Index: llvm/trunk/lib/Target/X86/X86ISelLowering.cpp =================================================================== --- llvm/trunk/lib/Target/X86/X86ISelLowering.cpp +++ llvm/trunk/lib/Target/X86/X86ISelLowering.cpp @@ -34349,6 +34349,20 @@ return SDValue(); } +/// Detect a pattern of truncation with signed saturation. +/// The types should allow to use VPMOVSS* instruction on AVX512. +/// Return the source value to be truncated or SDValue() if the pattern was not +/// matched. +static SDValue detectAVX512SSatPattern(SDValue In, EVT VT, + const X86Subtarget &Subtarget, + const TargetLowering &TLI) { + if (!TLI.isTypeLegal(In.getValueType())) + return SDValue(); + if (!isSATValidOnAVX512Subtarget(In.getValueType(), VT, Subtarget)) + return SDValue(); + return detectSSatPattern(In, VT); +} + /// Detect a pattern of truncation with saturation: /// (truncate (umin (x, unsigned_max_of_dest_type)) to dest_type). /// The types should allow to use VPMOVUS* instruction on AVX512. @@ -34988,6 +35002,12 @@ const TargetLowering &TLI = DAG.getTargetLoweringInfo(); if (SDValue Val = + detectAVX512SSatPattern(St->getValue(), St->getMemoryVT(), Subtarget, + TLI)) + return EmitTruncSStore(true /* Signed saturation */, St->getChain(), + dl, Val, St->getBasePtr(), + St->getMemoryVT(), St->getMemOperand(), DAG); + if (SDValue Val = detectAVX512USatPattern(St->getValue(), St->getMemoryVT(), Subtarget, TLI)) return EmitTruncSStore(false /* Unsigned saturation */, St->getChain(), Index: llvm/trunk/test/CodeGen/X86/vector-trunc-ssat.ll =================================================================== --- llvm/trunk/test/CodeGen/X86/vector-trunc-ssat.ll +++ llvm/trunk/test/CodeGen/X86/vector-trunc-ssat.ll @@ -2042,9 +2042,7 @@ ; ; AVX512-LABEL: trunc_ssat_v8i64_v8i8_store: ; AVX512: # %bb.0: -; AVX512-NEXT: vpminsq {{.*}}(%rip){1to8}, %zmm0, %zmm0 -; AVX512-NEXT: vpmaxsq {{.*}}(%rip){1to8}, %zmm0, %zmm0 -; AVX512-NEXT: vpmovqb %zmm0, (%rdi) +; AVX512-NEXT: vpmovsqb %zmm0, (%rdi) ; AVX512-NEXT: vzeroupper ; AVX512-NEXT: retq %1 = icmp slt <8 x i64> %a0, @@ -2998,9 +2996,7 @@ ; ; AVX512VL-LABEL: trunc_ssat_v8i32_v8i8_store: ; AVX512VL: # %bb.0: -; AVX512VL-NEXT: vpminsd {{.*}}(%rip){1to8}, %ymm0, %ymm0 -; AVX512VL-NEXT: vpmaxsd {{.*}}(%rip){1to8}, %ymm0, %ymm0 -; AVX512VL-NEXT: vpmovdb %ymm0, (%rdi) +; AVX512VL-NEXT: vpmovsdb %ymm0, (%rdi) ; AVX512VL-NEXT: vzeroupper ; AVX512VL-NEXT: retq ; @@ -3018,9 +3014,7 @@ ; ; AVX512BWVL-LABEL: trunc_ssat_v8i32_v8i8_store: ; AVX512BWVL: # %bb.0: -; AVX512BWVL-NEXT: vpminsd {{.*}}(%rip){1to8}, %ymm0, %ymm0 -; AVX512BWVL-NEXT: vpmaxsd {{.*}}(%rip){1to8}, %ymm0, %ymm0 -; AVX512BWVL-NEXT: vpmovdb %ymm0, (%rdi) +; AVX512BWVL-NEXT: vpmovsdb %ymm0, (%rdi) ; AVX512BWVL-NEXT: vzeroupper ; AVX512BWVL-NEXT: retq %1 = icmp slt <8 x i32> %a0,