llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT--> @llvm/pr-subscribers-clang Author: Kerry McLaughlin (kmclaughlin-arm) <details> <summary>Changes</summary> Adds the following SME2 builtins: - svzip (x2 & x4) - svzipq (x2 & x4) - svuzp (x2 & x4) - svuzpq (x2 & x4) See https://github.com/ARM-software/acle/pull/217/files Patch by David Sherwood <david.sherwood@<!-- -->arm.com> --- Patch is 338.19 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/74841.diff 5 Files Affected: - (modified) clang/include/clang/Basic/arm_sve.td (+16) - (added) clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_uzpx2.c (+651) - (added) clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_uzpx4.c (+939) - (added) clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_zipx2.c (+651) - (added) clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_zipx4.c (+938) ``````````diff diff --git a/clang/include/clang/Basic/arm_sve.td b/clang/include/clang/Basic/arm_sve.td index 61788cb5cd9bd..1fe85b0fa1769 100644 --- a/clang/include/clang/Basic/arm_sve.td +++ b/clang/include/clang/Basic/arm_sve.td @@ -2243,3 +2243,19 @@ let TargetGuard = "sme2" in { def SVQCVTN_U16_U64_X4 : SInst<"svqcvtn_u16[_{d}_x4]", "b4.d", "Ul", MergeNone, "aarch64_sve_uqcvtn_x4", [IsStreaming], []>; def SVQCVTN_U16_S64_X4 : SInst<"svqcvtn_u16[_{d}_x4]", "b4.d", "l", MergeNone, "aarch64_sve_sqcvtun_x4", [IsStreaming], []>; } + +// +// Multi-vector zip/unzip +// + +let TargetGuard = "sme2" in { + def SVZIP_X2 : SInst<"svzip[_{d}_x2]", "22", "cUcsUsiUilUlbhfd", MergeNone, "aarch64_sve_zip_x2", [IsStreaming], []>; + def SVZIPQ_X2 : SInst<"svzipq[_{d}_x2]", "22", "cUcsUsiUilUlbhfd", MergeNone, "aarch64_sve_zipq_x2", [IsStreaming], []>; + def SVZIP_X4 : SInst<"svzip[_{d}_x4]", "44", "cUcsUsiUilUlbhfd", MergeNone, "aarch64_sve_zip_x4", [IsStreaming], []>; + def SVZIPQ_X4 : SInst<"svzipq[_{d}_x4]", "44", "cUcsUsiUilUlbhfd", MergeNone, "aarch64_sve_zipq_x4", [IsStreaming], []>; + + def SVUZP_X2 : SInst<"svuzp[_{d}_x2]", "22", "cUcsUsiUilUlbhfd", MergeNone, "aarch64_sve_uzp_x2", [IsStreaming], []>; + def SVUZPQ_X2 : SInst<"svuzpq[_{d}_x2]", "22", "cUcsUsiUilUlbhfd", MergeNone, "aarch64_sve_uzpq_x2", [IsStreaming], []>; + def SVUZP_X4 : SInst<"svuzp[_{d}_x4]", "44", "cUcsUsiUilUlbhfd", MergeNone, "aarch64_sve_uzp_x4", [IsStreaming], []>; + def SVUZPQ_X4 : SInst<"svuzpq[_{d}_x4]", "44", "cUcsUsiUilUlbhfd", MergeNone, "aarch64_sve_uzpq_x4", [IsStreaming], []>; +} diff --git a/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_uzpx2.c b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_uzpx2.c new file mode 100644 index 0000000000000..118b73972575f --- /dev/null +++ b/clang/test/CodeGen/aarch64-sme2-intrinsics/acle_sme2_uzpx2.c @@ -0,0 +1,651 @@ +// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py +// REQUIRES: aarch64-registered-target + +// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s +// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK +// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s +// RUN: %clang_cc1 -DSVE_OVERLOADED_FORMS -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -emit-llvm -o - -x c++ %s | opt -S -p mem2reg,instcombine,tailcallelim | FileCheck %s -check-prefix=CPP-CHECK +// RUN: %clang_cc1 -triple aarch64-none-linux-gnu -target-feature +sve -target-feature +sme2 -S -disable-O0-optnone -Werror -Wall -o /dev/null %s + +#include <arm_sme_draft_spec_subject_to_change.h> + +#ifdef SVE_OVERLOADED_FORMS +// A simple used,unused... macro, long enough to represent any SVE builtin. +#define SVE_ACLE_FUNC(A1,A2_UNUSED) A1 +#else +#define SVE_ACLE_FUNC(A1,A2) A1##A2 +#endif + +// 8-bit UZPs + +// CHECK-LABEL: @test_svuzp_s8_x2( +// CHECK-NEXT: entry: +// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0) +// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16) +// CHECK-NEXT: [[TMP2:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.uzp.x2.nxv16i8(<vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]]) +// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP2]], 