From: Craig Topper Date: Thu, 23 Jun 2016 07:37:33 +0000 (+0000) Subject: [AVX512] Remove masked unpack intrinsics and autoupgrade to vectorshuffle and selects. X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=12d48c9c949f852da7f72046decdb2f3952be59b;p=llvm [AVX512] Remove masked unpack intrinsics and autoupgrade to vectorshuffle and selects. git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@273543 91177308-0d34-0410-b5e6-96231b3b80d8 --- diff --git a/include/llvm/IR/IntrinsicsX86.td b/include/llvm/IR/IntrinsicsX86.td index 084f8dad993..d507795a184 100644 --- a/include/llvm/IR/IntrinsicsX86.td +++ b/include/llvm/IR/IntrinsicsX86.td @@ -4650,225 +4650,6 @@ let TargetPrefix = "x86" in { // All intrinsics start with "llvm.x86.". llvm_v32i16_ty, llvm_i32_ty], [IntrNoMem]>; } -// Unpack ops. -let TargetPrefix = "x86" in { // All intrinsics start with "llvm.x86.". - def int_x86_avx512_mask_unpckh_pd_128 : - GCCBuiltin<"__builtin_ia32_unpckhpd128_mask">, - Intrinsic<[llvm_v2f64_ty], - [llvm_v2f64_ty, llvm_v2f64_ty, llvm_v2f64_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_unpckh_pd_256 : - GCCBuiltin<"__builtin_ia32_unpckhpd256_mask">, - Intrinsic<[llvm_v4f64_ty], - [llvm_v4f64_ty, llvm_v4f64_ty, llvm_v4f64_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_unpckh_pd_512 : - GCCBuiltin<"__builtin_ia32_unpckhpd512_mask">, - Intrinsic<[llvm_v8f64_ty], - [llvm_v8f64_ty, llvm_v8f64_ty, llvm_v8f64_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_unpckh_ps_128 : - GCCBuiltin<"__builtin_ia32_unpckhps128_mask">, - Intrinsic<[llvm_v4f32_ty], - [llvm_v4f32_ty, llvm_v4f32_ty, llvm_v4f32_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_unpckh_ps_256 : - GCCBuiltin<"__builtin_ia32_unpckhps256_mask">, - Intrinsic<[llvm_v8f32_ty], - [llvm_v8f32_ty, llvm_v8f32_ty, llvm_v8f32_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_unpckh_ps_512 : - GCCBuiltin<"__builtin_ia32_unpckhps512_mask">, - Intrinsic<[llvm_v16f32_ty], - [llvm_v16f32_ty, llvm_v16f32_ty, llvm_v16f32_ty, llvm_i16_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_unpckl_pd_128 : - GCCBuiltin<"__builtin_ia32_unpcklpd128_mask">, - Intrinsic<[llvm_v2f64_ty], - [llvm_v2f64_ty, llvm_v2f64_ty, llvm_v2f64_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_unpckl_pd_256 : - GCCBuiltin<"__builtin_ia32_unpcklpd256_mask">, - Intrinsic<[llvm_v4f64_ty], - [llvm_v4f64_ty, llvm_v4f64_ty, llvm_v4f64_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_unpckl_pd_512 : - GCCBuiltin<"__builtin_ia32_unpcklpd512_mask">, - Intrinsic<[llvm_v8f64_ty], - [llvm_v8f64_ty, llvm_v8f64_ty, llvm_v8f64_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_unpckl_ps_128 : - GCCBuiltin<"__builtin_ia32_unpcklps128_mask">, - Intrinsic<[llvm_v4f32_ty], - [llvm_v4f32_ty, llvm_v4f32_ty, llvm_v4f32_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_unpckl_ps_256 : - GCCBuiltin<"__builtin_ia32_unpcklps256_mask">, - Intrinsic<[llvm_v8f32_ty], - [llvm_v8f32_ty, llvm_v8f32_ty, llvm_v8f32_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_unpckl_ps_512 : - GCCBuiltin<"__builtin_ia32_unpcklps512_mask">, - Intrinsic<[llvm_v16f32_ty], - [llvm_v16f32_ty, llvm_v16f32_ty, llvm_v16f32_ty, llvm_i16_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpckhb_w_128 : - GCCBuiltin<"__builtin_ia32_punpckhbw128_mask">, - Intrinsic<[llvm_v16i8_ty], - [llvm_v16i8_ty, llvm_v16i8_ty, llvm_v16i8_ty, llvm_i16_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpckhb_w_256 : - GCCBuiltin<"__builtin_ia32_punpckhbw256_mask">, - Intrinsic<[llvm_v32i8_ty], - [llvm_v32i8_ty, llvm_v32i8_ty, llvm_v32i8_ty, llvm_i32_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpckhb_w_512 : - GCCBuiltin<"__builtin_ia32_punpckhbw512_mask">, - Intrinsic<[llvm_v64i8_ty], - [llvm_v64i8_ty, llvm_v64i8_ty, llvm_v64i8_ty, llvm_i64_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpckhd_q_128 : - GCCBuiltin<"__builtin_ia32_punpckhdq128_mask">, - Intrinsic<[llvm_v4i32_ty], - [llvm_v4i32_ty, llvm_v4i32_ty, llvm_v4i32_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpckhd_q_256 : - GCCBuiltin<"__builtin_ia32_punpckhdq256_mask">, - Intrinsic<[llvm_v8i32_ty], - [llvm_v8i32_ty, llvm_v8i32_ty, llvm_v8i32_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpckhd_q_512 : - GCCBuiltin<"__builtin_ia32_punpckhdq512_mask">, - Intrinsic<[llvm_v16i32_ty], - [llvm_v16i32_ty, llvm_v16i32_ty, llvm_v16i32_ty, llvm_i16_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpckhqd_q_128 : - GCCBuiltin<"__builtin_ia32_punpckhqdq128_mask">, - Intrinsic<[llvm_v2i64_ty], - [llvm_v2i64_ty, llvm_v2i64_ty, llvm_v2i64_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpckhqd_q_256 : - GCCBuiltin<"__builtin_ia32_punpckhqdq256_mask">, - Intrinsic<[llvm_v4i64_ty], - [llvm_v4i64_ty, llvm_v4i64_ty, llvm_v4i64_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpckhqd_q_512 : - GCCBuiltin<"__builtin_ia32_punpckhqdq512_mask">, - Intrinsic<[llvm_v8i64_ty], - [llvm_v8i64_ty, llvm_v8i64_ty, llvm_v8i64_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpckhw_d_128 : - GCCBuiltin<"__builtin_ia32_punpckhwd128_mask">, - Intrinsic<[llvm_v8i16_ty], - [llvm_v8i16_ty, llvm_v8i16_ty, llvm_v8i16_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpckhw_d_256 : - GCCBuiltin<"__builtin_ia32_punpckhwd256_mask">, - Intrinsic<[llvm_v16i16_ty], - [llvm_v16i16_ty, llvm_v16i16_ty, llvm_v16i16_ty, llvm_i16_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpckhw_d_512 : - GCCBuiltin<"__builtin_ia32_punpckhwd512_mask">, - Intrinsic<[llvm_v32i16_ty], - [llvm_v32i16_ty, llvm_v32i16_ty, llvm_v32i16_ty, llvm_i32_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpcklb_w_128 : - GCCBuiltin<"__builtin_ia32_punpcklbw128_mask">, - Intrinsic<[llvm_v16i8_ty], - [llvm_v16i8_ty, llvm_v16i8_ty, llvm_v16i8_ty, llvm_i16_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpcklb_w_256 : - GCCBuiltin<"__builtin_ia32_punpcklbw256_mask">, - Intrinsic<[llvm_v32i8_ty], - [llvm_v32i8_ty, llvm_v32i8_ty, llvm_v32i8_ty, llvm_i32_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpcklb_w_512 : - GCCBuiltin<"__builtin_ia32_punpcklbw512_mask">, - Intrinsic<[llvm_v64i8_ty], - [llvm_v64i8_ty, llvm_v64i8_ty, llvm_v64i8_ty, llvm_i64_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpckld_q_128 : - GCCBuiltin<"__builtin_ia32_punpckldq128_mask">, - Intrinsic<[llvm_v4i32_ty], - [llvm_v4i32_ty, llvm_v4i32_ty, llvm_v4i32_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpckld_q_256 : - GCCBuiltin<"__builtin_ia32_punpckldq256_mask">, - Intrinsic<[llvm_v8i32_ty], - [llvm_v8i32_ty, llvm_v8i32_ty, llvm_v8i32_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpckld_q_512 : - GCCBuiltin<"__builtin_ia32_punpckldq512_mask">, - Intrinsic<[llvm_v16i32_ty], - [llvm_v16i32_ty, llvm_v16i32_ty, llvm_v16i32_ty, llvm_i16_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpcklqd_q_128 : - GCCBuiltin<"__builtin_ia32_punpcklqdq128_mask">, - Intrinsic<[llvm_v2i64_ty], - [llvm_v2i64_ty, llvm_v2i64_ty, llvm_v2i64_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpcklqd_q_256 : - GCCBuiltin<"__builtin_ia32_punpcklqdq256_mask">, - Intrinsic<[llvm_v4i64_ty], - [llvm_v4i64_ty, llvm_v4i64_ty, llvm_v4i64_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpcklqd_q_512 : - GCCBuiltin<"__builtin_ia32_punpcklqdq512_mask">, - Intrinsic<[llvm_v8i64_ty], - [llvm_v8i64_ty, llvm_v8i64_ty, llvm_v8i64_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpcklw_d_128 : - GCCBuiltin<"__builtin_ia32_punpcklwd128_mask">, - Intrinsic<[llvm_v8i16_ty], - [llvm_v8i16_ty, llvm_v8i16_ty, llvm_v8i16_ty, llvm_i8_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpcklw_d_256 : - GCCBuiltin<"__builtin_ia32_punpcklwd256_mask">, - Intrinsic<[llvm_v16i16_ty], - [llvm_v16i16_ty, llvm_v16i16_ty, llvm_v16i16_ty, llvm_i16_ty], - [IntrNoMem]>; - - def int_x86_avx512_mask_punpcklw_d_512 : - GCCBuiltin<"__builtin_ia32_punpcklwd512_mask">, - Intrinsic<[llvm_v32i16_ty], - [llvm_v32i16_ty, llvm_v32i16_ty, llvm_v32i16_ty, llvm_i32_ty], - [IntrNoMem]>; -} - // Vector convert let TargetPrefix = "x86" in { // All intrinsics start with "llvm.x86.". def int_x86_avx512_mask_cvtdq2pd_128 : diff --git a/lib/IR/AutoUpgrade.cpp b/lib/IR/AutoUpgrade.cpp index a4fc03cf029..8ba7ef4d26e 100644 --- a/lib/IR/AutoUpgrade.cpp +++ b/lib/IR/AutoUpgrade.cpp @@ -197,6 +197,10 @@ static bool UpgradeIntrinsicFunction1(Function *F, Function *&NewFn) { Name.startswith("x86.avx512.mask.pshuf.d.") || Name.startswith("x86.avx512.mask.pshufl.w.") || Name.startswith("x86.avx512.mask.pshufh.w.") || + Name.startswith("x86.avx512.mask.punpckl") || + Name.startswith("x86.avx512.mask.punpckh") || + Name.startswith("x86.avx512.mask.unpckl.") || + Name.startswith("x86.avx512.mask.unpckh.") || Name.startswith("x86.sse41.pmovsx") || Name.startswith("x86.sse41.pmovzx") || Name.startswith("x86.avx2.pmovsx") || @@ -1034,6 +1038,38 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) { if (CI->getNumArgOperands() == 4) Rep = EmitX86Select(Builder, CI->getArgOperand(3), Rep, CI->getArgOperand(2)); + } else if (Name.startswith("llvm.x86.avx512.mask.punpckl") || + Name.startswith("llvm.x86.avx512.mask.unpckl.")) { + Value *Op0 = CI->getArgOperand(0); + Value *Op1 = CI->getArgOperand(1); + int NumElts = CI->getType()->getVectorNumElements(); + int NumLaneElts = 128/CI->getType()->getScalarSizeInBits(); + + SmallVector Idxs(NumElts); + for (int l = 0; l != NumElts; l += NumLaneElts) + for (int i = 0; i != NumLaneElts; ++i) + Idxs[i + l] = l + (i / 2) + NumElts * (i % 2); + + Rep = Builder.CreateShuffleVector(Op0, Op1, Idxs); + + Rep = EmitX86Select(Builder, CI->getArgOperand(3), Rep, + CI->getArgOperand(2)); + } else if (Name.startswith("llvm.x86.avx512.mask.punpckh") || + Name.startswith("llvm.x86.avx512.mask.unpckh.")) { + Value *Op0 = CI->getArgOperand(0); + Value *Op1 = CI->getArgOperand(1); + int NumElts = CI->getType()->getVectorNumElements(); + int NumLaneElts = 128/CI->getType()->getScalarSizeInBits(); + + SmallVector Idxs(NumElts); + for (int l = 0; l != NumElts; l += NumLaneElts) + for (int i = 0; i != NumLaneElts; ++i) + Idxs[i + l] = (NumLaneElts / 2) + l + (i / 2) + NumElts * (i % 2); + + Rep = Builder.CreateShuffleVector(Op0, Op1, Idxs); + + Rep = EmitX86Select(Builder, CI->getArgOperand(3), Rep, + CI->getArgOperand(2)); } else { llvm_unreachable("Unknown function for CallInst upgrade."); } diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp index 54e5536c36e..f0b1901d77a 100644 --- a/lib/Target/X86/X86ISelLowering.cpp +++ b/lib/Target/X86/X86ISelLowering.cpp @@ -11822,6 +11822,11 @@ static SDValue lowerV32I16VectorShuffle(const SDLoc &DL, ArrayRef Mask, assert(Mask.size() == 32 && "Unexpected mask size for v32 shuffle!"); assert(Subtarget.hasBWI() && "We can only lower v32i16 with AVX-512-BWI!"); + // Use dedicated unpack instructions for masks that match their pattern. + if (SDValue V = + lowerVectorShuffleWithUNPCK(DL, MVT::v32i16, Mask, V1, V2, DAG)) + return V; + // Try to use shift instructions. if (SDValue Shift = lowerVectorShuffleAsShift(DL, MVT::v32i16, V1, V2, Mask, Subtarget, DAG)) @@ -11856,6 +11861,11 @@ static SDValue lowerV64I8VectorShuffle(const SDLoc &DL, ArrayRef Mask, assert(Mask.size() == 64 && "Unexpected mask size for v64 shuffle!"); assert(Subtarget.hasBWI() && "We can