maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/clang/test/CodeGen/aarch64-sve-intrinsics/acle_sve_cmpne.c

Ignoring revisions in .git-blame-ignore-revs. Click here to bypass and see the normal blame view.

580 lines
41 KiB
C
Raw Normal View History

[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py
[SVE] Add more warnings checks to clang and LLVM SVE tests There are now more SVE tests in LLVM and Clang that do not emit warnings related to invalid use of EVT::getVectorNumElements() and VectorType::getNumElements(). For these tests I have added additional checks that there are no warnings in order to prevent any future regressions. Differential Revision: https://reviews.llvm.org/D82943
2020-06-19 16:06:21 +08:00
// REQUIRES: aarch64-registered-target
[AArch64][SVE] Enable __ARM_FEATURE_SVE macros. This patch enables the following macros when their corresponding target attributes are set: __ARM_FEATURE_SVE (+sve) __ARM_FEATURE_SVE2 (+sve2) __ARM_FEATURE_SVE2_AES (+sve2-aes) __ARM_FEATURE_SVE2_BITPERM (+sve2-bitperm) __ARM_FEATURE_SVE2_SHA3 (+sve2-sha3) __ARM_FEATURE_SVE2_SM4 (+sve2-sm4) This implies that the base SVE and SVE2 ACLE (00bet2) are now feature complete, meaning that all intrinsics are implemented in LLVM and Clang. Disclaimer: To implement the ACLE we have had to fix up many parts of LLVM to make it support scalable vectors. We have also used many target-specific intrinsics to reduce reliance on parts of LLVM where we know scalable vectors may not yet be handled properly (e.g. some transformation might drop the 'scalable' flag on a vector type). While we've done a best effort with the limited testing that is available to us, we're still working to improve the stability of the implementation. Additionally, Clang may print warnings that code may have miscompiled. We find this often to be a false alarm where the wrong interfaces have been used in LLVM and where resulting code is not actually incorrect. However, this warrants a bug report and investigation. If you find any bugs or issues, please raise them on bugs.llvm.org and let us know! Reviewers: rengolin, efriedma, david-arm, SjoerdMeijer Reviewed By: SjoerdMeijer Differential Revision: https://reviews.llvm.org/D81725
2020-06-24 23:52:41 +08:00
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
[AArch64][SVE] Enable __ARM_FEATURE_SVE macros. This patch enables the following macros when their corresponding target attributes are set: __ARM_FEATURE_SVE (+sve) __ARM_FEATURE_SVE2 (+sve2) __ARM_FEATURE_SVE2_AES (+sve2-aes) __ARM_FEATURE_SVE2_BITPERM (+sve2-bitperm) __ARM_FEATURE_SVE2_SHA3 (+sve2-sha3) __ARM_FEATURE_SVE2_SM4 (+sve2-sm4) This implies that the base SVE and SVE2 ACLE (00bet2) are now feature complete, meaning that all intrinsics are implemented in LLVM and Clang. Disclaimer: To implement the ACLE we have had to fix up many parts of LLVM to make it support scalable vectors. We have also used many target-specific intrinsics to reduce reliance on parts of LLVM where we know scalable vectors may not yet be handled properly (e.g. some transformation might drop the 'scalable' flag on a vector type). While we've done a best effort with the limited testing that is available to us, we're still working to improve the stability of the implementation. Additionally, Clang may print warnings that code may have miscompiled. We find this often to be a false alarm where the wrong interfaces have been used in LLVM and where resulting code is not actually incorrect. However, this warrants a bug report and investigation. If you find any bugs or issues, please raise them on bugs.llvm.org and let us know! Reviewers: rengolin, efriedma, david-arm, SjoerdMeijer Reviewed By: SjoerdMeijer Differential Revision: https://reviews.llvm.org/D81725
2020-06-24 23:52:41 +08:00
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - %s | FileCheck %s
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -Wall -emit-llvm -o - -x c++ %s | FileCheck %s -check-prefix=CPP-CHECK
// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -fallow-half-arguments-and-returns -S -O1 -Werror -o /dev/null %s
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
#include <arm_sve.h>
#ifdef SVE_OVERLOADED_FORMS
// A simple used,unused... macro, long enough to represent any SVE builtin.
