfsub_vf_nxv8f32_unmasked(<vscale x 8 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv8f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v28
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 8 x float> %elt.head, <vscale x 8 x float> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x float> @llvm.vp.fsub.nxv8f32(<vscale x 8 x float> %va, <vscale x 8 x float> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x float> %v
+}
+
+declare <vscale x 16 x float> @llvm.vp.fsub.nxv16f32(<vscale x 16 x float>, <vscale x 16 x float>, <vscale x 16 x i1>, i32)
+
+define <vscale x 16 x float> @vfsub_vv_nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 16 x float> @llvm.vp.fsub.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+define <vscale x 16 x float> @vfsub_vv_nxv16f32_unmasked(<vscale x 16 x float> %va, <vscale x 16 x float> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fsub.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %b, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+define <vscale x 16 x float> @vfsub_vf_nxv16f32(<vscale x 16 x float> %va, float %b, <vscale x 16 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv16f32:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fsub.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+define <vscale x 16 x float> @vfsub_vf_nxv16f32_unmasked(<vscale x 16 x float> %va, float %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv16f32_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e32, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e32, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 16 x float> undef, float %b, i32 0
+  %vb = shufflevector <vscale x 16 x float> %elt.head, <vscale x 16 x float> undef, <vscale x 16 x i32> zeroinitializer
+  %head = insertelement <vscale x 16 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 16 x i1> %head, <vscale x 16 x i1> undef, <vscale x 16 x i32> zeroinitializer
+  %v = call <vscale x 16 x float> @llvm.vp.fsub.nxv16f32(<vscale x 16 x float> %va, <vscale x 16 x float> %vb, <vscale x 16 x i1> %m, i32 %evl)
+  ret <vscale x 16 x float> %v
+}
+
+declare <vscale x 1 x double> @llvm.vp.fsub.nxv1f64(<vscale x 1 x double>, <vscale x 1 x double>, <vscale x 1 x i1>, i32)
+
+define <vscale x 1 x double> @vfsub_vv_nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 1 x double> @llvm.vp.fsub.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+define <vscale x 1 x double> @vfsub_vv_nxv1f64_unmasked(<vscale x 1 x double> %va, <vscale x 1 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv1f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v9
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fsub.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %b, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+define <vscale x 1 x double> @vfsub_vf_nxv1f64(<vscale x 1 x double> %va, double %b, <vscale x 1 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv1f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fsub.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+define <vscale x 1 x double> @vfsub_vf_nxv1f64_unmasked(<vscale x 1 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv1f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m1, ta, mu
+; CHECK-NEXT:    vfmv.v.f v25, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m1, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v25
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 1 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 1 x double> %elt.head, <vscale x 1 x double> undef, <vscale x 1 x i32> zeroinitializer
+  %head = insertelement <vscale x 1 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 1 x i1> %head, <vscale x 1 x i1> undef, <vscale x 1 x i32> zeroinitializer
+  %v = call <vscale x 1 x double> @llvm.vp.fsub.nxv1f64(<vscale x 1 x double> %va, <vscale x 1 x double> %vb, <vscale x 1 x i1> %m, i32 %evl)
+  ret <vscale x 1 x double> %v
+}
+
+declare <vscale x 2 x double> @llvm.vp.fsub.nxv2f64(<vscale x 2 x double>, <vscale x 2 x double>, <vscale x 2 x i1>, i32)
+
+define <vscale x 2 x double> @vfsub_vv_nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v10, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 2 x double> @llvm.vp.fsub.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+define <vscale x 2 x double> @vfsub_vv_nxv2f64_unmasked(<vscale x 2 x double> %va, <vscale x 2 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v10
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fsub.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %b, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+define <vscale x 2 x double> @vfsub_vf_nxv2f64(<vscale x 2 x double> %va, double %b, <vscale x 2 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv2f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v26, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fsub.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+define <vscale x 2 x double> @vfsub_vf_nxv2f64_unmasked(<vscale x 2 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv2f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m2, ta, mu