only lower v64i8 with AVX-512-BWI!"); + // Use dedicated unpack instructions for masks that match their pattern. + if (SDValue V = + lowerVectorShuffleWithUNPCK(DL, MVT::v64i8, Mask, V1, V2, DAG)) + return V; + // Try to use shift instructions. if (SDValue Shift = lowerVectorShuffleAsShift(DL, MVT::v64i8, V1, V2, Mask, Subtarget, DAG)) diff --git a/lib/Target/X86/X86IntrinsicsInfo.h b/lib/Target/X86/X86IntrinsicsInfo.h index 071709a1a73..341775cf4eb 100644 --- a/lib/Target/X86/X86IntrinsicsInfo.h +++ b/lib/Target/X86/X86IntrinsicsInfo.h @@ -1415,54 +1415,6 @@ static const IntrinsicData IntrinsicsWithoutChain[] = { X86ISD::VPTERNLOG, 0), X86_INTRINSIC_DATA(avx512_mask_pternlog_q_512, TERLOG_OP_MASK, X86ISD::VPTERNLOG, 0), - X86_INTRINSIC_DATA(avx512_mask_punpckhb_w_128, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_punpckhb_w_256, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_punpckhb_w_512, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_punpckhd_q_128, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_punpckhd_q_256, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_punpckhd_q_512, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_punpckhqd_q_128, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_punpckhqd_q_256, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_punpckhqd_q_512, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_punpckhw_d_128, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_punpckhw_d_256, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_punpckhw_d_512, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_punpcklb_w_128, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_punpcklb_w_256, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_punpcklb_w_512, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_punpckld_q_128, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_punpckld_q_256, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_punpckld_q_512, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_punpcklqd_q_128, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_punpcklqd_q_256, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_punpcklqd_q_512, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_punpcklw_d_128, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_punpcklw_d_256, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_punpcklw_d_512, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), X86_INTRINSIC_DATA(avx512_mask_pxor_d_128, INTR_TYPE_2OP_MASK, ISD::XOR, 0), X86_INTRINSIC_DATA(avx512_mask_pxor_d_256, INTR_TYPE_2OP_MASK, ISD::XOR, 0), X86_INTRINSIC_DATA(avx512_mask_pxor_d_512, INTR_TYPE_2OP_MASK, ISD::XOR, 0), @@ -1575,30 +1527,6 @@ static const IntrinsicData IntrinsicsWithoutChain[] = { X86_INTRINSIC_DATA(avx512_mask_ucmp_w_128, CMP_MASK_CC, X86ISD::CMPMU, 0), X86_INTRINSIC_DATA(avx512_mask_ucmp_w_256, CMP_MASK_CC, X86ISD::CMPMU, 0), X86_INTRINSIC_DATA(avx512_mask_ucmp_w_512, CMP_MASK_CC, X86ISD::CMPMU, 0), - X86_INTRINSIC_DATA(avx512_mask_unpckh_pd_128, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_unpckh_pd_256, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_unpckh_pd_512, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_unpckh_ps_128, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_unpckh_ps_256, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_unpckh_ps_512, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKH, 0), - X86_INTRINSIC_DATA(avx512_mask_unpckl_pd_128, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_unpckl_pd_256, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_unpckl_pd_512, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_unpckl_ps_128, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_unpckl_ps_256, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), - X86_INTRINSIC_DATA(avx512_mask_unpckl_ps_512, INTR_TYPE_2OP_MASK, - X86ISD::UNPCKL, 0), X86_INTRINSIC_DATA(avx512_mask_valign_d_128, INTR_TYPE_3OP_IMM8_MASK, X86ISD::VALIGN, 0), X86_INTRINSIC_DATA(avx512_mask_valign_d_256, INTR_TYPE_3OP_IMM8_MASK, diff --git a/test/CodeGen/X86/avx512-intrinsics-upgrade.ll b/test/CodeGen/X86/avx512-intrinsics-upgrade.ll index d8ebd5538ee..8ffc984b0cc 100644 --- a/test/CodeGen/X86/avx512-intrinsics-upgrade.ll +++ b/test/CodeGen/X86/avx512-intrinsics-upgrade.ll @@ -368,3 +368,136 @@ define i8 @test_mask_pcmpgt_q(<8 x i64> %a, <8 x i64> %b, i8 %mask) { } declare i8 @llvm.x86.avx512.mask.pcmpgt.q.512(<8 x i64>, <8 x i64>, i8) + +declare <8 x double> @llvm.x86.avx512.mask.unpckh.pd.512(<8 x double>, <8 x double>, <8 x double>, i8) + +define <8 x double>@test_int_x86_avx512_mask_unpckh_pd_512(<8 x double> %x0, <8 x double> %x1, <8 x double> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_unpckh_pd_512: +; CHECK: ## BB#0: +; CHECK-NEXT: vunpckhpd {{.*#+}} zmm3 = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7] +; CHECK-NEXT: kmovw %edi, %k1 +; CHECK-NEXT: vunpckhpd {{.*#+}} zmm2 = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7] +; CHECK-NEXT: vaddpd %zmm3, %zmm2, %zmm0 +; CHECK-NEXT: retq + %res = call <8 x double> @llvm.x86.avx512.mask.unpckh.pd.512(<8 x double> %x0, <8 x double> %x1, <8 x double> %x2, i8 %x3) + %res1 = call <8 x double> @llvm.x86.avx512.mask.unpckh.pd.512(<8 x double> %x0, <8 x double> %x1, <8 x double> %x2, i8 -1) + %res2 = fadd <8 x double> %res, %res1 + ret <8 x double> %res2 +} + +declare <16 x float> @llvm.x86.avx512.mask.unpckh.ps.512(<16 x float>, <16 x float>, <16 x float>, i16) + +define <16 x float>@test_int_x86_avx512_mask_unpckh_ps_512(<16 x float> %x0, <16 x float> %x1, <16 x float> %x2, i16 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_unpckh_ps_512: +; CHECK: ## BB#0: +; CHECK-NEXT: vunpckhps {{.*#+}} zmm3 = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15] +; CHECK-NEXT: kmovw %edi, %k1 +; CHECK-NEXT: vunpckhps {{.*#+}} zmm2 = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15] +; CHECK-NEXT: vaddps %zmm3, %zmm2, %zmm0 +; CHECK-NEXT: retq + %res = call <16 x float> @llvm.x86.avx512.mask.unpckh.ps.512(<16 x float> %x0, <16 x float> %x1, <16 x float> %x2, i16 %x3) + %res1 = call <16 x float> @llvm.x86.avx512.mask.unpckh.ps.512(<16 x float> %x0, <16 x float> %x1, <16 x float> %x2, i16 -1) + %res2 = fadd <16 x float> %res, %res1 + ret <16 x float> %res2 +} + +declare <8 x double> @llvm.x86.avx512.mask.unpckl.pd.512(<8 x double>, <8 x double>, <8 x double>, i8) + +define <8 x double>@test_int_x86_avx512_mask_unpckl_pd_512(<8 x double> %x0, <8 x double> %x1, <8 x double> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_unpckl_pd_512: +; CHECK: ## BB#0: +; CHECK-NEXT: vunpcklpd {{.*#+}} zmm3 = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6] +; CHECK-NEXT: kmovw %edi, %k1 +; CHECK-NEXT: vunpcklpd {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6] +; CHECK-NEXT: vaddpd %zmm3, %zmm2, %zmm0 +; CHECK-NEXT: retq + %res = call <8 x double> @llvm.x86.avx512.mask.unpckl.pd.512(<8 x double> %x0, <8 x double> %x1, <8 x double> %x2, i8 %x3) + %res1 = call <8 x double> @llvm.x86.avx512.mask.unpckl.pd.512(<8 x double> %x0, <8 x double> %x1, <8 x double> %x2, i8 -1) + %res2 = fadd <8 x double> %res, %res1 + ret <8 x double> %res2 +} + +declare <16 x float> @llvm.x86.avx512.mask.unpckl.ps.512(<16 x float>, <16 x float>, <16 x float>, i16) + +define <16 x float>@test_int_x86_avx512_mask_unpckl_ps_512(<16 x float> %x0, <16 x float> %x1, <16 x float> %x2, i16 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_unpckl_ps_512: +; CHECK: ## BB#0: +; CHECK-NEXT: vunpcklps {{.*#+}} zmm3 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13] +; CHECK-NEXT: kmovw %edi, %k1 +; CHECK-NEXT: vunpcklps {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13] +; CHECK-NEXT: vaddps %zmm3, %zmm2, %zmm0 +; CHECK-NEXT: retq + %res = call <16 x float> @llvm.x86.avx512.mask.unpckl.ps.512(<16 x float> %x0, <16 x float> %x1, <16 x float> %x2, i16 %x3) + %res1 = call <16 x float> @llvm.x86.avx512.mask.unpckl.ps.512(<16 x float> %x0, <16 x float> %x1, <16 x float> %x2, i16 -1) + %res2 = fadd <16 x float> %res, %res1 + ret <16 x float> %res2 +} + +declare <8 x i64> @llvm.x86.avx512.mask.punpcklqd.q.512(<8 x i64>, <8 x i64>, <8 x i64>, i8) + +define <8 x i64>@test_int_x86_avx512_mask_punpcklqd_q_512(<8 x i64> %x0, <8 x i64> %x1, <8 x i64> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpcklqd_q_512: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpcklqdq {{.*#+}} zmm3 = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6] +; CHECK-NEXT: kmovw %edi, %k1 +; CHECK-NEXT: vpunpcklqdq {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6] +; CHECK-NEXT: vpunpcklqdq {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6] +; CHECK-NEXT: vpaddq %zmm3, %zmm2, %zmm1 +; CHECK-NEXT: vpaddq %zmm1, %zmm0, %zmm0 +; CHECK-NEXT: retq + %res = call <8 x i64> @llvm.x86.avx512.mask.punpcklqd.q.512(<8 x i64> %x0, <8 x i64> %x1, <8 x i64> %x2, i8 %x3) + %res1 = call <8 x i64> @llvm.x86.avx512.mask.punpcklqd.q.512(<8 x i64> %x0, <8 x i64> %x1, <8 x i64> %x2, i8 -1) + %res2 = call <8 x i64> @llvm.x86.avx512.mask.punpcklqd.q.512(<8 x i64> %x0, <8 x i64> %x1, <8 x i64> zeroinitializer,i8 %x3) + %res3 = add <8 x i64> %res, %res1 + %res4 = add <8 x i64> %res2, %res3 + ret <8 x i64> %res4 +} + +declare <8 x i64> @llvm.x86.avx512.mask.punpckhqd.q.512(<8 x i64>, <8 x i64>, <8 x i64>, i8) + +define <8 x i64>@test_int_x86_avx512_mask_punpckhqd_q_512(<8 x i64> %x0, <8 x i64> %x1, <8 x i64> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpckhqd_q_512: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpckhqdq {{.*#+}} zmm3 = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7] +; CHECK-NEXT: kmovw %edi, %k1 +; CHECK-NEXT: vpunpckhqdq {{.*#+}} zmm2 = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7] +; CHECK-NEXT: vpaddq %zmm3, %zmm2, %zmm0 +; CHECK-NEXT: retq + %res = call <8 x i64> @llvm.x86.avx512.mask.punpckhqd.q.512(<8 x i64> %x0, <8 x i64> %x1, <8 x i64> %x2, i8 %x3) + %res1 = call <8 x i64> @llvm.x86.avx512.mask.punpckhqd.q.512(<8 x i64> %x0, <8 x i64> %x1, <8 x i64> %x2, i8 -1) + %res2 = add <8 x i64> %res, %res1 + ret <8 x i64> %res2 +} + +declare <16 x i32> @llvm.x86.avx512.mask.punpckhd.q.512(<16 x i32>, <16 x i32>, <16 x i32>, i16) + +define <16 x i32>@test_int_x86_avx512_mask_punpckhd_q_512(<16 x i32> %x0, <16 x i32> %x1, <16 x i32> %x2, i16 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpckhd_q_512: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpckhdq {{.*#+}} zmm3 = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15] +; CHECK-NEXT: kmovw %edi, %k1 +; CHECK-NEXT: vpunpckhdq {{.*#+}} zmm2 = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15] +; CHECK-NEXT: vpaddd %zmm3, %zmm2, %zmm0 +; CHECK-NEXT: retq + %res = call <16 x i32> @llvm.x86.avx512.mask.punpckhd.q.512(<16 x i32> %x0, <16 x i32> %x1, <16 x i32> %x2, i16 %x3) + %res1 = call <16 x i32> @llvm.x86.avx512.mask.punpckhd.q.512(<16 x i32> %x0, <16 x i32> %x1, <16 x i32> %x2, i16 -1) + %res2 = add <16 x i32> %res, %res1 + ret <16 x i32> %res2 +} + +declare <16 x i32> @llvm.x86.avx512.mask.punpckld.q.512(<16 x i32>, <16 x i32>, <16 x i32>, i16) + +define <16 x i32>@test_int_x86_avx512_mask_punpckld_q_512(<16 x i32> %x0, <16 x i32> %x1, <16 x i32> %x2, i16 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpckld_q_512: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpckldq {{.