#define SVE_ACLE_FUNC(A1,A2_UNUSED,A3,A4_UNUSED) A1##A3
#else
#define SVE_ACLE_FUNC(A1,A2,A3,A4) A1##A2##A3##A4
#endif
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_s8(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z15test_svcmpne_s8u10__SVBool_tu10__SVInt8_tu10__SVInt8_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP0]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_s8(svbool_t pg, svint8_t op1, svint8_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_s8,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]])
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP1]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z16test_svcmpne_s16u10__SVBool_tu11__SVInt16_tu11__SVInt16_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_s16(svbool_t pg, svint16_t op1, svint16_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_s16,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]])
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP1]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z16test_svcmpne_s32u10__SVBool_tu11__SVInt32_tu11__SVInt32_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_s32(svbool_t pg, svint32_t op1, svint32_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_s32,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_s64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]])
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP1]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z16test_svcmpne_s64u10__SVBool_tu11__SVInt64_tu11__SVInt64_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_s64(svbool_t pg, svint64_t op1, svint64_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_s64,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_u8(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z15test_svcmpne_u8u10__SVBool_tu11__SVUint8_tu11__SVUint8_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[OP2:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP0]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_u8(svbool_t pg, svuint8_t op1, svuint8_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_u8,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]])
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP1]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z16test_svcmpne_u16u10__SVBool_tu12__SVUint16_tu12__SVUint16_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[OP2:%.*]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_u16(svbool_t pg, svuint16_t op1, svuint16_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_u16,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]])
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP1]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z16test_svcmpne_u32u10__SVBool_tu12__SVUint32_tu12__SVUint32_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[OP2:%.*]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_u32(svbool_t pg, svuint32_t op1, svuint32_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_u32,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_u64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]])
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP1]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z16test_svcmpne_u64u10__SVBool_tu12__SVUint64_tu12__SVUint64_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_u64(svbool_t pg, svuint64_t op1, svuint64_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_u64,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_n_s64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_s64u10__SVBool_tu11__SVInt64_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_n_s64(svbool_t pg, svint64_t op1, int64_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_n_s64,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_n_u64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_u64u10__SVBool_tu12__SVUint64_tm(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.cmpne.nxv2i64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x i64> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_n_u64(svbool_t pg, svuint64_t op1, uint64_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_n_u64,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_wide_s8(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP0]]
//
// CPP-CHECK-LABEL: @_Z20test_svcmpne_wide_s8u10__SVBool_tu10__SVInt8_tu11__SVInt64_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP0]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_wide_s8(svbool_t pg, svint8_t op1, svint64_t op2)
{
return SVE_ACLE_FUNC(svcmpne_wide,_s8,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_wide_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]])
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP1]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z21test_svcmpne_wide_s16u10__SVBool_tu11__SVInt16_tu11__SVInt64_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_wide_s16(svbool_t pg, svint16_t op1, svint64_t op2)
{
return SVE_ACLE_FUNC(svcmpne_wide,_s16,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_wide_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]])
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP1]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z21test_svcmpne_wide_s32u10__SVBool_tu11__SVInt32_tu11__SVInt64_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[OP2:%.*]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_wide_s32(svbool_t pg, svint32_t op1, svint64_t op2)
{