0 +// CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP3]], i64 0) +// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP2]], 1 +// CHECK-NEXT: [[TMP6:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 16) +// CHECK-NEXT: ret <vscale x 32 x i8> [[TMP6]] +// +// CPP-CHECK-LABEL: @_Z16test_svuzp_s8_x210svint8x2_t( +// CPP-CHECK-NEXT: entry: +// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0) +// CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16) +// CPP-CHECK-NEXT: [[TMP2:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.uzp.x2.nxv16i8(<vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]]) +// CPP-CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP2]], 0 +// CPP-CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP3]], i64 0) +// CPP-CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP2]], 1 +// CPP-CHECK-NEXT: [[TMP6:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 16) +// CPP-CHECK-NEXT: ret <vscale x 32 x i8> [[TMP6]] +// +svint8x2_t test_svuzp_s8_x2(svint8x2_t zn) __arm_streaming { + return SVE_ACLE_FUNC(svuzp,_s8_x2)(zn); +} + +// CHECK-LABEL: @test_svuzp_u8_x2( +// CHECK-NEXT: entry: +// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0) +// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16) +// CHECK-NEXT: [[TMP2:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.uzp.x2.nxv16i8(<vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]]) +// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP2]], 0 +// CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP3]], i64 0) +// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP2]], 1 +// CHECK-NEXT: [[TMP6:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 16) +// CHECK-NEXT: ret <vscale x 32 x i8> [[TMP6]] +// +// CPP-CHECK-LABEL: @_Z16test_svuzp_u8_x211svuint8x2_t( +// CPP-CHECK-NEXT: entry: +// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN:%.*]], i64 0) +// CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 16 x i8> @llvm.vector.extract.nxv16i8.nxv32i8(<vscale x 32 x i8> [[ZN]], i64 16) +// CPP-CHECK-NEXT: [[TMP2:%.*]] = tail call { <vscale x 16 x i8>, <vscale x 16 x i8> } @llvm.aarch64.sve.uzp.x2.nxv16i8(<vscale x 16 x i8> [[TMP0]], <vscale x 16 x i8> [[TMP1]]) +// CPP-CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP2]], 0 +// CPP-CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> poison, <vscale x 16 x i8> [[TMP3]], i64 0) +// CPP-CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 16 x i8>, <vscale x 16 x i8> } [[TMP2]], 1 +// CPP-CHECK-NEXT: [[TMP6:%.*]] = tail call <vscale x 32 x i8> @llvm.vector.insert.nxv32i8.nxv16i8(<vscale x 32 x i8> [[TMP4]], <vscale x 16 x i8> [[TMP5]], i64 16) +// CPP-CHECK-NEXT: ret <vscale x 32 x i8> [[TMP6]] +// +svuint8x2_t test_svuzp_u8_x2(svuint8x2_t zn) __arm_streaming { + return SVE_ACLE_FUNC(svuzp,_u8_x2)(zn); +} + +// 16-bit UZPs + +// CHECK-LABEL: @test_svuzp_s16_x2( +// CHECK-NEXT: entry: +// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0) +// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8) +// CHECK-NEXT: [[TMP2:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.uzp.x2.nxv8i16(<vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]]) +// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP2]], 0 +// CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP3]], i64 0) +// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP2]], 1 +// CHECK-NEXT: [[TMP6:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 8) +// CHECK-NEXT: ret <vscale x 16 x i16> [[TMP6]] +// +// CPP-CHECK-LABEL: @_Z17test_svuzp_s16_x211svint16x2_t( +// CPP-CHECK-NEXT: entry: +// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0) +// CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8) +// CPP-CHECK-NEXT: [[TMP2:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.uzp.x2.nxv8i16(<vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]]) +// CPP-CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP2]], 0 +// CPP-CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP3]], i64 