*#+}} zmm3 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13] +; CHECK-NEXT: kmovw %edi, %k1 +; CHECK-NEXT: vpunpckldq {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13] +; CHECK-NEXT: vpaddd %zmm3, %zmm2, %zmm0 +; CHECK-NEXT: retq + %res = call <16 x i32> @llvm.x86.avx512.mask.punpckld.q.512(<16 x i32> %x0, <16 x i32> %x1, <16 x i32> %x2, i16 %x3) + %res1 = call <16 x i32> @llvm.x86.avx512.mask.punpckld.q.512(<16 x i32> %x0, <16 x i32> %x1, <16 x i32> %x2, i16 -1) + %res2 = add <16 x i32> %res, %res1 + ret <16 x i32> %res2 +} + diff --git a/test/CodeGen/X86/avx512-intrinsics.ll b/test/CodeGen/X86/avx512-intrinsics.ll index 015dbd32c7d..a1460a43012 100644 --- a/test/CodeGen/X86/avx512-intrinsics.ll +++ b/test/CodeGen/X86/avx512-intrinsics.ll @@ -4373,138 +4373,6 @@ define <16 x float>@test_int_x86_avx512_mask_scalef_ps_512(<16 x float> %x0, <16 ret <16 x float> %res2 } -declare <8 x double> @llvm.x86.avx512.mask.unpckh.pd.512(<8 x double>, <8 x double>, <8 x double>, i8) - -define <8 x double>@test_int_x86_avx512_mask_unpckh_pd_512(<8 x double> %x0, <8 x double> %x1, <8 x double> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_unpckh_pd_512: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 -; CHECK-NEXT: vunpckhpd {{.*#+}} zmm2 = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7] -; CHECK-NEXT: vunpckhpd {{.*#+}} zmm0 = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7] -; CHECK-NEXT: vaddpd %zmm0, %zmm2, %zmm0 -; CHECK-NEXT: retq - %res = call <8 x double> @llvm.x86.avx512.mask.unpckh.pd.512(<8 x double> %x0, <8 x double> %x1, <8 x double> %x2, i8 %x3) - %res1 = call <8 x double> @llvm.x86.avx512.mask.unpckh.pd.512(<8 x double> %x0, <8 x double> %x1, <8 x double> %x2, i8 -1) - %res2 = fadd <8 x double> %res, %res1 - ret <8 x double> %res2 -} - -declare <16 x float> @llvm.x86.avx512.mask.unpckh.ps.512(<16 x float>, <16 x float>, <16 x float>, i16) - -define <16 x float>@test_int_x86_avx512_mask_unpckh_ps_512(<16 x float> %x0, <16 x float> %x1, <16 x float> %x2, i16 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_unpckh_ps_512: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 -; CHECK-NEXT: vunpckhps {{.*#+}} zmm2 = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15] -; CHECK-NEXT: vunpckhps {{.*#+}} zmm0 = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15] -; CHECK-NEXT: vaddps %zmm0, %zmm2, %zmm0 -; CHECK-NEXT: retq - %res = call <16 x float> @llvm.x86.avx512.mask.unpckh.ps.512(<16 x float> %x0, <16 x float> %x1, <16 x float> %x2, i16 %x3) - %res1 = call <16 x float> @llvm.x86.avx512.mask.unpckh.ps.512(<16 x float> %x0, <16 x float> %x1, <16 x float> %x2, i16 -1) - %res2 = fadd <16 x float> %res, %res1 - ret <16 x float> %res2 -} - -declare <8 x double> @llvm.x86.avx512.mask.unpckl.pd.512(<8 x double>, <8 x double>, <8 x double>, i8) - -define <8 x double>@test_int_x86_avx512_mask_unpckl_pd_512(<8 x double> %x0, <8 x double> %x1, <8 x double> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_unpckl_pd_512: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 -; CHECK-NEXT: vunpcklpd {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6] -; CHECK-NEXT: vunpcklpd {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6] -; CHECK-NEXT: vaddpd %zmm0, %zmm2, %zmm0 -; CHECK-NEXT: retq - %res = call <8 x double> @llvm.x86.avx512.mask.unpckl.pd.512(<8 x double> %x0, <8 x double> %x1, <8 x double> %x2, i8 %x3) - %res1 = call <8 x double> @llvm.x86.avx512.mask.unpckl.pd.512(<8 x double> %x0, <8 x double> %x1, <8 x double> %x2, i8 -1) - %res2 = fadd <8 x double> %res, %res1 - ret <8 x double> %res2 -} - -declare <16 x float> @llvm.x86.avx512.mask.unpckl.ps.512(<16 x float>, <16 x float>, <16 x float>, i16) - -define <16 x float>@test_int_x86_avx512_mask_unpckl_ps_512(<16 x float> %x0, <16 x float> %x1, <16 x float> %x2, i16 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_unpckl_ps_512: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 -; CHECK-NEXT: vunpcklps {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13] -; CHECK-NEXT: vunpcklps {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13] -; CHECK-NEXT: vaddps %zmm0, %zmm2, %zmm0 -; CHECK-NEXT: retq - %res = call <16 x float> @llvm.x86.avx512.mask.unpckl.ps.512(<16 x float> %x0, <16 x float> %x1, <16 x float> %x2, i16 %x3) - %res1 = call <16 x float> @llvm.x86.avx512.mask.unpckl.ps.512(<16 x float> %x0, <16 x float> %x1, <16 x float> %x2, i16 -1) - %res2 = fadd <16 x float> %res, %res1 - ret <16 x float> %res2 -} - -declare <8 x i64> @llvm.x86.avx512.mask.punpcklqd.q.512(<8 x i64>, <8 x i64>, <8 x i64>, i8) - -define <8 x i64>@test_int_x86_avx512_mask_punpcklqd_q_512(<8 x i64> %x0, <8 x i64> %x1, <8 x i64> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpcklqd_q_512: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 -; CHECK-NEXT: vpunpcklqdq {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6] -; CHECK-NEXT: vpunpcklqdq {{.*#+}} zmm3 = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6] -; CHECK-NEXT: vpunpcklqdq {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[2],zmm1[2],zmm0[4],zmm1[4],zmm0[6],zmm1[6] -; CHECK-NEXT: vpaddq %zmm0, %zmm2, %zmm0 -; CHECK-NEXT: vpaddq %zmm0, %zmm3, %zmm0 -; CHECK-NEXT: retq - %res = call <8 x i64> @llvm.x86.avx512.mask.punpcklqd.q.512(<8 x i64> %x0, <8 x i64> %x1, <8 x i64> %x2, i8 %x3) - %res1 = call <8 x i64> @llvm.x86.avx512.mask.punpcklqd.q.512(<8 x i64> %x0, <8 x i64> %x1, <8 x i64> %x2, i8 -1) - %res2 = call <8 x i64> @llvm.x86.avx512.mask.punpcklqd.q.512(<8 x i64> %x0, <8 x i64> %x1, <8 x i64> zeroinitializer,i8 %x3) - %res3 = add <8 x i64> %res, %res1 - %res4 = add <8 x i64> %res2, %res3 - ret <8 x i64> %res4 -} - -declare <8 x i64> @llvm.x86.avx512.mask.punpckhqd.q.512(<8 x i64>, <8 x i64>, <8 x i64>, i8) - -define <8 x i64>@test_int_x86_avx512_mask_punpckhqd_q_512(<8 x i64> %x0, <8 x i64> %x1, <8 x i64> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpckhqd_q_512: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 -; CHECK-NEXT: vpunpckhqdq {{.*#+}} zmm2 = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7] -; CHECK-NEXT: vpunpckhqdq {{.*#+}} zmm0 = zmm0[1],zmm1[1],zmm0[3],zmm1[3],zmm0[5],zmm1[5],zmm0[7],zmm1[7] -; CHECK-NEXT: vpaddq %zmm0, %zmm2, %zmm0 -; CHECK-NEXT: retq - %res = call <8 x i64> @llvm.x86.avx512.mask.punpckhqd.q.512(<8 x i64> %x0, <8 x i64> %x1, <8 x i64> %x2, i8 %x3) - %res1 = call <8 x i64> @llvm.x86.avx512.mask.punpckhqd.q.512(<8 x i64> %x0, <8 x i64> %x1, <8 x i64> %x2, i8 -1) - %res2 = add <8 x i64> %res, %res1 - ret <8 x i64> %res2 -} - -declare <16 x i32> @llvm.x86.avx512.mask.punpckhd.q.512(<16 x i32>, <16 x i32>, <16 x i32>, i16) - -define <16 x i32>@test_int_x86_avx512_mask_punpckhd_q_512(<16 x i32> %x0, <16 x i32> %x1, <16 x i32> %x2, i16 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpckhd_q_512: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 -; CHECK-NEXT: vpunpckhdq {{.*#+}} zmm2 = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15] -; CHECK-NEXT: vpunpckhdq {{.*#+}} zmm0 = zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[14],zmm1[14],zmm0[15],zmm1[15] -; CHECK-NEXT: vpaddd %zmm0, %zmm2, %zmm0 -; CHECK-NEXT: retq - %res = call <16 x i32> @llvm.x86.avx512.mask.punpckhd.q.512(<16 x i32> %x0, <16 x i32> %x1, <16 x i32> %x2, i16 %x3) - %res1 = call <16 x i32> @llvm.x86.avx512.mask.punpckhd.q.512(<16 x i32> %x0, <16 x i32> %x1, <16 x i32> %x2, i16 -1) - %res2 = add <16 x i32> %res, %res1 - ret <16 x i32> %res2 -} - -declare <16 x i32> @llvm.x86.avx512.mask.punpckld.q.512(<16 x i32>, <16 x i32>, <16 x i32>, i16) - -define <16 x i32>@test_int_x86_avx512_mask_punpckld_q_512(<16 x i32> %x0, <16 x i32> %x1, <16 x i32> %x2, i16 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpckld_q_512: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 -; CHECK-NEXT: vpunpckldq {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13] -; CHECK-NEXT: vpunpckldq {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[12],zmm1[12],zmm0[13],zmm1[13] -; CHECK-NEXT: vpaddd %zmm0, %zmm2, %zmm0 -; CHECK-NEXT: retq - %res = call <16 x i32> @llvm.x86.avx512.mask.punpckld.q.512(<16 x i32> %x0, <16 x i32> %x1, <16 x i32> %x2, i16 %x3) - %res1 = call <16 x i32> @llvm.x86.avx512.mask.punpckld.q.512(<16 x i32> %x0, <16 x i32> %x1, <16 x i32> %x2, i16 -1) - %res2 = add <16 x i32> %res, %res1 - ret <16 x i32> %res2 -} - declare <16 x i8> @llvm.x86.avx512.mask.pmov.qb.512(<8 x i64>, <16 x i8>, i8) define <16 x i8>@test_int_x86_avx512_mask_pmov_qb_512(<8 x i64> %x0, <16 x i8> %x1, i8 %x2) { diff --git a/test/CodeGen/X86/avx512bw-intrinsics-upgrade.ll b/test/CodeGen/X86/avx512bw-intrinsics-upgrade.ll index 9c70139ce59..f8e51d35043 100644 --- a/test/CodeGen/X86/avx512bw-intrinsics-upgrade.ll +++ b/test/CodeGen/X86/avx512bw-intrinsics-upgrade.ll @@ -440,3 +440,99 @@ define i32 @test_mask_pcmpgt_w(<32 x i16> %a, <32 x i16> %b, i32 %mask) { declare i32 @llvm.x86.avx512.mask.pcmpgt.w.512(<32 x i16>, <32 x i16>, i32) +declare <64 x i8> @llvm.x86.avx512.mask.punpckhb.w.512(<64 x i8>, <64 x i8>, <64 x i8>, i64) + +define <64 x i8>@test_int_x86_avx512_mask_punpckhb_w_512(<64 x i8> %x0, <64 x i8> %x1, <64 x i8> %x2, i64 %x3) { +; AVX512BW-LABEL: test_int_x86_avx512_mask_punpckhb_w_512: +; AVX512BW: ## BB#0: +; AVX512BW-NEXT: vpunpckhbw {{.*#+}} zmm3 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63] +; AVX512BW-NEXT: kmovq %rdi, %k1 +; AVX512BW-NEXT: vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63] +; AVX512BW-NEXT: vpaddb %zmm3, %zmm2, %zmm0 +; AVX512BW-NEXT: retq +; +; AVX512F-32-LABEL: test_int_x86_avx512_mask_punpckhb_w_512: +; AVX512F-32: # BB#0: +; AVX512F-32-NEXT: vpunpckhbw {{.*#+}} zmm3 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63] +; AVX512F-32-NEXT: kmovq {{[0-9]+}}(%esp), %k1 +; AVX512F-32-NEXT: vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63] +; AVX512F-32-NEXT: vpaddb %zmm3, %zmm2, %zmm0 +; AVX512F-32-NEXT: retl + %res = call <64 x i8> @llvm.x86.avx512.mask.punpckhb.w.512(<64 x i8> %x0, <64 x i8> %x1, <64 x i8> %x2, i64 %x3) + %res1 = call <64 x i8> @llvm.x86.avx512.mask.punpckhb.w.512(<64 x i8> %x0, <64 x i8> %x1, <64 x i8> %x2, i64 -1) + %res2 = add <64 x i8> %res, %res1 + ret <64 x i8> %res2 +} + +declare <64 x i8> @llvm.x86.avx512.mask.punpcklb.w.512(<64 x i8>, <64 x i8>, <64 x i8>, i64) + +define <64 x i8>@test_int_x86_avx512_mask_punpcklb_w_512(<64 x i8> %x0, <64 x i8> %x1, <64 x i8> %x2, i64 %x3) { +; AVX512BW-LABEL: test_int_x86_avx512_mask_punpcklb_w_512: +; AVX512BW: ## BB#0: +; AVX512BW-NEXT: vpunpcklbw {{.*#+}} zmm3 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55] +; AVX512BW-NEXT: kmovq %rdi, %k1 +; AVX512BW-NEXT: vpunpcklbw {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55] +; AVX512BW-NEXT: vpaddb %zmm3, %zmm2, %zmm0 +; AVX512BW-NEXT: retq +; +; AVX512F-32-LABEL: test_int_x86_avx512_mask_punpcklb_w_512: +; AVX512F-32: # BB#0: +; AVX512F-32-NEXT: vpunpcklbw {{.