return SVE_ACLE_FUNC(svcmpne_wide,_s32,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_n_s8(
// CHECK-NEXT: entry:
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpne_n_s8u10__SVBool_tu10__SVInt8_ta(
// CPP-CHECK-NEXT: entry:
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_n_s8(svbool_t pg, svint8_t op1, int8_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_n_s8,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_s16u10__SVBool_tu11__SVInt16_ts(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_n_s16(svbool_t pg, svint16_t op1, int16_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_n_s16,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_s32u10__SVBool_tu11__SVInt32_ti(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_n_s32(svbool_t pg, svint32_t op1, int32_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_n_s32,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_n_u8(
// CHECK-NEXT: entry:
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z17test_svcmpne_n_u8u10__SVBool_tu11__SVUint8_th(
// CPP-CHECK-NEXT: entry:
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 16 x i8> poison, i8 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 16 x i8> [[DOTSPLATINSERT]], <vscale x 16 x i8> poison, <vscale x 16 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 16 x i8> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_n_u8(svbool_t pg, svuint8_t op1, uint8_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_n_u8,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_n_u16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_u16u10__SVBool_tu12__SVUint16_tt(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x i16> poison, i16 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x i16> [[DOTSPLATINSERT]], <vscale x 8 x i16> poison, <vscale x 8 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 8 x i16> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_n_u16(svbool_t pg, svuint16_t op1, uint16_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_n_u16,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_n_u32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_u32u10__SVBool_tu12__SVUint32_tj(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x i32> poison, i32 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x i32> [[DOTSPLATINSERT]], <vscale x 4 x i32> poison, <vscale x 4 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 4 x i32> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_n_u32(svbool_t pg, svuint32_t op1, uint32_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_n_u32,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpne.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]])
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP1]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z16test_svcmpne_f16u10__SVBool_tu13__SVFloat16_tu13__SVFloat16_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpne.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[OP2:%.*]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_f16(svbool_t pg, svfloat16_t op1, svfloat16_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_f16,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpne.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]])
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP1]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z16test_svcmpne_f32u10__SVBool_tu13__SVFloat32_tu13__SVFloat32_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpne.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[OP2:%.*]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_f32(svbool_t pg, svfloat32_t op1, svfloat32_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_f32,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpne.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]])
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP1]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
// CPP-CHECK-LABEL: @_Z16test_svcmpne_f64u10__SVBool_tu13__SVFloat64_tu13__SVFloat64_t(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpne.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[OP2:%.*]])
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP1]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP2]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_f64(svbool_t pg, svfloat64_t op1, svfloat64_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_f64,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_n_f16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpne.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_f16u10__SVBool_tu13__SVFloat16_tDh(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 8 x half> poison, half [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 8 x half> [[DOTSPLATINSERT]], <vscale x 8 x half> poison, <vscale x 8 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.fcmpne.nxv8f16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x half> [[OP1:%.*]], <vscale x 8 x half> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_n_f16(svbool_t pg, svfloat16_t op1, float16_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_n_f16,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_n_f32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpne.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_f32u10__SVBool_tu13__SVFloat32_tf(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 4 x float> poison, float [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 4 x float> [[DOTSPLATINSERT]], <vscale x 4 x float> poison, <vscale x 4 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.fcmpne.nxv4f32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x float> [[OP1:%.