llvmbot wrote:
<!--LLVM PR SUMMARY COMMENT--> @llvm/pr-subscribers-backend-x86 Author: Zihao Wang (hhy3) <details> <summary>Changes</summary> #<!-- -->163732 --- Patch is 36.81 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/187656.diff 11 Files Affected: - (modified) clang/include/clang/Basic/BuiltinsX86.td (+15-15) - (modified) clang/lib/AST/ByteCode/InterpBuiltin.cpp (+51) - (modified) clang/lib/AST/ExprConstant.cpp (+40) - (modified) clang/lib/Headers/avx512fintrin.h (+16-24) - (modified) clang/lib/Headers/avx512vbmi2intrin.h (+8-12) - (modified) clang/lib/Headers/avx512vlintrin.h (+32-32) - (modified) clang/lib/Headers/avx512vlvbmi2intrin.h (+16-24) - (modified) clang/test/CodeGen/X86/avx512f-builtins.c (+10-1) - (modified) clang/test/CodeGen/X86/avx512vbmi2-builtins.c (+5) - (modified) clang/test/CodeGen/X86/avx512vl-builtins.c (+20-2) - (modified) clang/test/CodeGen/X86/avx512vlvbmi2-builtins.c (+10) ``````````diff diff --git a/clang/include/clang/Basic/BuiltinsX86.td b/clang/include/clang/Basic/BuiltinsX86.td index 09b4d1c9970fd..4cadd570cc6c4 100644 --- a/clang/include/clang/Basic/BuiltinsX86.td +++ b/clang/include/clang/Basic/BuiltinsX86.td @@ -1365,51 +1365,51 @@ let Features = "avx512f", Attributes = [NoThrow, Const, RequiredVectorWidth<128> def subsd_round_mask : X86Builtin<"_Vector<2, double>(_Vector<2, double>, _Vector<2, double>, _Vector<2, double>, unsigned char, _Constant int)">; } -let Features = "avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in { +let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in { def compressdf128_mask : X86Builtin<"_Vector<2, double>(_Vector<2, double>, _Vector<2, double>, unsigned char)">; } -let Features = "avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in { +let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in { def compressdf256_mask : X86Builtin<"_Vector<4, double>(_Vector<4, double>, _Vector<4, double>, unsigned char)">; } -let Features = "avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in { +let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in { def compressdi128_mask : X86Builtin<"_Vector<2, long long int>(_Vector<2, long long int>, _Vector<2, long long int>, unsigned char)">; } -let Features = "avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in { +let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in { def compressdi256_mask : X86Builtin<"_Vector<4, long long int>(_Vector<4, long long int>, _Vector<4, long long int>, unsigned char)">; } -let Features = "avx512vl,avx512vbmi2", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in { +let Features = "avx512vl,avx512vbmi2", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in { def compresshi128_mask : X86Builtin<"_Vector<8, short>(_Vector<8, short>, _Vector<8, short>, unsigned char)">; } -let Features = "avx512vl,avx512vbmi2", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in { +let Features = "avx512vl,avx512vbmi2", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in { def compresshi256_mask : X86Builtin<"_Vector<16, short>(_Vector<16, short>, _Vector<16, short>, unsigned short)">; } -let Features = "avx512vl,avx512vbmi2", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in { +let Features = "avx512vl,avx512vbmi2", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in { def compressqi128_mask : X86Builtin<"_Vector<16, char>(_Vector<16, char>, _Vector<16, char>, unsigned short)">; } -let Features = "avx512vl,avx512vbmi2", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in { +let Features = "avx512vl,avx512vbmi2", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in { def compressqi256_mask : X86Builtin<"_Vector<32, char>(_Vector<32, char>, _Vector<32, char>, unsigned int)">; } -let Features = "avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in { +let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in { def compresssf128_mask : X86Builtin<"_Vector<4, float>(_Vector<4, float>, _Vector<4, float>, unsigned char)">; } -let Features = "avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in { +let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in { def compresssf256_mask : X86Builtin<"_Vector<8, float>(_Vector<8, float>, _Vector<8, float>, unsigned char)">; } -let Features = "avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<128>] in { +let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<128>] in { def compresssi128_mask : X86Builtin<"_Vector<4, int>(_Vector<4, int>, _Vector<4, int>, unsigned char)">; } -let Features = "avx512vl", Attributes = [NoThrow, Const, RequiredVectorWidth<256>] in { +let Features = "avx512vl", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<256>] in { def compresssi256_mask : X86Builtin<"_Vector<8, int>(_Vector<8, int>, _Vector<8, int>, unsigned char)">; } @@ -3210,17 +3210,17 @@ let Features = "avx512bw", Attributes = [NoThrow, Const, RequiredVectorWidth<512 def psadbw512 : X86Builtin<"_Vector<8, long long int>(_Vector<64, char>, _Vector<64, char>)">; } -let Features = "avx512f", Attributes = [NoThrow, Const, RequiredVectorWidth<512>] in { +let Features = "avx512f", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<512>] in { def compressdf512_mask : X86Builtin<"_Vector<8, double>(_Vector<8, double>, _Vector<8, double>, unsigned char)">; def compressdi512_mask : X86Builtin<"_Vector<8, long long int>(_Vector<8, long long int>, _Vector<8, long long int>, unsigned char)">; } -let Features = "avx512vbmi2", Attributes = [NoThrow, Const, RequiredVectorWidth<512>] in { +let Features = "avx512vbmi2", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<512>] in { def compresshi512_mask : X86Builtin<"_Vector<32, short>(_Vector<32, short>, _Vector<32, short>, unsigned int)">; def compressqi512_mask : X86Builtin<"_Vector<64, char>(_Vector<64, char>, _Vector<64, char>, unsigned long long int)">; } -let Features = "avx512f", Attributes = [NoThrow, Const, RequiredVectorWidth<512>] in { +let Features = "avx512f", Attributes = [NoThrow, Const, Constexpr, RequiredVectorWidth<512>] in { def compresssf512_mask : X86Builtin<"_Vector<16, float>(_Vector<16, float>, _Vector<16, float>, unsigned short)">; def compresssi512_mask : X86Builtin<"_Vector<16, int>(_Vector<16, int>, _Vector<16, int>, unsigned short)">; } diff --git a/clang/lib/AST/ByteCode/InterpBuiltin.cpp b/clang/lib/AST/ByteCode/InterpBuiltin.cpp index abc746af81306..5a9a8e7ae95b7 100644 --- a/clang/lib/AST/ByteCode/InterpBuiltin.cpp +++ b/clang/lib/AST/ByteCode/InterpBuiltin.cpp @@ -3550,6 +3550,38 @@ static bool interp__builtin_ia32_vpconflict(InterpState &S, CodePtr