*#+}} zmm3 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55] +; AVX512F-32-NEXT: kmovq {{[0-9]+}}(%esp), %k1 +; AVX512F-32-NEXT: vpunpcklbw {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55] +; AVX512F-32-NEXT: vpaddb %zmm3, %zmm2, %zmm0 +; AVX512F-32-NEXT: retl + %res = call <64 x i8> @llvm.x86.avx512.mask.punpcklb.w.512(<64 x i8> %x0, <64 x i8> %x1, <64 x i8> %x2, i64 %x3) + %res1 = call <64 x i8> @llvm.x86.avx512.mask.punpcklb.w.512(<64 x i8> %x0, <64 x i8> %x1, <64 x i8> %x2, i64 -1) + %res2 = add <64 x i8> %res, %res1 + ret <64 x i8> %res2 +} + +declare <32 x i16> @llvm.x86.avx512.mask.punpckhw.d.512(<32 x i16>, <32 x i16>, <32 x i16>, i32) + +define <32 x i16>@test_int_x86_avx512_mask_punpckhw_d_512(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 %x3) { +; AVX512BW-LABEL: test_int_x86_avx512_mask_punpckhw_d_512: +; AVX512BW: ## BB#0: +; AVX512BW-NEXT: vpunpckhwd {{.*#+}} zmm3 = zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31] +; AVX512BW-NEXT: kmovd %edi, %k1 +; AVX512BW-NEXT: vpunpckhwd {{.*#+}} zmm2 = zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31] +; AVX512BW-NEXT: vpaddw %zmm3, %zmm2, %zmm0 +; AVX512BW-NEXT: retq +; +; AVX512F-32-LABEL: test_int_x86_avx512_mask_punpckhw_d_512: +; AVX512F-32: # BB#0: +; AVX512F-32-NEXT: vpunpckhwd {{.*#+}} zmm3 = zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31] +; AVX512F-32-NEXT: kmovd {{[0-9]+}}(%esp), %k1 +; AVX512F-32-NEXT: vpunpckhwd {{.*#+}} zmm2 = zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31] +; AVX512F-32-NEXT: vpaddw %zmm3, %zmm2, %zmm0 +; AVX512F-32-NEXT: retl + %res = call <32 x i16> @llvm.x86.avx512.mask.punpckhw.d.512(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 %x3) + %res1 = call <32 x i16> @llvm.x86.avx512.mask.punpckhw.d.512(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 -1) + %res2 = add <32 x i16> %res, %res1 + ret <32 x i16> %res2 +} + +declare <32 x i16> @llvm.x86.avx512.mask.punpcklw.d.512(<32 x i16>, <32 x i16>, <32 x i16>, i32) + +define <32 x i16>@test_int_x86_avx512_mask_punpcklw_d_512(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 %x3) { +; AVX512BW-LABEL: test_int_x86_avx512_mask_punpcklw_d_512: +; AVX512BW: ## BB#0: +; AVX512BW-NEXT: vpunpcklwd {{.*#+}} zmm3 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27] +; AVX512BW-NEXT: kmovd %edi, %k1 +; AVX512BW-NEXT: vpunpcklwd {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27] +; AVX512BW-NEXT: vpaddw %zmm3, %zmm2, %zmm0 +; AVX512BW-NEXT: retq +; +; AVX512F-32-LABEL: test_int_x86_avx512_mask_punpcklw_d_512: +; AVX512F-32: # BB#0: +; AVX512F-32-NEXT: vpunpcklwd {{.*#+}} zmm3 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27] +; AVX512F-32-NEXT: kmovd {{[0-9]+}}(%esp), %k1 +; AVX512F-32-NEXT: vpunpcklwd {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27] +; AVX512F-32-NEXT: vpaddw %zmm3, %zmm2, %zmm0 +; AVX512F-32-NEXT: retl + %res = call <32 x i16> @llvm.x86.avx512.mask.punpcklw.d.512(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 %x3) + %res1 = call <32 x i16> @llvm.x86.avx512.mask.punpcklw.d.512(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 -1) + %res2 = add <32 x i16> %res, %res1 + ret <32 x i16> %res2 +} + diff --git a/test/CodeGen/X86/avx512bw-intrinsics.ll b/test/CodeGen/X86/avx512bw-intrinsics.ll index b5e586e2bda..149f2917ad0 100644 --- a/test/CodeGen/X86/avx512bw-intrinsics.ll +++ b/test/CodeGen/X86/avx512bw-intrinsics.ll @@ -2301,106 +2301,6 @@ define <16 x i32>@test_int_x86_avx512_mask_pmaddw_d_512(<32 x i16> %x0, <32 x i1 ret <16 x i32> %res2 } -declare <64 x i8> @llvm.x86.avx512.mask.punpckhb.w.512(<64 x i8>, <64 x i8>, <64 x i8>, i64) - -define <64 x i8>@test_int_x86_avx512_mask_punpckhb_w_512(<64 x i8> %x0, <64 x i8> %x1, <64 x i8> %x2, i64 %x3) { -; AVX512BW-LABEL: test_int_x86_avx512_mask_punpckhb_w_512: -; AVX512BW: ## BB#0: -; AVX512BW-NEXT: kmovq %rdi, %k1 -; AVX512BW-NEXT: vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63] -; AVX512BW-NEXT: vpunpckhbw {{.*#+}} zmm0 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63] -; AVX512BW-NEXT: vpaddb %zmm0, %zmm2, %zmm0 -; AVX512BW-NEXT: retq -; -; AVX512F-32-LABEL: test_int_x86_avx512_mask_punpckhb_w_512: -; AVX512F-32: # BB#0: -; AVX512F-32-NEXT: kmovd {{[0-9]+}}(%esp), %k0 -; AVX512F-32-NEXT: kmovd {{[0-9]+}}(%esp), %k1 -; AVX512F-32-NEXT: kunpckdq %k0, %k1, %k1 -; AVX512F-32-NEXT: vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63] -; AVX512F-32-NEXT: vpunpckhbw {{.*#+}} zmm0 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63] -; AVX512F-32-NEXT: vpaddb %zmm0, %zmm2, %zmm0 -; AVX512F-32-NEXT: retl - %res = call <64 x i8> @llvm.x86.avx512.mask.punpckhb.w.512(<64 x i8> %x0, <64 x i8> %x1, <64 x i8> %x2, i64 %x3) - %res1 = call <64 x i8> @llvm.x86.avx512.mask.punpckhb.w.512(<64 x i8> %x0, <64 x i8> %x1, <64 x i8> %x2, i64 -1) - %res2 = add <64 x i8> %res, %res1 - ret <64 x i8> %res2 -} - -declare <64 x i8> @llvm.x86.avx512.mask.punpcklb.w.512(<64 x i8>, <64 x i8>, <64 x i8>, i64) - -define <64 x i8>@test_int_x86_avx512_mask_punpcklb_w_512(<64 x i8> %x0, <64 x i8> %x1, <64 x i8> %x2, i64 %x3) { -; AVX512BW-LABEL: test_int_x86_avx512_mask_punpcklb_w_512: -; AVX512BW: ## BB#0: -; AVX512BW-NEXT: kmovq %rdi, %k1 -; AVX512BW-NEXT: vpunpcklbw {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55] -; AVX512BW-NEXT: vpunpcklbw {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55] -; AVX512BW-NEXT: vpaddb %zmm0, %zmm2, %zmm0 -; AVX512BW-NEXT: retq -; -; AVX512F-32-LABEL: test_int_x86_avx512_mask_punpcklb_w_512: -; AVX512F-32: # BB#0: -; AVX512F-32-NEXT: kmovd {{[0-9]+}}(%esp), %k0 -; AVX512F-32-NEXT: kmovd {{[0-9]+}}(%esp), %k1 -; AVX512F-32-NEXT: kunpckdq %k0, %k1, %k1 -; AVX512F-32-NEXT: vpunpcklbw {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55] -; AVX512F-32-NEXT: vpunpcklbw {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55] -; AVX512F-32-NEXT: vpaddb %zmm0, %zmm2, %zmm0 -; AVX512F-32-NEXT: retl - %res = call <64 x i8> @llvm.x86.avx512.mask.punpcklb.w.512(<64 x i8> %x0, <64 x i8> %x1, <64 x i8> %x2, i64 %x3) - %res1 = call <64 x i8> @llvm.x86.avx512.mask.punpcklb.w.512(<64 x i8> %x0, <64 x i8> %x1, <64 x i8> %x2, i64 -1) - %res2 = add <64 x i8> %res, %res1 - ret <64 x i8> %res2 -} - -declare <32 x i16> @llvm.x86.avx512.mask.punpckhw.d.512(<32 x i16>, <32 x i16>, <32 x i16>, i32) - -define <32 x i16>@test_int_x86_avx512_mask_punpckhw_d_512(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 %x3) { -; AVX512BW-LABEL: test_int_x86_avx512_mask_punpckhw_d_512: -; AVX512BW: ## BB#0: -; AVX512BW-NEXT: kmovd %edi, %k1 -; AVX512BW-NEXT: vpunpckhwd {{.*#+}} zmm2 = zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31] -; AVX512BW-NEXT: vpunpckhwd {{.*#+}} zmm0 = zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31] -; AVX512BW-NEXT: vpaddw %zmm0, %zmm2, %zmm0 -; AVX512BW-NEXT: retq -; -; AVX512F-32-LABEL: test_int_x86_avx512_mask_punpckhw_d_512: -; AVX512F-32: # BB#0: -; AVX512F-32-NEXT: kmovd {{[0-9]+}}(%esp), %k1 -; AVX512F-32-NEXT: vpunpckhwd {{.*#+}} zmm2 = zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31] -; AVX512F-32-NEXT: vpunpckhwd {{.*#+}} zmm0 = zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31] -; AVX512F-32-NEXT: vpaddw %zmm0, %zmm2, %zmm0 -; AVX512F-32-NEXT: retl - %res = call <32 x i16> @llvm.x86.avx512.mask.punpckhw.d.512(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 %x3) - %res1 = call <32 x i16> @llvm.x86.avx512.mask.punpckhw.d.512(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 -1) - %res2 = add <32 x i16> %res, %res1 - ret <32 x i16> %res2 -} - -declare <32 x i16> @llvm.x86.avx512.mask.punpcklw.d.512(<32 x i16>, <32 x i16>, <32 x i16>, i32) - -define <32 x i16>@test_int_x86_avx512_mask_punpcklw_d_512(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 %x3) { -; AVX512BW-LABEL: test_int_x86_avx512_mask_punpcklw_d_512: -; AVX512BW: ## BB#0: -; AVX512BW-NEXT: kmovd %edi, %k1 -; AVX512BW-NEXT: vpunpcklwd {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27] -; AVX512BW-NEXT: vpunpcklwd {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27] -; AVX512BW-NEXT: vpaddw %zmm0, %zmm2, %zmm0 -; AVX512BW-NEXT: retq -; -; AVX512F-32-LABEL: test_int_x86_avx512_mask_punpcklw_d_512: -; AVX512F-32: # BB#0: -; AVX512F-32-NEXT: kmovd {{[0-9]+}}(%esp), %k1 -; AVX512F-32-NEXT: vpunpcklwd {{.*#+}} zmm2 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27] -; AVX512F-32-NEXT: vpunpcklwd {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27] -; AVX512F-32-NEXT: vpaddw %zmm0, %zmm2, %zmm0 -; AVX512F-32-NEXT: retl - %res = call <32 x i16> @llvm.x86.avx512.mask.punpcklw.d.512(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 %x3) - %res1 = call <32 x i16> @llvm.x86.avx512.mask.punpcklw.d.512(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 -1) - %res2 = add <32 x i16> %res, %res1 - ret <32 x i16> %res2 -} - declare <32 x i16> @llvm.x86.avx512.mask.dbpsadbw.512(<64 x i8>, <64 x i8>, i32, <32 x i16>, i32) define <32 x i16>@test_int_x86_avx512_mask_dbpsadbw_512(<64 x i8> %x0, <64 x i8> %x1, <32 x i16> %x3, i32 %x4) { diff --git a/test/CodeGen/X86/avx512bwvl-intrinsics-upgrade.ll b/test/CodeGen/X86/avx512bwvl-intrinsics-upgrade.ll index 6520805262c..ab65c412109 100644 --- a/test/CodeGen/X86/avx512bwvl-intrinsics-upgrade.ll +++ b/test/CodeGen/X86/avx512bwvl-intrinsics-upgrade.ll @@ -359,3 +359,147 @@ define i16 @test_mask_pcmpgt_w_256(<16 x i16> %a, <16 x i16> %b, i16 %mask) { declare i16 @llvm.x86.avx512.mask.pcmpgt.w.256(<16 x i16>, <16 x i16>, i16) +declare <16 x i8> @llvm.x86.avx512.mask.punpckhb.w.128(<16 x i8>, <16 x i8>, <16 x i8>, i16) + +define <16 x i8>@test_int_x86_avx512_mask_punpckhb_w_128(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %x2, i16 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpckhb_w_128: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpckhbw %xmm1, %xmm0, %xmm3 ## encoding: [0x62,0xf1,0x7d,0x08,0x68,0xd9] +; CHECK-NEXT: ## xmm3 = xmm0[8],xmm1[8],xmm0[9],xmm1[9],xmm0[10],xmm1[10],xmm0[11],xmm1[11],xmm0[12],xmm1[12],xmm0[13],xmm1[13],xmm0[14],xmm1[14],xmm0[15],xmm1[15] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vpunpckhbw %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x09,0x68,0xd1] +; CHECK-NEXT: ## xmm2 = xmm0[8],xmm1[8],xmm0[9],xmm1[9],xmm0[10],xmm1[10],xmm0[11],xmm1[11],xmm0[12],xmm1[12],xmm0[13],xmm1[13],xmm0[14],xmm1[14],xmm0[15],xmm1[15] +; CHECK-NEXT: vpaddb %xmm3, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6d,0x08,0xfc,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <16 x i8> @llvm.x86.avx512.mask.punpckhb.w.128(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %x2, i16 %x3) + %res1 = call <16 x i8> @llvm.x86.avx512.mask.punpckhb.w.128(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %x2, i16 -1) + %res2 = add <16 x i8> %res, %res1 + ret <16 x i8> %res2 +} + +declare <16 x i8> @llvm.x86.avx512.mask.punpcklb.w.128(<16 x i8>, <16 x i8>, <16 x i8>, i16) + +define <16 x i8>@test_int_x86_avx512_mask_punpcklb_w_128(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %x2, i16 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpcklb_w_128: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpcklbw %xmm1, %xmm0, %xmm3 ## encoding: [0x62,0xf1,0x7d,0x08,0x60,0xd9] +; CHECK-NEXT: ## xmm3 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vpunpcklbw %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x09,0x60,0xd1] +; CHECK-NEXT: ## xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7] +; CHECK-NEXT: vpaddb %xmm3, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6d,0x08,0xfc,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <16 x i8> @llvm.x86.avx512.mask.punpcklb.w.128(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %x2, i16 %x3) + %res1 = call <16 x i8> @llvm.x86.avx512.mask.punpcklb.w.128(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %x2, i16 -1) + %res2 = add <16 x i8> %res, %res1 + ret <16 x i8> %res2 +} + +declare <32 x i8> @llvm.x86.avx512.mask.punpckhb.w.256(<32 x i8>, <32 x i8>, <32 x i8>, i32) + +define <32 x i8>@test_int_x86_avx512_mask_punpckhb_w_256(<32 x i8> %x0, <32 x i8> %x1, <32 x i8> %x2, i32 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpckhb_w_256: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpckhbw %ymm1, %ymm0, %ymm3 ## encoding: [0x62,0xf1,0x7d,0x28,0x68,0xd9] +; CHECK-NEXT: ## ymm3 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31] +; CHECK-NEXT: kmovd %edi, %k1 ## encoding: [0xc5,0xfb,0x92,0xcf] +; CHECK-NEXT: vpunpckhbw %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x29,0x68,0xd1] +; CHECK-NEXT: ## ymm2 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31] +; CHECK-NEXT: vpaddb %ymm3, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6d,0x28,0xfc,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <32 x i8> @llvm.x86.avx512.mask.punpckhb.w.256(<32 x i8> %x0, <32 x i8> %x1, <32 x i8> %x2, i32 %x3) + %res1 = call <32 x i8> @llvm.x86.avx512.mask.punpckhb.w.256(<32 x i8> %x0, <32 x i8> %x1, <32 x i8> %x2, i32 -1) + %res2 = add <32 x i8> %res, %res1 + ret <32 x i8> %res2 +} + +declare <32 x i8> @llvm.x86.avx512.mask.punpcklb.w.256(<32 x i8>, <32 x i8>, <32 x i8>, i32) + +define <32 x i8>@test_int_x86_avx512_mask_punpcklb_w_256(<32 x i8> %x0, <32 x i8> %x1, <32 x i8> %x2, i32 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpcklb_w_256: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpcklbw %ymm1, %ymm0, %ymm3 ## encoding: [0x62,0xf1,0x7d,0x28,0x60,0xd9] +; CHECK-NEXT: ## ymm3 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23] +; CHECK-NEXT: kmovd %edi, %k1 ## encoding: [0xc5,0xfb,0x92,0xcf] +; CHECK-NEXT: vpunpcklbw %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x29,0x60,0xd1] +; CHECK-NEXT: ## ymm2 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23] +; CHECK-NEXT: vpaddb %ymm3, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6d,0x28,0xfc,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <32 x i8> @llvm.x86.avx512.mask.punpcklb.w.256(<32 x i8> %x0, <32 x i8> %x1, <32 x i8> %x2, i32 %x3) + %res1 = call <32 x i8> @llvm.x86.avx512.mask.punpcklb.w.256(<32 x i8> %x0, <32 x i8> %x1, <32 x i8> %x2, i32 -1) + %res2 = add <32 x i8> %res, %res1 + ret <32 x i8> %res2 +} + +declare <8 x i16> @llvm.x86.avx512.mask.punpcklw.d.128(<8 x i16>, <8 x i16>, <8 x i16>, i8) + +define <8 x i16>@test_int_x86_avx512_mask_punpcklw_d_128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpcklw_d_128: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpcklwd %xmm1, %xmm0, %xmm3 ## encoding: [0x62,0xf1,0x7d,0x08,0x61,0xd9] +; CHECK-NEXT: ## xmm3 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vpunpcklwd %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x09,0x61,0xd1] +; CHECK-NEXT: ## xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] +; CHECK-NEXT: vpaddw %xmm3, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6d,0x08,0xfd,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <8 x i16> @llvm.x86.avx512.mask.punpcklw.d.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3) + %res1 = call <8 x i16> @llvm.x86.avx512.mask.punpcklw.d.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 -1) + %res2 = add <8 x i16> %res, %res1 + ret <8 x i16> %res2 +} + +declare <8 x i16> @llvm.x86.avx512.mask.punpckhw.d.128(<8 x i16>, <8 x i16>, <8 x i16>, i8) + +define <8 x i16>@test_int_x86_avx512_mask_punpckhw_d_128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpckhw_d_128: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpckhwd %xmm1, %xmm0, %xmm3 ## encoding: [0x62,0xf1,0x7d,0x08,0x69,0xd9] +; CHECK-NEXT: ## xmm3 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vpunpckhwd %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x09,0x69,0xd1] +; CHECK-NEXT: ## xmm2 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7] +; CHECK-NEXT: vpaddw %xmm3, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6d,0x08,0xfd,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <8 x i16> @llvm.x86.avx512.mask.punpckhw.d.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3) + %res1 = call <8 x i16> @llvm.x86.avx512.mask.punpckhw.d.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 -1) + %res2 = add <8 x i16> %res, %res1 + ret <8 x i16> %res2 +} + +declare <16 x i16> @llvm.x86.avx512.mask.punpcklw.d.256(<16 x i16>, <16 x i16>, <16 x i16>, i16) + +define <16 x i16>@test_int_x86_avx512_mask_punpcklw_d_256(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpcklw_d_256: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpcklwd %ymm1, %ymm0, %ymm3 ## encoding: [0x62,0xf1,0x7d,0x28,0x61,0xd9] +; CHECK-NEXT: ## ymm3 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vpunpcklwd %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x29,0x61,0xd1] +; CHECK-NEXT: ## ymm2 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11] +; CHECK-NEXT: vpaddw %ymm3, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6d,0x28,0xfd,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <16 x i16> @llvm.x86.avx512.mask.punpcklw.d.256(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 %x3) + %res1 = call <16 x i16> @llvm.x86.avx512.mask.punpcklw.d.256(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 -1) + %res2 = add <16 x i16> %res, %res1 + ret <16 x i16> %res2 +} + +declare <16 x i16> @llvm.x86.avx512.mask.punpckhw.d.256(<16 x i16>, <16 x i16>, <16 x i16>, i16) + +define <16 x i16>@test_int_x86_avx512_mask_punpckhw_d_256(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpckhw_d_256: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpckhwd %ymm1, %ymm0, %ymm3 ## encoding: [0x62,0xf1,0x7d,0x28,0x69,0xd9] +; CHECK-NEXT: ## ymm3 = ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vpunpckhwd %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x29,0x69,0xd1] +; CHECK-NEXT: ## ymm2 = ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15] +; CHECK-NEXT: vpaddw %ymm3, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6d,0x28,0xfd,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <16 x i16> @llvm.x86.avx512.mask.punpckhw.d.256(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 %x3) + %res1 = call <16 x i16> @llvm.x86.avx512.mask.punpckhw.d.256(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 -1) + %res2 = add <16 x i16> %res, %res1 + ret <16 x i16> %res2 +} + diff --git a/test/CodeGen/X86/avx512bwvl-intrinsics.ll b/test/CodeGen/X86/avx512bwvl-intrinsics.ll index 6017a112e06..6ecaf1d80e4 100644 --- a/test/CodeGen/X86/avx512bwvl-intrinsics.ll +++ b/test/CodeGen/X86/avx512bwvl-intrinsics.ll @@ -5172,150 +5172,6 @@ define <16 x i16>@test_int_x86_avx512_mask_pmaddubs_w_256(<32 x i8> %x0, <32 x i ret <16 x i16> %res2 } -declare <16 x i8> @llvm.x86.avx512.mask.punpckhb.w.128(<16 x i8>, <16 x i8>, <16 x i8>, i16) - -define <16 x i8>@test_int_x86_avx512_mask_punpckhb_w_128(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %x2, i16 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpckhb_w_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vpunpckhbw %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x09,0x68,0xd1] -; CHECK-NEXT: ## xmm2 = xmm0[8],xmm1[8],xmm0[9],xmm1[9],xmm0[10],xmm1[10],xmm0[11],xmm1[11],xmm0[12],xmm1[12],xmm0[13],xmm1[13],xmm0[14],xmm1[14],xmm0[15],xmm1[15] -; CHECK-NEXT: vpunpckhbw %xmm1, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0x7d,0x08,0x68,0xc1] -; CHECK-NEXT: ## xmm0 = xmm0[8],xmm1[8],xmm0[9],xmm1[9],xmm0[10],xmm1[10],xmm0[11],xmm1[11],xmm0[12],xmm1[12],xmm0[13],xmm1[13],xmm0[14],xmm1[14],xmm0[15],xmm1[15] -; CHECK-NEXT: vpaddb %xmm0, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6d,0x08,0xfc,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <16 x i8> @llvm.x86.avx512.mask.punpckhb.w.128(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %x2, i16 %x3) - %res1 = call <16 x i8> @llvm.x86.avx512.mask.punpckhb.w.128(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %x2, i16 -1) - %res2 = add <16 x i8> %res, %res1 - ret <16 x i8> %res2 -} - -declare <16 x i8> @llvm.x86.avx512.mask.punpcklb.w.128(<16 x i8>, <16 x i8>, <16 x i8>, i16) - -define <16 x i8>@test_int_x86_avx512_mask_punpcklb_w_128(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %x2, i16 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpcklb_w_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vpunpcklbw %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x09,0x60,0xd1] -; CHECK-NEXT: ## xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7] -; CHECK-NEXT: vpunpcklbw %xmm1, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0x7d,0x08,0x60,0xc1] -; CHECK-NEXT: ## xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7] -; CHECK-NEXT: vpaddb %xmm0, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6d,0x08,0xfc,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <16 x i8> @llvm.x86.avx512.mask.punpcklb.w.128(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %x2, i16 %x3) - %res1 = call <16 x i8> @llvm.x86.avx512.mask.punpcklb.w.128(<16 x i8> %x0, <16 x i8> %x1, <16 x i8> %x2, i16 -1) - %res2 = add <16 x i8> %res, %res1 - ret <16 x i8> %res2 -} - -declare <32 x i8> @llvm.x86.avx512.mask.punpckhb.w.256(<32 x i8>, <32 x i8>, <32 x i8>, i32) - -define <32 x i8>@test_int_x86_avx512_mask_punpckhb_w_256(<32 x i8> %x0, <32 x i8> %x1, <32 x i8> %x2, i32 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpckhb_w_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovd %edi, %k1 ## encoding: [0xc5,0xfb,0x92,0xcf] -; CHECK-NEXT: vpunpckhbw %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x29,0x68,0xd1] -; CHECK-NEXT: ## ymm2 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31] -; CHECK-NEXT: vpunpckhbw %ymm1, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0x7d,0x28,0x68,0xc1] -; CHECK-NEXT: ## ymm0 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31] -; CHECK-NEXT: vpaddb %ymm0, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6d,0x28,0xfc,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <32 x i8> @llvm.x86.avx512.mask.punpckhb.w.256(<32 x i8> %x0, <32 x i8> %x1, <32 x i8> %x2, i32 %x3) - %res1 = call <32 x i8> @llvm.x86.avx512.mask.punpckhb.w.256(<32 x i8> %x0, <32 x i8> %x1, <32 x i8> %x2, i32 -1) - %res2 = add <32 x i8> %res, %res1 - ret <32 x i8> %res2 -} - -declare <32 x i8> @llvm.x86.avx512.mask.punpcklb.w.256(<32 x i8>, <32 x i8>, <32 x i8>, i32) - -define <32 x i8>@test_int_x86_avx512_mask_punpcklb_w_256(<32 x i8> %x0, <32 x i8> %x1, <32 x i8> %x2, i32 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpcklb_w_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovd %edi, %k1 ## encoding: [0xc5,0xfb,0x92,0xcf] -; CHECK-NEXT: vpunpcklbw %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x29,0x60,0xd1] -; CHECK-NEXT: ## ymm2 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23] -; CHECK-NEXT: vpunpcklbw %ymm1, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0x7d,0x28,0x60,0xc1] -; CHECK-NEXT: ## ymm0 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23] -; CHECK-NEXT: vpaddb %ymm0, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6d,0x28,0xfc,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <32 x i8> @llvm.x86.avx512.mask.punpcklb.w.256(<32 x i8> %x0, <32 x i8> %x1, <32 x i8> %x2, i32 %x3) - %res1 = call <32 x i8> @llvm.x86.avx512.mask.punpcklb.w.256(<32 x i8> %x0, <32 x i8> %x1, <32 x i8> %x2, i32 -1) - %res2 = add <32 x i8> %res, %res1 - ret <32 x i8> %res2 -} - -declare <8 x i16> @llvm.x86.avx512.mask.punpcklw.d.128(<8 x i16>, <8 x i16>, <8 x i16>, i8) - -define <8 x i16>@test_int_x86_avx512_mask_punpcklw_d_128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpcklw_d_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vpunpcklwd %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x09,0x61,0xd1] -; CHECK-NEXT: ## xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; CHECK-NEXT: vpunpcklwd %xmm1, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0x7d,0x08,0x61,0xc1] -; CHECK-NEXT: ## xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; CHECK-NEXT: vpaddw %xmm0, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6d,0x08,0xfd,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <8 x i16> @llvm.x86.avx512.mask.punpcklw.d.