*]], <vscale x 4 x float> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_n_f32(svbool_t pg, svfloat32_t op1, float32_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_n_f32,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_n_f64(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpne.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z18test_svcmpne_n_f64u10__SVBool_tu13__SVFloat64_td(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv2i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x double> poison, double [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x double> [[DOTSPLATINSERT]], <vscale x 2 x double> poison, <vscale x 2 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 2 x i1> @llvm.aarch64.sve.fcmpne.nxv2f64(<vscale x 2 x i1> [[TMP0]], <vscale x 2 x double> [[OP1:%.*]], <vscale x 2 x double> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv2i1(<vscale x 2 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_n_f64(svbool_t pg, svfloat64_t op1, float64_t op2)
{
return SVE_ACLE_FUNC(svcmpne,_n_f64,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_wide_n_s8(
// CHECK-NEXT: entry:
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
// CPP-CHECK-LABEL: @_Z22test_svcmpne_wide_n_s8u10__SVBool_tu10__SVInt8_tl(
// CPP-CHECK-NEXT: entry:
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CPP-CHECK-NEXT: [[TMP0:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CPP-CHECK-NEXT: [[TMP1:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.cmpne.wide.nxv16i8(<vscale x 16 x i1> [[PG:%.*]], <vscale x 16 x i8> [[OP1:%.*]], <vscale x 2 x i64> [[TMP0]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP1]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_wide_n_s8(svbool_t pg, svint8_t op1, int64_t op2)
{
return SVE_ACLE_FUNC(svcmpne_wide,_n_s8,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_wide_n_s16(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z23test_svcmpne_wide_n_s16u10__SVBool_tu11__SVInt16_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv8i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 8 x i1> @llvm.aarch64.sve.cmpne.wide.nxv8i16(<vscale x 8 x i1> [[TMP0]], <vscale x 8 x i16> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv8i1(<vscale x 8 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_wide_n_s16(svbool_t pg, svint16_t op1, int64_t op2)
{
return SVE_ACLE_FUNC(svcmpne_wide,_n_s16,,)(pg, op1, op2);
}
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-LABEL: @test_svcmpne_wide_n_s32(
// CHECK-NEXT: entry:
// CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
// CPP-CHECK-LABEL: @_Z23test_svcmpne_wide_n_s32u10__SVBool_tu11__SVInt32_tl(
// CPP-CHECK-NEXT: entry:
// CPP-CHECK-NEXT: [[TMP0:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.convert.from.svbool.nxv4i1(<vscale x 16 x i1> [[PG:%.*]])
[instcombine] Canonicalize constant index type to i64 for extractelement/insertelement The basic idea to this is that a) having a single canonical type makes CSE easier, and b) many of our transforms are inconsistent about which types we end up with based on visit order. I'm restricting this to constants as for non-constants, we'd have to decide whether the simplicity was worth extra instructions. For constants, there are no extra instructions. We chose the canonical type as i64 arbitrarily. We might consider changing this to something else in the future if we have cause. Differential Revision: https://reviews.llvm.org/D115387
2021-12-14 08:29:19 +08:00
// CPP-CHECK-NEXT: [[DOTSPLATINSERT:%.*]] = insertelement <vscale x 2 x i64> poison, i64 [[OP2:%.*]], i64 0
Revert "Revert "[AArch64][SVE][InstCombine] Canonicalize aarch64_sve_dup_x intrinsic to IR splat operation"" This reverts commit eee7d225ded98f42d37c05ec292bbb18560ce06b. Effectively relanding 98c37247d81dfc967ecc49eee7a15612b6510f67 after fixing the failing tests. Change-Id: I5d7461aeb820a2d5f1895457d824a8de4d316ee5
2021-09-11 08:57:29 +08:00
// CPP-CHECK-NEXT: [[TMP1:%.*]] = shufflevector <vscale x 2 x i64> [[DOTSPLATINSERT]], <vscale x 2 x i64> poison, <vscale x 2 x i32> zeroinitializer
[NFC] Recommit "Regenerate SVE ACLE intrinsics tests"" 2nd try This reverts 61ddc3d3db9b6c6b163774c7ce9f4d611cbd7c80 to reapply 91eda9c30f33da6ec6da70b59a5f5da6c6397039 after fixing the " |& " causing failures on windows. Change-Id: Ib646c803b2274f0f24f9a8932de7aa97003529c5
2021-09-08 08:09:42 +08:00
// CPP-CHECK-NEXT: [[TMP2:%.*]] = call <vscale x 4 x i1> @llvm.aarch64.sve.cmpne.wide.nxv4i32(<vscale x 4 x i1> [[TMP0]], <vscale x 4 x i32> [[OP1:%.*]], <vscale x 2 x i64> [[TMP1]])
// CPP-CHECK-NEXT: [[TMP3:%.*]] = call <vscale x 16 x i1> @llvm.aarch64.sve.convert.to.svbool.nxv4i1(<vscale x 4 x i1> [[TMP2]])
// CPP-CHECK-NEXT: ret <vscale x 16 x i1> [[TMP3]]
//
[SveEmitter] Add builtins for compares and ReverseCompare flag. The IsReverseCompare flag tells CGBuiltin to swap the operands, so that a LT/LE intrinsics can be expressed in terms of GE/GT intrinsics. This patch also adds builtins for the wide-variants of the compares. Reviewers: SjoerdMeijer, efriedma, ctetreau Reviewed By: efriedma Tags: #clang Differential Revision: https://reviews.llvm.org/D78747
2020-04-24 18:31:34 +08:00
svbool_t test_svcmpne_wide_n_s32(svbool_t pg, svint32_t op1, int64_t op2)
{
return SVE_ACLE_FUNC(svcmpne_wide,_n_s32,,)(pg, op1, op2);
}