OpPC, return true; } +static bool interp__builtin_ia32_compress(InterpState &S, CodePtr OpPC, + const CallExpr *Call) { + assert(Call->getNumArgs() == 3); + + APSInt Mask = popToAPSInt(S, Call->getArg(2)); + const Pointer &Passthru = S.Stk.pop<Pointer>(); + const Pointer &Source = S.Stk.pop<Pointer>(); + const Pointer &Dst = S.Stk.peek<Pointer>(); + + unsigned NumElems = Source.getNumElems(); + PrimType ElemT = Source.getFieldDesc()->getPrimType(); + + unsigned J = 0; + for (unsigned I = 0; I != NumElems; ++I) { + if (Mask[I]) { + if (ElemT == PT_Float) + Dst.elem<Floating>(J) = Source.elem<Floating>(I); + else + INT_TYPE_SWITCH_NO_BOOL(ElemT, { Dst.elem<T>(J) = Source.elem<T>(I); }); + ++J; + } + } + for (unsigned I = J; I != NumElems; ++I) { + if (ElemT == PT_Float) + Dst.elem<Floating>(I) = Passthru.elem<Floating>(I); + else + INT_TYPE_SWITCH_NO_BOOL(ElemT, { Dst.elem<T>(I) = Passthru.elem<T>(I); }); + } + Dst.initializeAllElements(); + return true; +} + static bool interp__builtin_ia32_cvt_vec2mask(InterpState &S, CodePtr OpPC, const CallExpr *Call, unsigned ID) { @@ -5112,6 +5144,25 @@ bool InterpretBuiltin(InterpState &S, CodePtr OpPC, const CallExpr *Call, case X86::BI__builtin_ia32_vpconflictdi_256: case X86::BI__builtin_ia32_vpconflictdi_512: return interp__builtin_ia32_vpconflict(S, OpPC, Call); + case X86::BI__builtin_ia32_compressdf128_mask: + case X86::BI__builtin_ia32_compressdf256_mask: + case X86::BI__builtin_ia32_compressdf512_mask: + case X86::BI__builtin_ia32_compressdi128_mask: + case X86::BI__builtin_ia32_compressdi256_mask: + case X86::BI__builtin_ia32_compressdi512_mask: + case X86::BI__builtin_ia32_compresshi128_mask: + case X86::BI__builtin_ia32_compresshi256_mask: + case X86::BI__builtin_ia32_compresshi512_mask: + case X86::BI__builtin_ia32_compressqi128_mask: + case X86::BI__builtin_ia32_compressqi256_mask: + case X86::BI__builtin_ia32_compressqi512_mask: + case X86::BI__builtin_ia32_compresssf128_mask: + case X86::BI__builtin_ia32_compresssf256_mask: + case X86::BI__builtin_ia32_compresssf512_mask: + case X86::BI__builtin_ia32_compresssi128_mask: + case X86::BI__builtin_ia32_compresssi256_mask: + case X86::BI__builtin_ia32_compresssi512_mask: + return interp__builtin_ia32_compress(S, OpPC, Call); case clang::X86::BI__builtin_ia32_blendpd: case clang::X86::BI__builtin_ia32_blendpd256: case clang::X86::BI__builtin_ia32_blendps: diff --git a/clang/lib/AST/ExprConstant.cpp b/clang/lib/AST/ExprConstant.cpp index 429fef0a1afa8..b459d4043e13d 100644 --- a/clang/lib/AST/ExprConstant.cpp +++ b/clang/lib/AST/ExprConstant.cpp @@ -12870,6 +12870,46 @@ bool VectorExprEvaluator::VisitCallExpr(const CallExpr *E) { return Success(APValue(ResultElements.data(), ResultElements.size()), E); } + case X86::BI__builtin_ia32_compressdf128_mask: + case X86::BI__builtin_ia32_compressdf256_mask: + case X86::BI__builtin_ia32_compressdf512_mask: + case X86::BI__builtin_ia32_compressdi128_mask: + case X86::BI__builtin_ia32_compressdi256_mask: + case X86::BI__builtin_ia32_compressdi512_mask: + case X86::BI__builtin_ia32_compresshi128_mask: + case X86::BI__builtin_ia32_compresshi256_mask: + case X86::BI__builtin_ia32_compresshi512_mask: + case X86::BI__builtin_ia32_compressqi128_mask: + case X86::BI__builtin_ia32_compressqi256_mask: + case X86::BI__builtin_ia32_compressqi512_mask: + case X86::BI__builtin_ia32_compresssf128_mask: + case X86::BI__builtin_ia32_compresssf256_mask: + case X86::BI__builtin_ia32_compresssf512_mask: + case X86::BI__builtin_ia32_compresssi128_mask: + case X86::BI__builtin_ia32_compresssi256_mask: + case X86::BI__builtin_ia32_compresssi512_mask: { + APValue Source, Passthru; + if (!EvaluateAsRValue(Info, E->getArg(0), Source) || + !EvaluateAsRValue(Info, E->getArg(1), Passthru)) + return false; + APSInt Mask; + if (!EvaluateInteger(E->getArg(2), Mask, Info)) + return false; + + unsigned NumElts = Source.getVectorLength(); + SmallVector<APValue, 64> ResultElements; + ResultElements.reserve(NumElts); + + for (unsigned I = 0; I != NumElts; ++I) { + if (Mask[I]) + ResultElements.push_back(Source.getVectorElt(I)); + } + for (unsigned I = ResultElements.size(); I != NumElts; ++I) { + ResultElements.push_back(Passthru.getVectorElt(I)); + } + + return Success(APValue(ResultElements.data(), ResultElements.size()), E); + } case X86::BI__builtin_ia32_vpconflictsi_128: case X86::BI__builtin_ia32_vpconflictsi_256: case X86::BI__builtin_ia32_vpconflictsi_512: diff --git a/clang/lib/Headers/avx512fintrin.h b/clang/lib/Headers/avx512fintrin.h index 51f8369296e53..a5d2f1726cff4 100644 --- a/clang/lib/Headers/avx512fintrin.h +++ b/clang/lib/Headers/avx512fintrin.h @@ -8096,68 +8096,60 @@ _mm512_stream_ps (void *__P, __m512 __A) __builtin_nontemporal_store((__v16sf_aligned)__A, (__v16sf_aligned*)__P); } -static __inline__ __m512d __DEFAULT_FN_ATTRS512 -_mm512_mask_compress_pd (__m512d __W, __mmask8 __U, __m512d __A) -{ +static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR +_mm512_mask_compress_pd(__m512d __W, __mmask8 __U, __m512d __A) { return (__m512d) __builtin_ia32_compressdf512_mask ((__v8df) __A, (__v8df) __W, (__mmask8) __U); } -static __inline__ __m512d __DEFAULT_FN_ATTRS512 -_mm512_maskz_compress_pd (__mmask8 __U, __m512d __A) -{ +static __inline__ __m512d __DEFAULT_FN_ATTRS512_CONSTEXPR +_mm512_maskz_compress_pd(__mmask8 __U, __m512d __A) { return (__m512d) __builtin_ia32_compressdf512_mask ((__v8df) __A, (__v8df) _mm512_setzero_pd (), (__mmask8) __U); } -static __inline__ __m512i __DEFAULT_FN_ATTRS512 -_mm512_mask_compress_epi64 (__m512i __W, __mmask8 __U, __m512i __A) -{ +static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR +_mm512_mask_compress_epi64(__m512i __W, __mmask8 __U, __m512i __A) { return (__m512i) __builtin_ia32_compressdi512_mask ((__v8di) __A, (__v8di) __W, (__mmask8) __U); } -static __inline__ __m512i __DEFAULT_FN_ATTRS512 -_mm512_maskz_compress_epi64 (__mmask8 __U, __m512i __A) -{ +static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR +_mm512_maskz_compress_epi64(__mmask8 __U, __m512i __A) { return (__m512i) __builtin_ia32_compressdi512_mask ((__v8di) __A, (__v8di) _mm512_setzero_si512 (), (__mmask8) __U); } -static __inline__ __m512 __DEFAULT_FN_ATTRS512 -_mm512_mask_compress_ps (__m512 __W, __mmask16 __U, __m512 __A) -{ +static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR +_mm512_mask_compress_ps(__m512 __W, __mmask16 __U, __m512 __A) { return (__m512) __builtin_ia32_compresssf512_mask ((__v16sf) __A, (__v16sf) __W, (__mmask16) __U); } -static __inline__ __m512 __DEFAULT_FN_ATTRS512 -_mm512_maskz_compress_ps (__mmask16 __U, __m512 __A) -{ +static __inline__ __m512 __DEFAULT_FN_ATTRS512_CONSTEXPR +_mm512_maskz_compress_ps(__mmask16 __U, __m512 __A) { return (__m512) __builtin_ia32_compresssf512_mask ((__v16sf) __A, (__v16sf) _mm512_setzero_ps (), (__mmask16) __U); } -static __inline__ __m512i __DEFAULT_FN_ATTRS512 -_mm512_mask_compress_epi32 (__m512i __W, __mmask16 __U, __m512i __A) -{ +static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR +_mm512_mask_compress_epi32(__m512i __W, __mmask16 __U, __m512i __A) { return (__m512i) __builtin_ia32_compresssi512_mask ((__v16si) __A, (__v16si) __W, (__mmask16) __U); } -static __inline__ __m512i __DEFAULT_FN_ATTRS512 -_mm512_maskz_compress_epi32 (__mmask16 __U, __m512i __A) -{ +static __inline__ __m512i __DEFAULT_FN_ATTRS512_CONSTEXPR +_mm512_maskz_compress_epi32(__mmask16 __U, __m512i __A) { return (__m512i) __builtin_ia32_compresssi512_mask ((__v16si) __A, (__v16si) _mm512_setzero_si512 (), diff --git a/clang/lib/Headers/avx512vbmi2intrin.h b/clang/lib/Headers/avx512vbmi2intrin.h index a24b6e592197c..380d1eeb5c38b 100644 --- a/clang/lib/Headers/avx512vbmi2intrin.h +++ b/clang/lib/Headers/avx512vbmi2intrin.h @@ -25,33 +25,29 @@ #define __DEFAULT_FN_ATTRS_CONSTEXPR __DEFAULT_FN_ATTRS #endif -static __inline__ __m512i __DEFAULT_FN_ATTRS -_mm512_mask_compress_epi16(__m512i __S, __mmask32 __U, __m512i __D) -{ +static __inline__ __m512i __DEFAULT_FN_ATTRS_CONSTEXPR +_mm512_mask_compress_epi16(__m512i __S, __mmask32 __U, __m512i __D) { return (__m512i) __builtin_ia32_compresshi512_mask ((__v32hi) __D, (__v32hi) __S, __U); } -static __inline__ __m512i __DEFAULT_FN_ATTRS -_mm512_maskz_compress_epi16(__mmask32 __U, __m512i __D) -{ +static __inline__ __m512i __DEFAULT_FN_ATTRS_CONSTEXPR +_mm512_maskz_compress_epi16(__mmask32 __U, __m512i __D) { return (__m512i) __builtin_ia32_compresshi512_mask ((__v32hi) __D, (__v32hi) _mm512_setzero_si512(), __U); } -static __inline__ __m512i __DEFAULT_FN_ATTRS -_mm512_mask_compress_epi8(__m512i __S, __mmask64 __U, __m512i __D) -{ +static __inline__ __m512i __DEFAULT_FN_ATTRS_CONSTEXPR +_mm512_mask_compress_epi8(__m512i __S, __mmask64 __U, __m512i __D) { return (__m512i) __builtin_ia32_compressqi512_mask ((__v64qi) __D, (__v64qi) __S, __U); } -static __inline__ __m512i __DEFAULT_FN_ATTRS -_mm512_maskz_compress_epi8(__mmask64 __U, __m512i __D) -{ +static __inline__ __m512i __DEFAULT_FN_ATTRS_CONSTEXPR +_mm512_maskz_compress_epi8(__mmask64 __U, __m512i __D) { return (__m512i) __builtin_ia32_compressqi512_mask ((__v64qi) __D, (__v64qi) _mm512_setzero_si512(), __U); diff --git a/clang/lib/Headers/avx512vlintrin.h b/clang/lib/Headers/avx512vlintrin.h index ea43046240cc0..03d132839d9a9 100644 --- a/clang/lib/Headers/avx512vlintrin.h +++ b/clang/lib/Headers/avx512vlintrin.h @@ -1530,120 +1530,120 @@ _mm256_mask_blend_epi64(__mmask8 __U, __m256i __A, __m256i __W) { (__v4di) __A); } -static __inline__ __m128d __DEFAULT_FN_ATTRS128 -_mm_mask_compress_pd (__m128d __W, __mmask8 __U, __m128d __A) { +static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR +_mm_mask_compress_pd(__m128d __W, __mmask8 __U, __m128d __A) { return (__m128d) __builtin_ia32_compressdf128_mask ((__v2df) __A, (__v2df) __W, (__mmask8) __U); } -static __inline__ __m128d __DEFAULT_FN_ATTRS128 -_mm_maskz_compress_pd (__mmask8 __U, __m128d __A) { +static __inline__ __m128d __DEFAULT_FN_ATTRS128_CONSTEXPR +_mm_maskz_compress_pd(__mmask8 __U, __m128d __A) { return (__m128d) __builtin_ia32_compressdf128_mask ((__v2df) __A, (__v2df) _mm_setzero_pd (), (__mmask8) __U); } -static __inline__ __m256d __DEFAULT_FN_ATTRS256 -_mm256_mask_compress_pd (__m256d __W, __mmask8 __U, __m256d __A) { +static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR +_mm256_mask_compress_pd(__m256d __W, __mmask8 __U, __m256d __A) { return (__m256d) __builtin_ia32_compressdf256_mask ((__v4df) __A, (__v4df) __W, (__mmask8) __U); } -static __inline__ __m256d __DEFAULT_FN_ATTRS256 -_mm256_maskz_compress_pd (__mmask8 __U, __m256d __A) { +static __inline__ __m256d __DEFAULT_FN_ATTRS256_CONSTEXPR +_mm256_maskz_compress_pd(__mmask8 __U, __m256d __A) { return (__m256d) __builtin_ia32_compressdf256_mask ((__v4df) __A, (__v4df) _mm256_setzero_pd (), (__mmask8) __U); } -static __inline__ __m128i __DEFAULT_FN_ATTRS128 -_mm_mask_compress_epi64 (__m128i __W, __mmask8 __U, __m128i __A) { +static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR +_mm_mask_compress_epi64(__m128i __W, __mmask8 __U, __m128i __A) { return (__m128i) __builtin_ia32_compressdi128_mask ((__v2di) __A, (__v2di) __W, (__mmask8) __U); } -static __inline__ __m128i __DEFAULT_FN_ATTRS128 -_mm_maskz_compress_epi64 (__mmask8 __U, __m128i __A) { +static __inline__ __m128i __DEFAULT_FN_ATTRS128_CONSTEXPR +_mm_maskz_compress_epi64(__mmask8 __U, __m128i __A) { return (__m128i) __builtin_ia32_compressdi128_mask ((__v2di) __A, (__v2di) _mm_setzero_si128 (), (__mmask8) __U); } -static __inline__ __m256i __DEFAULT_FN_ATTRS256 -_mm256_mask_compress_epi64 (__m256i __W, __mmask8 __U, __m256i __A) { +static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR +_mm256_mask_compress_epi64(__m256i __W, __mmask8 __U, __m256i __A) { return (__m256i) __builtin_ia32_compressdi256_mask ((__v4di) __A, (__v4di) __W, (__mmask8) __U); } -static __inline__ __m256i __DEFAULT_FN_ATTRS256 -_mm256_maskz_compress_epi64 (__mmask8 __U, __m256i __A) { +static __inline__ __m256i __DEFAULT_FN_ATTRS256_CONSTEXPR +_mm256_maskz_compress_epi64(__mmask8 __U, __m256i __A) { return (__m256i) __builtin_ia32_compressdi256_mask ((__v4di) __A, (__v4di) _mm256_setzero_si256 (), (__mmask8) __U); } -static __inline__ __m128 __DEFAULT_FN_ATTRS128 -_mm_mask_compress_ps (__m128 __W, __mmask8 __U, __m128 __A) { +static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR +_mm_mask_compress_ps(__m128 __W, __mmask8 __U, __m128 __A) { return (__m128) __builtin_ia32_compresssf128_mask ((__v4sf) __A, (__v4sf) __W, (__mmask8) __U); } -static __inline__ __m128 __DEFAULT_FN_ATTRS128 -_mm_maskz_compress_ps (__mmask8 __U, __m128 __A) { +static __inline__ __m128 __DEFAULT_FN_ATTRS128_CONSTEXPR +_mm_maskz_compress_ps(__mmask8 __U, __m128 __A) { return (__m128) __builtin_ia32_compresssf128_mask ((__v4sf) __A, (__v4sf) _mm_setzero_ps (), (__mmask8) __U); } -static __inline__ _... [truncated] `````````` </details> https://github.com/llvm/llvm-project/pull/187656 _______________________________________________ cfe-commits mailing list [email protected] https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