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3) - %res1 = call <8 x i16> @llvm.x86.avx512.mask.punpcklw.d.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 -1) - %res2 = add <8 x i16> %res, %res1 - ret <8 x i16> %res2 -} - -declare <8 x i16> @llvm.x86.avx512.mask.punpckhw.d.128(<8 x i16>, <8 x i16>, <8 x i16>, i8) - -define <8 x i16>@test_int_x86_avx512_mask_punpckhw_d_128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpckhw_d_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vpunpckhwd %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x09,0x69,0xd1] -; CHECK-NEXT: ## xmm2 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7] -; CHECK-NEXT: vpunpckhwd %xmm1, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0x7d,0x08,0x69,0xc1] -; CHECK-NEXT: ## xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7] -; CHECK-NEXT: vpaddw %xmm0, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6d,0x08,0xfd,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <8 x i16> @llvm.x86.avx512.mask.punpckhw.d.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3) - %res1 = call <8 x i16> @llvm.x86.avx512.mask.punpckhw.d.128(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 -1) - %res2 = add <8 x i16> %res, %res1 - ret <8 x i16> %res2 -} - -declare <16 x i16> @llvm.x86.avx512.mask.punpcklw.d.256(<16 x i16>, <16 x i16>, <16 x i16>, i16) - -define <16 x i16>@test_int_x86_avx512_mask_punpcklw_d_256(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpcklw_d_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vpunpcklwd %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x29,0x61,0xd1] -; CHECK-NEXT: ## ymm2 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11] -; CHECK-NEXT: vpunpcklwd %ymm1, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0x7d,0x28,0x61,0xc1] -; CHECK-NEXT: ## ymm0 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11] -; CHECK-NEXT: vpaddw %ymm0, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6d,0x28,0xfd,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <16 x i16> @llvm.x86.avx512.mask.punpcklw.d.256(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 %x3) - %res1 = call <16 x i16> @llvm.x86.avx512.mask.punpcklw.d.256(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 -1) - %res2 = add <16 x i16> %res, %res1 - ret <16 x i16> %res2 -} - -declare <16 x i16> @llvm.x86.avx512.mask.punpckhw.d.256(<16 x i16>, <16 x i16>, <16 x i16>, i16) - -define <16 x i16>@test_int_x86_avx512_mask_punpckhw_d_256(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpckhw_d_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vpunpckhwd %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x29,0x69,0xd1] -; CHECK-NEXT: ## ymm2 = ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15] -; CHECK-NEXT: vpunpckhwd %ymm1, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0x7d,0x28,0x69,0xc1] -; CHECK-NEXT: ## ymm0 = ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15] -; CHECK-NEXT: vpaddw %ymm0, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6d,0x28,0xfd,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <16 x i16> @llvm.x86.avx512.mask.punpckhw.d.256(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 %x3) - %res1 = call <16 x i16> @llvm.x86.avx512.mask.punpckhw.d.256(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 -1) - %res2 = add <16 x i16> %res, %res1 - ret <16 x i16> %res2 -} - declare <8 x i16> @llvm.x86.avx512.mask.dbpsadbw.128(<16 x i8>, <16 x i8>, i32, <8 x i16>, i8) define <8 x i16>@test_int_x86_avx512_mask_dbpsadbw_128(<16 x i8> %x0, <16 x i8> %x1, <8 x i16> %x3, i8 %x4) { diff --git a/test/CodeGen/X86/avx512vl-intrinsics-upgrade.ll b/test/CodeGen/X86/avx512vl-intrinsics-upgrade.ll index 3089e598893..4a0235e09ad 100644 --- a/test/CodeGen/X86/avx512vl-intrinsics-upgrade.ll +++ b/test/CodeGen/X86/avx512vl-intrinsics-upgrade.ll @@ -742,3 +742,291 @@ define i8 @test_mask_pcmpgt_q_128(<2 x i64> %a, <2 x i64> %b, i8 %mask) { } declare i8 @llvm.x86.avx512.mask.pcmpgt.q.128(<2 x i64>, <2 x i64>, i8) + +declare <2 x double> @llvm.x86.avx512.mask.unpckh.pd.128(<2 x double>, <2 x double>, <2 x double>, i8) + +define <2 x double>@test_int_x86_avx512_mask_unpckh_pd_128(<2 x double> %x0, <2 x double> %x1, <2 x double> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_unpckh_pd_128: +; CHECK: ## BB#0: +; CHECK-NEXT: vunpckhpd %xmm1, %xmm0, %xmm3 ## encoding: [0x62,0xf1,0xfd,0x08,0x15,0xd9] +; CHECK-NEXT: ## xmm3 = xmm0[1],xmm1[1] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vunpckhpd %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x09,0x15,0xd1] +; CHECK-NEXT: ## xmm2 = xmm0[1],xmm1[1] +; CHECK-NEXT: vaddpd %xmm3, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0xed,0x08,0x58,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <2 x double> @llvm.x86.avx512.mask.unpckh.pd.128(<2 x double> %x0, <2 x double> %x1, <2 x double> %x2, i8 %x3) + %res1 = call <2 x double> @llvm.x86.avx512.mask.unpckh.pd.128(<2 x double> %x0, <2 x double> %x1, <2 x double> %x2, i8 -1) + %res2 = fadd <2 x double> %res, %res1 + ret <2 x double> %res2 +} + +declare <4 x double> @llvm.x86.avx512.mask.unpckh.pd.256(<4 x double>, <4 x double>, <4 x double>, i8) + +define <4 x double>@test_int_x86_avx512_mask_unpckh_pd_256(<4 x double> %x0, <4 x double> %x1, <4 x double> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_unpckh_pd_256: +; CHECK: ## BB#0: +; CHECK-NEXT: vunpckhpd %ymm1, %ymm0, %ymm3 ## encoding: [0x62,0xf1,0xfd,0x28,0x15,0xd9] +; CHECK-NEXT: ## ymm3 = ymm0[1],ymm1[1],ymm0[3],ymm1[3] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vunpckhpd %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x29,0x15,0xd1] +; CHECK-NEXT: ## ymm2 = ymm0[1],ymm1[1],ymm0[3],ymm1[3] +; CHECK-NEXT: vaddpd %ymm3, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0xed,0x28,0x58,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <4 x double> @llvm.x86.avx512.mask.unpckh.pd.256(<4 x double> %x0, <4 x double> %x1, <4 x double> %x2, i8 %x3) + %res1 = call <4 x double> @llvm.x86.avx512.mask.unpckh.pd.256(<4 x double> %x0, <4 x double> %x1, <4 x double> %x2, i8 -1) + %res2 = fadd <4 x double> %res, %res1 + ret <4 x double> %res2 +} + +declare <4 x float> @llvm.x86.avx512.mask.unpckh.ps.128(<4 x float>, <4 x float>, <4 x float>, i8) + +define <4 x float>@test_int_x86_avx512_mask_unpckh_ps_128(<4 x float> %x0, <4 x float> %x1, <4 x float> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_unpckh_ps_128: +; CHECK: ## BB#0: +; CHECK-NEXT: vunpckhps %xmm1, %xmm0, %xmm3 ## encoding: [0x62,0xf1,0x7c,0x08,0x15,0xd9] +; CHECK-NEXT: ## xmm3 = xmm0[2],xmm1[2],xmm0[3],xmm1[3] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vunpckhps %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7c,0x09,0x15,0xd1] +; CHECK-NEXT: ## xmm2 = xmm0[2],xmm1[2],xmm0[3],xmm1[3] +; CHECK-NEXT: vaddps %xmm3, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6c,0x08,0x58,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <4 x float> @llvm.x86.avx512.mask.unpckh.ps.128(<4 x float> %x0, <4 x float> %x1, <4 x float> %x2, i8 %x3) + %res1 = call <4 x float> @llvm.x86.avx512.mask.unpckh.ps.128(<4 x float> %x0, <4 x float> %x1, <4 x float> %x2, i8 -1) + %res2 = fadd <4 x float> %res, %res1 + ret <4 x float> %res2 +} + +declare <8 x float> @llvm.x86.avx512.mask.unpckh.ps.256(<8 x float>, <8 x float>, <8 x float>, i8) + +define <8 x float>@test_int_x86_avx512_mask_unpckh_ps_256(<8 x float> %x0, <8 x float> %x1, <8 x float> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_unpckh_ps_256: +; CHECK: ## BB#0: +; CHECK-NEXT: vunpckhps %ymm1, %ymm0, %ymm3 ## encoding: [0x62,0xf1,0x7c,0x28,0x15,0xd9] +; CHECK-NEXT: ## ymm3 = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vunpckhps %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7c,0x29,0x15,0xd1] +; CHECK-NEXT: ## ymm2 = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7] +; CHECK-NEXT: vaddps %ymm3, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6c,0x28,0x58,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <8 x float> @llvm.x86.avx512.mask.unpckh.ps.256(<8 x float> %x0, <8 x float> %x1, <8 x float> %x2, i8 %x3) + %res1 = call <8 x float> @llvm.x86.avx512.mask.unpckh.ps.256(<8 x float> %x0, <8 x float> %x1, <8 x float> %x2, i8 -1) + %res2 = fadd <8 x float> %res, %res1 + ret <8 x float> %res2 +} + +declare <2 x double> @llvm.x86.avx512.mask.unpckl.pd.128(<2 x double>, <2 x double>, <2 x double>, i8) + +define <2 x double>@test_int_x86_avx512_mask_unpckl_pd_128(<2 x double> %x0, <2 x double> %x1, <2 x double> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_unpckl_pd_128: +; CHECK: ## BB#0: +; CHECK-NEXT: vunpcklpd %xmm1, %xmm0, %xmm3 ## encoding: [0x62,0xf1,0xfd,0x08,0x14,0xd9] +; CHECK-NEXT: ## xmm3 = xmm0[0],xmm1[0] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vunpcklpd %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x09,0x14,0xd1] +; CHECK-NEXT: ## xmm2 = xmm0[0],xmm1[0] +; CHECK-NEXT: vaddpd %xmm3, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0xed,0x08,0x58,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <2 x double> @llvm.x86.avx512.mask.unpckl.pd.128(<2 x double> %x0, <2 x double> %x1, <2 x double> %x2, i8 %x3) + %res1 = call <2 x double> @llvm.x86.avx512.mask.unpckl.pd.128(<2 x double> %x0, <2 x double> %x1, <2 x double> %x2, i8 -1) + %res2 = fadd <2 x double> %res, %res1 + ret <2 x double> %res2 +} + +declare <4 x double> @llvm.x86.avx512.mask.unpckl.pd.256(<4 x double>, <4 x double>, <4 x double>, i8) + +define <4 x double>@test_int_x86_avx512_mask_unpckl_pd_256(<4 x double> %x0, <4 x double> %x1, <4 x double> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_unpckl_pd_256: +; CHECK: ## BB#0: +; CHECK-NEXT: vunpcklpd %ymm1, %ymm0, %ymm3 ## encoding: [0x62,0xf1,0xfd,0x28,0x14,0xd9] +; CHECK-NEXT: ## ymm3 = ymm0[0],ymm1[0],ymm0[2],ymm1[2] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vunpcklpd %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x29,0x14,0xd1] +; CHECK-NEXT: ## ymm2 = ymm0[0],ymm1[0],ymm0[2],ymm1[2] +; CHECK-NEXT: vaddpd %ymm3, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0xed,0x28,0x58,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <4 x double> @llvm.x86.avx512.mask.unpckl.pd.256(<4 x double> %x0, <4 x double> %x1, <4 x double> %x2, i8 %x3) + %res1 = call <4 x double> @llvm.x86.avx512.mask.unpckl.pd.256(<4 x double> %x0, <4 x double> %x1, <4 x double> %x2, i8 -1) + %res2 = fadd <4 x double> %res, %res1 + ret <4 x double> %res2 +} + +declare <4 x float> @llvm.x86.avx512.mask.unpckl.ps.128(<4 x float>, <4 x float>, <4 x float>, i8) + +define <4 x float>@test_int_x86_avx512_mask_unpckl_ps_128(<4 x float> %x0, <4 x float> %x1, <4 x float> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_unpckl_ps_128: +; CHECK: ## BB#0: +; CHECK-NEXT: vunpcklps %xmm1, %xmm0, %xmm3 ## encoding: [0x62,0xf1,0x7c,0x08,0x14,0xd9] +; CHECK-NEXT: ## xmm3 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vunpcklps %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7c,0x09,0x14,0xd1] +; CHECK-NEXT: ## xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] +; CHECK-NEXT: vaddps %xmm3, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6c,0x08,0x58,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <4 x float> @llvm.x86.avx512.mask.unpckl.ps.128(<4 x float> %x0, <4 x float> %x1, <4 x float> %x2, i8 %x3) + %res1 = call <4 x float> @llvm.x86.avx512.mask.unpckl.ps.128(<4 x float> %x0, <4 x float> %x1, <4 x float> %x2, i8 -1) + %res2 = fadd <4 x float> %res, %res1 + ret <4 x float> %res2 +} + +declare <8 x float> @llvm.x86.avx512.mask.unpckl.ps.256(<8 x float>, <8 x float>, <8 x float>, i8) + +define <8 x float>@test_int_x86_avx512_mask_unpckl_ps_256(<8 x float> %x0, <8 x float> %x1, <8 x float> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_unpckl_ps_256: +; CHECK: ## BB#0: +; CHECK-NEXT: vunpcklps %ymm1, %ymm0, %ymm3 ## encoding: [0x62,0xf1,0x7c,0x28,0x14,0xd9] +; CHECK-NEXT: ## ymm3 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vunpcklps %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7c,0x29,0x14,0xd1] +; CHECK-NEXT: ## ymm2 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5] +; CHECK-NEXT: vaddps %ymm3, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6c,0x28,0x58,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <8 x float> @llvm.x86.avx512.mask.unpckl.ps.256(<8 x float> %x0, <8 x float> %x1, <8 x float> %x2, i8 %x3) + %res1 = call <8 x float> @llvm.x86.avx512.mask.unpckl.ps.256(<8 x float> %x0, <8 x float> %x1, <8 x float> %x2, i8 -1) + %res2 = fadd <8 x float> %res, %res1 + ret <8 x float> %res2 +} + +declare <4 x i32> @llvm.x86.avx512.mask.punpckhd.q.128(<4 x i32>, <4 x i32>, <4 x i32>, i8) + +define <4 x i32>@test_int_x86_avx512_mask_punpckhd_q_128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpckhd_q_128: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpckhdq %xmm1, %xmm0, %xmm3 ## encoding: [0x62,0xf1,0x7d,0x08,0x6a,0xd9] +; CHECK-NEXT: ## xmm3 = xmm0[2],xmm1[2],xmm0[3],xmm1[3] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vpunpckhdq %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x09,0x6a,0xd1] +; CHECK-NEXT: ## xmm2 = xmm0[2],xmm1[2],xmm0[3],xmm1[3] +; CHECK-NEXT: vpaddd %xmm3, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6d,0x08,0xfe,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <4 x i32> @llvm.x86.avx512.mask.punpckhd.q.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 %x3) + %res1 = call <4 x i32> @llvm.x86.avx512.mask.punpckhd.q.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 -1) + %res2 = add <4 x i32> %res, %res1 + ret <4 x i32> %res2 +} + +declare <4 x i32> @llvm.x86.avx512.mask.punpckld.q.128(<4 x i32>, <4 x i32>, <4 x i32>, i8) + +define <4 x i32>@test_int_x86_avx512_mask_punpckld_q_128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpckld_q_128: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpckldq %xmm1, %xmm0, %xmm3 ## encoding: [0x62,0xf1,0x7d,0x08,0x62,0xd9] +; CHECK-NEXT: ## xmm3 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vpunpckldq %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x09,0x62,0xd1] +; CHECK-NEXT: ## xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] +; CHECK-NEXT: vpaddd %xmm3, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6d,0x08,0xfe,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <4 x i32> @llvm.x86.avx512.mask.punpckld.q.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 %x3) + %res1 = call <4 x i32> @llvm.x86.avx512.mask.punpckld.q.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 -1) + %res2 = add <4 x i32> %res, %res1 + ret <4 x i32> %res2 +} + +declare <8 x i32> @llvm.x86.avx512.mask.punpckhd.q.256(<8 x i32>, <8 x i32>, <8 x i32>, i8) + +define <8 x i32>@test_int_x86_avx512_mask_punpckhd_q_256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpckhd_q_256: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpckhdq %ymm1, %ymm0, %ymm3 ## encoding: [0x62,0xf1,0x7d,0x28,0x6a,0xd9] +; CHECK-NEXT: ## ymm3 = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vpunpckhdq %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x29,0x6a,0xd1] +; CHECK-NEXT: ## ymm2 = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7] +; CHECK-NEXT: vpaddd %ymm3, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6d,0x28,0xfe,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <8 x i32> @llvm.x86.avx512.mask.punpckhd.q.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2, i8 %x3) + %res1 = call <8 x i32> @llvm.x86.avx512.mask.punpckhd.q.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2, i8 -1) + %res2 = add <8 x i32> %res, %res1 + ret <8 x i32> %res2 +} + +declare <8 x i32> @llvm.x86.avx512.mask.punpckld.q.256(<8 x i32>, <8 x i32>, <8 x i32>, i8) + +define <8 x i32>@test_int_x86_avx512_mask_punpckld_q_256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpckld_q_256: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpckldq %ymm1, %ymm0, %ymm3 ## encoding: [0x62,0xf1,0x7d,0x28,0x62,0xd9] +; CHECK-NEXT: ## ymm3 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vpunpckldq %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x29,0x62,0xd1] +; CHECK-NEXT: ## ymm2 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5] +; CHECK-NEXT: vpaddd %ymm3, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6d,0x28,0xfe,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <8 x i32> @llvm.x86.avx512.mask.punpckld.q.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2, i8 %x3) + %res1 = call <8 x i32> @llvm.x86.avx512.mask.punpckld.q.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2, i8 -1) + %res2 = add <8 x i32> %res, %res1 + ret <8 x i32> %res2 +} + +declare <2 x i64> @llvm.x86.avx512.mask.punpckhqd.q.128(<2 x i64>, <2 x i64>, <2 x i64>, i8) + +define <2 x i64>@test_int_x86_avx512_mask_punpckhqd_q_128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpckhqd_q_128: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpckhqdq %xmm1, %xmm0, %xmm3 ## encoding: [0x62,0xf1,0xfd,0x08,0x6d,0xd9] +; CHECK-NEXT: ## xmm3 = xmm0[1],xmm1[1] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vpunpckhqdq %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x09,0x6d,0xd1] +; CHECK-NEXT: ## xmm2 = xmm0[1],xmm1[1] +; CHECK-NEXT: vpaddq %xmm3, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0xed,0x08,0xd4,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <2 x i64> @llvm.x86.avx512.mask.punpckhqd.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3) + %res1 = call <2 x i64> @llvm.x86.avx512.mask.punpckhqd.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 -1) + %res2 = add <2 x i64> %res, %res1 + ret <2 x i64> %res2 +} + +declare <2 x i64> @llvm.x86.avx512.mask.punpcklqd.q.128(<2 x i64>, <2 x i64>, <2 x i64>, i8) + +define <2 x i64>@test_int_x86_avx512_mask_punpcklqd_q_128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpcklqd_q_128: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpcklqdq %xmm1, %xmm0, %xmm3 ## encoding: [0x62,0xf1,0xfd,0x08,0x6c,0xd9] +; CHECK-NEXT: ## xmm3 = xmm0[0],xmm1[0] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vpunpcklqdq %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x09,0x6c,0xd1] +; CHECK-NEXT: ## xmm2 = xmm0[0],xmm1[0] +; CHECK-NEXT: vpaddq %xmm3, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0xed,0x08,0xd4,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <2 x i64> @llvm.x86.avx512.mask.punpcklqd.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3) + %res1 = call <2 x i64> @llvm.x86.avx512.mask.punpcklqd.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 -1) + %res2 = add <2 x i64> %res, %res1 + ret <2 x i64> %res2 +} + +declare <4 x i64> @llvm.x86.avx512.mask.punpcklqd.q.256(<4 x i64>, <4 x i64>, <4 x i64>, i8) + +define <4 x i64>@test_int_x86_avx512_mask_punpcklqd_q_256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpcklqd_q_256: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpcklqdq %ymm1, %ymm0, %ymm3 ## encoding: [0x62,0xf1,0xfd,0x28,0x6c,0xd9] +; CHECK-NEXT: ## ymm3 = ymm0[0],ymm1[0],ymm0[2],ymm1[2] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vpunpcklqdq %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x29,0x6c,0xd1] +; CHECK-NEXT: ## ymm2 = ymm0[0],ymm1[0],ymm0[2],ymm1[2] +; CHECK-NEXT: vpaddq %ymm3, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0xed,0x28,0xd4,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <4 x i64> @llvm.x86.avx512.mask.punpcklqd.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3) + %res1 = call <4 x i64> @llvm.x86.avx512.mask.punpcklqd.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 -1) + %res2 = add <4 x i64> %res, %res1 + ret <4 x i64> %res2 +} + +declare <4 x i64> @llvm.x86.avx512.mask.punpckhqd.q.256(<4 x i64>, <4 x i64>, <4 x i64>, i8) + +define <4 x i64>@test_int_x86_avx512_mask_punpckhqd_q_256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_punpckhqd_q_256: +; CHECK: ## BB#0: +; CHECK-NEXT: vpunpckhqdq %ymm1, %ymm0, %ymm3 ## encoding: [0x62,0xf1,0xfd,0x28,0x6d,0xd9] +; CHECK-NEXT: ## ymm3 = ymm0[1],ymm1[1],ymm0[3],ymm1[3] +; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] +; CHECK-NEXT: vpunpckhqdq %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x29,0x6d,0xd1] +; CHECK-NEXT: ## ymm2 = ymm0[1],ymm1[1],ymm0[3],ymm1[3] +; CHECK-NEXT: vpaddq %ymm3, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0xed,0x28,0xd4,0xc3] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <4 x i64> @llvm.x86.avx512.mask.punpckhqd.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3) + %res1 = call <4 x i64> @llvm.x86.avx512.mask.punpckhqd.