0) +// CPP-CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP2]], 1 +// CPP-CHECK-NEXT: [[TMP6:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 8) +// CPP-CHECK-NEXT: ret <vscale x 16 x i16> [[TMP6]] +// +svint16x2_t test_svuzp_s16_x2(svint16x2_t zn) __arm_streaming { + return SVE_ACLE_FUNC(svuzp,_s16_x2)(zn); +} + +// CHECK-LABEL: @test_svuzp_u16_x2( +// CHECK-NEXT: entry: +// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0) +// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8) +// CHECK-NEXT: [[TMP2:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.uzp.x2.nxv8i16(<vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]]) +// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP2]], 0 +// CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP3]], i64 0) +// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP2]], 1 +// CHECK-NEXT: [[TMP6:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 8) +// CHECK-NEXT: ret <vscale x 16 x i16> [[TMP6]] +// +// CPP-CHECK-LABEL: @_Z17test_svuzp_u16_x212svuint16x2_t( +// CPP-CHECK-NEXT: entry: +// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN:%.*]], i64 0) +// CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x i16> @llvm.vector.extract.nxv8i16.nxv16i16(<vscale x 16 x i16> [[ZN]], i64 8) +// CPP-CHECK-NEXT: [[TMP2:%.*]] = tail call { <vscale x 8 x i16>, <vscale x 8 x i16> } @llvm.aarch64.sve.uzp.x2.nxv8i16(<vscale x 8 x i16> [[TMP0]], <vscale x 8 x i16> [[TMP1]]) +// CPP-CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP2]], 0 +// CPP-CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> poison, <vscale x 8 x i16> [[TMP3]], i64 0) +// CPP-CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 8 x i16>, <vscale x 8 x i16> } [[TMP2]], 1 +// CPP-CHECK-NEXT: [[TMP6:%.*]] = tail call <vscale x 16 x i16> @llvm.vector.insert.nxv16i16.nxv8i16(<vscale x 16 x i16> [[TMP4]], <vscale x 8 x i16> [[TMP5]], i64 8) +// CPP-CHECK-NEXT: ret <vscale x 16 x i16> [[TMP6]] +// +svuint16x2_t test_svuzp_u16_x2(svuint16x2_t zn) __arm_streaming { + return SVE_ACLE_FUNC(svuzp,_u16_x2)(zn); +} + +// CHECK-LABEL: @test_svuzp_f16_x2( +// CHECK-NEXT: entry: +// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0) +// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8) +// CHECK-NEXT: [[TMP2:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.uzp.x2.nxv8f16(<vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]]) +// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP2]], 0 +// CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> poison, <vscale x 8 x half> [[TMP3]], i64 0) +// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP2]], 1 +// CHECK-NEXT: [[TMP6:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], i64 8) +// CHECK-NEXT: ret <vscale x 16 x half> [[TMP6]] +// +// CPP-CHECK-LABEL: @_Z17test_svuzp_f16_x213svfloat16x2_t( +// CPP-CHECK-NEXT: entry: +// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN:%.*]], i64 0) +// CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x half> @llvm.vector.extract.nxv8f16.nxv16f16(<vscale x 16 x half> [[ZN]], i64 8) +// CPP-CHECK-NEXT: [[TMP2:%.*]] = tail call { <vscale x 8 x half>, <vscale x 8 x half> } @llvm.aarch64.sve.uzp.x2.nxv8f16(<vscale x 8 x half> [[TMP0]], <vscale x 8 x half> [[TMP1]]) +// CPP-CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP2]], 0 +// CPP-CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> poison, <vscale x 8 x half> [[TMP3]], i64 0) +// CPP-CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 8 x half>, <vscale x 8 x half> } [[TMP2]], 1 +// CPP-CHECK-NEXT: [[TMP6:%.