+; CHECK-NEXT:    vfmv.v.f v26, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m2, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v26
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 2 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 2 x double> %elt.head, <vscale x 2 x double> undef, <vscale x 2 x i32> zeroinitializer
+  %head = insertelement <vscale x 2 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 2 x i1> %head, <vscale x 2 x i1> undef, <vscale x 2 x i32> zeroinitializer
+  %v = call <vscale x 2 x double> @llvm.vp.fsub.nxv2f64(<vscale x 2 x double> %va, <vscale x 2 x double> %vb, <vscale x 2 x i1> %m, i32 %evl)
+  ret <vscale x 2 x double> %v
+}
+
+declare <vscale x 4 x double> @llvm.vp.fsub.nxv4f64(<vscale x 4 x double>, <vscale x 4 x double>, <vscale x 4 x i1>, i32)
+
+define <vscale x 4 x double> @vfsub_vv_nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v12, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 4 x double> @llvm.vp.fsub.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+define <vscale x 4 x double> @vfsub_vv_nxv4f64_unmasked(<vscale x 4 x double> %va, <vscale x 4 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v12
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fsub.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %b, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+define <vscale x 4 x double> @vfsub_vf_nxv4f64(<vscale x 4 x double> %va, double %b, <vscale x 4 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv4f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v28, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fsub.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+define <vscale x 4 x double> @vfsub_vf_nxv4f64_unmasked(<vscale x 4 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv4f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m4, ta, mu
+; CHECK-NEXT:    vfmv.v.f v28, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m4, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v28
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 4 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 4 x double> %elt.head, <vscale x 4 x double> undef, <vscale x 4 x i32> zeroinitializer
+  %head = insertelement <vscale x 4 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 4 x i1> %head, <vscale x 4 x i1> undef, <vscale x 4 x i32> zeroinitializer
+  %v = call <vscale x 4 x double> @llvm.vp.fsub.nxv4f64(<vscale x 4 x double> %va, <vscale x 4 x double> %vb, <vscale x 4 x i1> %m, i32 %evl)
+  ret <vscale x 4 x double> %v
+}
+
+declare <vscale x 8 x double> @llvm.vp.fsub.nxv8f64(<vscale x 8 x double>, <vscale x 8 x double>, <vscale x 8 x i1>, i32)
+
+define <vscale x 8 x double> @vfsub_vv_nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %v = call <vscale x 8 x double> @llvm.vp.fsub.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}
+
+define <vscale x 8 x double> @vfsub_vv_nxv8f64_unmasked(<vscale x 8 x double> %va, <vscale x 8 x double> %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vv_nxv8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fsub.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %b, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}
+
+define <vscale x 8 x double> @vfsub_vf_nxv8f64(<vscale x 8 x double> %va, double %b, <vscale x 8 x i1> %m, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv8f64:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v16, v0.t
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fsub.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}
+
+define <vscale x 8 x double> @vfsub_vf_nxv8f64_unmasked(<vscale x 8 x double> %va, double %b, i32 zeroext %evl) {
+; CHECK-LABEL: vfsub_vf_nxv8f64_unmasked:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vsetvli a1, zero, e64, m8, ta, mu
+; CHECK-NEXT:    vfmv.v.f v16, fa0
+; CHECK-NEXT:    vsetvli zero, a0, e64, m8, ta, mu
+; CHECK-NEXT:    vfsub.vv v8, v8, v16
+; CHECK-NEXT:    ret
+  %elt.head = insertelement <vscale x 8 x double> undef, double %b, i32 0
+  %vb = shufflevector <vscale x 8 x double> %elt.head, <vscale x 8 x double> undef, <vscale x 8 x i32> zeroinitializer
+  %head = insertelement <vscale x 8 x i1> undef, i1 true, i32 0
+  %m = shufflevector <vscale x 8 x i1> %head, <vscale x 8 x i1> undef, <vscale x 8 x i32> zeroinitializer
+  %v = call <vscale x 8 x double> @llvm.vp.fsub.nxv8f64(<vscale x 8 x double> %va, <vscale x 8 x double> %vb, <vscale x 8 x i1> %m, i32 %evl)
+  ret <vscale x 8 x double> %v
+}