*]] = tail call <vscale x 16 x half> @llvm.vector.insert.nxv16f16.nxv8f16(<vscale x 16 x half> [[TMP4]], <vscale x 8 x half> [[TMP5]], i64 8) +// CPP-CHECK-NEXT: ret <vscale x 16 x half> [[TMP6]] +// +svfloat16x2_t test_svuzp_f16_x2(svfloat16x2_t zn) __arm_streaming { + return SVE_ACLE_FUNC(svuzp,_f16_x2)(zn); +} + +// CHECK-LABEL: @test_svuzp_bf16_x2( +// CHECK-NEXT: entry: +// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0) +// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8) +// CHECK-NEXT: [[TMP2:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.uzp.x2.nxv8bf16(<vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]]) +// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP2]], 0 +// CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> poison, <vscale x 8 x bfloat> [[TMP3]], i64 0) +// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP2]], 1 +// CHECK-NEXT: [[TMP6:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], i64 8) +// CHECK-NEXT: ret <vscale x 16 x bfloat> [[TMP6]] +// +// CPP-CHECK-LABEL: @_Z18test_svuzp_bf16_x214svbfloat16x2_t( +// CPP-CHECK-NEXT: entry: +// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN:%.*]], i64 0) +// CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 8 x bfloat> @llvm.vector.extract.nxv8bf16.nxv16bf16(<vscale x 16 x bfloat> [[ZN]], i64 8) +// CPP-CHECK-NEXT: [[TMP2:%.*]] = tail call { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } @llvm.aarch64.sve.uzp.x2.nxv8bf16(<vscale x 8 x bfloat> [[TMP0]], <vscale x 8 x bfloat> [[TMP1]]) +// CPP-CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP2]], 0 +// CPP-CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> poison, <vscale x 8 x bfloat> [[TMP3]], i64 0) +// CPP-CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 8 x bfloat>, <vscale x 8 x bfloat> } [[TMP2]], 1 +// CPP-CHECK-NEXT: [[TMP6:%.*]] = tail call <vscale x 16 x bfloat> @llvm.vector.insert.nxv16bf16.nxv8bf16(<vscale x 16 x bfloat> [[TMP4]], <vscale x 8 x bfloat> [[TMP5]], i64 8) +// CPP-CHECK-NEXT: ret <vscale x 16 x bfloat> [[TMP6]] +// +svbfloat16x2_t test_svuzp_bf16_x2(svbfloat16x2_t zn) __arm_streaming { + return SVE_ACLE_FUNC(svuzp,_bf16_x2)(zn); +} + +// 32-bit UZPs + +// CHECK-LABEL: @test_svuzp_s32_x2( +// CHECK-NEXT: entry: +// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[ZN:%.*]], i64 0) +// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[ZN]], i64 4) +// CHECK-NEXT: [[TMP2:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.uzp.x2.nxv4i32(<vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]]) +// CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP2]], 0 +// CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP3]], i64 0) +// CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP2]], 1 +// CHECK-NEXT: [[TMP6:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 4) +// CHECK-NEXT: ret <vscale x 8 x i32> [[TMP6]] +// +// CPP-CHECK-LABEL: @_Z17test_svuzp_s32_x211svint32x2_t( +// CPP-CHECK-NEXT: entry: +// CPP-CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[ZN:%.*]], i64 0) +// CPP-CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[ZN]], i64 4) +// CPP-CHECK-NEXT: [[TMP2:%.*]] = tail call { <vscale x 4 x i32>, <vscale x 4 x i32> } @llvm.aarch64.sve.uzp.x2.nxv4i32(<vscale x 4 x i32> [[TMP0]], <vscale x 4 x i32> [[TMP1]]) +// CPP-CHECK-NEXT: [[TMP3:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP2]], 0 +// CPP-CHECK-NEXT: [[TMP4:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> poison, <vscale x 4 x i32> [[TMP3]], i64 0) +// CPP-CHECK-NEXT: [[TMP5:%.*]] = extractvalue { <vscale x 4 x i32>, <vscale x 4 x i32> } [[TMP2]], 1 +// CPP-CHECK-NEXT: [[TMP6:%.*]] = tail call <vscale x 8 x i32> @llvm.vector.insert.nxv8i32.nxv4i32(<vscale x 8 x i32> [[TMP4]], <vscale x 4 x i32> [[TMP5]], i64 4) +// CPP-CHECK-NEXT: ret <vscale x 8 x i32> [[TMP6]] +// +svint32x2_t test_svuzp_s32_x2(svint32x2_t zn) __arm_streaming { + return SVE_ACLE_FUNC(svuzp,_s32_x2)(zn); +} + +// CHECK-LABEL: @test_svuzp_u32_x2( +// CHECK-NEXT: entry: +// CHECK-NEXT: [[TMP0:%.*]] = tail call <vscale x 4 x i32> @llvm.vector.extract.nxv4i32.nxv8i32(<vscale x 8 x i32> [[ZN:%.*]], i64 0) +// CHECK-NEXT: [[TMP1:%.*]] = tail call <vscale... [truncated] `````````` </details> https://github.com/llvm/llvm-project/pull/74841 _______________________________________________ cfe-commits mailing list cfe-commits@lists.llvm.org https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