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 -1) + %res2 = add <4 x i64> %res, %res1 + ret <4 x i64> %res2 +} diff --git a/test/CodeGen/X86/avx512vl-intrinsics.ll b/test/CodeGen/X86/avx512vl-intrinsics.ll index 4103782e397..b2397de135d 100644 --- a/test/CodeGen/X86/avx512vl-intrinsics.ll +++ b/test/CodeGen/X86/avx512vl-intrinsics.ll @@ -3876,294 +3876,6 @@ define <8 x float>@test_int_x86_avx512_mask_scalef_ps_256(<8 x float> %x0, <8 x ret <8 x float> %res2 } -declare <2 x double> @llvm.x86.avx512.mask.unpckh.pd.128(<2 x double>, <2 x double>, <2 x double>, i8) - -define <2 x double>@test_int_x86_avx512_mask_unpckh_pd_128(<2 x double> %x0, <2 x double> %x1, <2 x double> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_unpckh_pd_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vunpckhpd %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x09,0x15,0xd1] -; CHECK-NEXT: ## xmm2 = xmm0[1],xmm1[1] -; CHECK-NEXT: vunpckhpd %xmm1, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0xfd,0x08,0x15,0xc1] -; CHECK-NEXT: ## xmm0 = xmm0[1],xmm1[1] -; CHECK-NEXT: vaddpd %xmm0, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0xed,0x08,0x58,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <2 x double> @llvm.x86.avx512.mask.unpckh.pd.128(<2 x double> %x0, <2 x double> %x1, <2 x double> %x2, i8 %x3) - %res1 = call <2 x double> @llvm.x86.avx512.mask.unpckh.pd.128(<2 x double> %x0, <2 x double> %x1, <2 x double> %x2, i8 -1) - %res2 = fadd <2 x double> %res, %res1 - ret <2 x double> %res2 -} - -declare <4 x double> @llvm.x86.avx512.mask.unpckh.pd.256(<4 x double>, <4 x double>, <4 x double>, i8) - -define <4 x double>@test_int_x86_avx512_mask_unpckh_pd_256(<4 x double> %x0, <4 x double> %x1, <4 x double> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_unpckh_pd_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vunpckhpd %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x29,0x15,0xd1] -; CHECK-NEXT: ## ymm2 = ymm0[1],ymm1[1],ymm0[3],ymm1[3] -; CHECK-NEXT: vunpckhpd %ymm1, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0xfd,0x28,0x15,0xc1] -; CHECK-NEXT: ## ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3] -; CHECK-NEXT: vaddpd %ymm0, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0xed,0x28,0x58,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <4 x double> @llvm.x86.avx512.mask.unpckh.pd.256(<4 x double> %x0, <4 x double> %x1, <4 x double> %x2, i8 %x3) - %res1 = call <4 x double> @llvm.x86.avx512.mask.unpckh.pd.256(<4 x double> %x0, <4 x double> %x1, <4 x double> %x2, i8 -1) - %res2 = fadd <4 x double> %res, %res1 - ret <4 x double> %res2 -} - -declare <4 x float> @llvm.x86.avx512.mask.unpckh.ps.128(<4 x float>, <4 x float>, <4 x float>, i8) - -define <4 x float>@test_int_x86_avx512_mask_unpckh_ps_128(<4 x float> %x0, <4 x float> %x1, <4 x float> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_unpckh_ps_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vunpckhps %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7c,0x09,0x15,0xd1] -; CHECK-NEXT: ## xmm2 = xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; CHECK-NEXT: vunpckhps %xmm1, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0x7c,0x08,0x15,0xc1] -; CHECK-NEXT: ## xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; CHECK-NEXT: vaddps %xmm0, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6c,0x08,0x58,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <4 x float> @llvm.x86.avx512.mask.unpckh.ps.128(<4 x float> %x0, <4 x float> %x1, <4 x float> %x2, i8 %x3) - %res1 = call <4 x float> @llvm.x86.avx512.mask.unpckh.ps.128(<4 x float> %x0, <4 x float> %x1, <4 x float> %x2, i8 -1) - %res2 = fadd <4 x float> %res, %res1 - ret <4 x float> %res2 -} - -declare <8 x float> @llvm.x86.avx512.mask.unpckh.ps.256(<8 x float>, <8 x float>, <8 x float>, i8) - -define <8 x float>@test_int_x86_avx512_mask_unpckh_ps_256(<8 x float> %x0, <8 x float> %x1, <8 x float> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_unpckh_ps_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vunpckhps %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7c,0x29,0x15,0xd1] -; CHECK-NEXT: ## ymm2 = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7] -; CHECK-NEXT: vunpckhps %ymm1, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0x7c,0x28,0x15,0xc1] -; CHECK-NEXT: ## ymm0 = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7] -; CHECK-NEXT: vaddps %ymm0, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6c,0x28,0x58,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <8 x float> @llvm.x86.avx512.mask.unpckh.ps.256(<8 x float> %x0, <8 x float> %x1, <8 x float> %x2, i8 %x3) - %res1 = call <8 x float> @llvm.x86.avx512.mask.unpckh.ps.256(<8 x float> %x0, <8 x float> %x1, <8 x float> %x2, i8 -1) - %res2 = fadd <8 x float> %res, %res1 - ret <8 x float> %res2 -} - -declare <2 x double> @llvm.x86.avx512.mask.unpckl.pd.128(<2 x double>, <2 x double>, <2 x double>, i8) - -define <2 x double>@test_int_x86_avx512_mask_unpckl_pd_128(<2 x double> %x0, <2 x double> %x1, <2 x double> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_unpckl_pd_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vunpcklpd %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x09,0x14,0xd1] -; CHECK-NEXT: ## xmm2 = xmm0[0],xmm1[0] -; CHECK-NEXT: vunpcklpd %xmm1, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0xfd,0x08,0x14,0xc1] -; CHECK-NEXT: ## xmm0 = xmm0[0],xmm1[0] -; CHECK-NEXT: vaddpd %xmm0, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0xed,0x08,0x58,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <2 x double> @llvm.x86.avx512.mask.unpckl.pd.128(<2 x double> %x0, <2 x double> %x1, <2 x double> %x2, i8 %x3) - %res1 = call <2 x double> @llvm.x86.avx512.mask.unpckl.pd.128(<2 x double> %x0, <2 x double> %x1, <2 x double> %x2, i8 -1) - %res2 = fadd <2 x double> %res, %res1 - ret <2 x double> %res2 -} - -declare <4 x double> @llvm.x86.avx512.mask.unpckl.pd.256(<4 x double>, <4 x double>, <4 x double>, i8) - -define <4 x double>@test_int_x86_avx512_mask_unpckl_pd_256(<4 x double> %x0, <4 x double> %x1, <4 x double> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_unpckl_pd_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vunpcklpd %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x29,0x14,0xd1] -; CHECK-NEXT: ## ymm2 = ymm0[0],ymm1[0],ymm0[2],ymm1[2] -; CHECK-NEXT: vunpcklpd %ymm1, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0xfd,0x28,0x14,0xc1] -; CHECK-NEXT: ## ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2] -; CHECK-NEXT: vaddpd %ymm0, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0xed,0x28,0x58,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <4 x double> @llvm.x86.avx512.mask.unpckl.pd.256(<4 x double> %x0, <4 x double> %x1, <4 x double> %x2, i8 %x3) - %res1 = call <4 x double> @llvm.x86.avx512.mask.unpckl.pd.256(<4 x double> %x0, <4 x double> %x1, <4 x double> %x2, i8 -1) - %res2 = fadd <4 x double> %res, %res1 - ret <4 x double> %res2 -} - -declare <4 x float> @llvm.x86.avx512.mask.unpckl.ps.128(<4 x float>, <4 x float>, <4 x float>, i8) - -define <4 x float>@test_int_x86_avx512_mask_unpckl_ps_128(<4 x float> %x0, <4 x float> %x1, <4 x float> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_unpckl_ps_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vunpcklps %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7c,0x09,0x14,0xd1] -; CHECK-NEXT: ## xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; CHECK-NEXT: vunpcklps %xmm1, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0x7c,0x08,0x14,0xc1] -; CHECK-NEXT: ## xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; CHECK-NEXT: vaddps %xmm0, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6c,0x08,0x58,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <4 x float> @llvm.x86.avx512.mask.unpckl.ps.128(<4 x float> %x0, <4 x float> %x1, <4 x float> %x2, i8 %x3) - %res1 = call <4 x float> @llvm.x86.avx512.mask.unpckl.ps.128(<4 x float> %x0, <4 x float> %x1, <4 x float> %x2, i8 -1) - %res2 = fadd <4 x float> %res, %res1 - ret <4 x float> %res2 -} - -declare <8 x float> @llvm.x86.avx512.mask.unpckl.ps.256(<8 x float>, <8 x float>, <8 x float>, i8) - -define <8 x float>@test_int_x86_avx512_mask_unpckl_ps_256(<8 x float> %x0, <8 x float> %x1, <8 x float> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_unpckl_ps_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vunpcklps %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7c,0x29,0x14,0xd1] -; CHECK-NEXT: ## ymm2 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5] -; CHECK-NEXT: vunpcklps %ymm1, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0x7c,0x28,0x14,0xc1] -; CHECK-NEXT: ## ymm0 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5] -; CHECK-NEXT: vaddps %ymm0, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6c,0x28,0x58,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <8 x float> @llvm.x86.avx512.mask.unpckl.ps.256(<8 x float> %x0, <8 x float> %x1, <8 x float> %x2, i8 %x3) - %res1 = call <8 x float> @llvm.x86.avx512.mask.unpckl.ps.256(<8 x float> %x0, <8 x float> %x1, <8 x float> %x2, i8 -1) - %res2 = fadd <8 x float> %res, %res1 - ret <8 x float> %res2 -} - -declare <4 x i32> @llvm.x86.avx512.mask.punpckhd.q.128(<4 x i32>, <4 x i32>, <4 x i32>, i8) - -define <4 x i32>@test_int_x86_avx512_mask_punpckhd_q_128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpckhd_q_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vpunpckhdq %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x09,0x6a,0xd1] -; CHECK-NEXT: ## xmm2 = xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; CHECK-NEXT: vpunpckhdq %xmm1, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0x7d,0x08,0x6a,0xc1] -; CHECK-NEXT: ## xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3] -; CHECK-NEXT: vpaddd %xmm0, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6d,0x08,0xfe,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <4 x i32> @llvm.x86.avx512.mask.punpckhd.q.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 %x3) - %res1 = call <4 x i32> @llvm.x86.avx512.mask.punpckhd.q.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 -1) - %res2 = add <4 x i32> %res, %res1 - ret <4 x i32> %res2 -} - -declare <4 x i32> @llvm.x86.avx512.mask.punpckld.q.128(<4 x i32>, <4 x i32>, <4 x i32>, i8) - -define <4 x i32>@test_int_x86_avx512_mask_punpckld_q_128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpckld_q_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vpunpckldq %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x09,0x62,0xd1] -; CHECK-NEXT: ## xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; CHECK-NEXT: vpunpckldq %xmm1, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0x7d,0x08,0x62,0xc1] -; CHECK-NEXT: ## xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1] -; CHECK-NEXT: vpaddd %xmm0, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0x6d,0x08,0xfe,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <4 x i32> @llvm.x86.avx512.mask.punpckld.q.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 %x3) - %res1 = call <4 x i32> @llvm.x86.avx512.mask.punpckld.q.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 -1) - %res2 = add <4 x i32> %res, %res1 - ret <4 x i32> %res2 -} - -declare <8 x i32> @llvm.x86.avx512.mask.punpckhd.q.256(<8 x i32>, <8 x i32>, <8 x i32>, i8) - -define <8 x i32>@test_int_x86_avx512_mask_punpckhd_q_256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpckhd_q_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vpunpckhdq %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x29,0x6a,0xd1] -; CHECK-NEXT: ## ymm2 = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7] -; CHECK-NEXT: vpunpckhdq %ymm1, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0x7d,0x28,0x6a,0xc1] -; CHECK-NEXT: ## ymm0 = ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[6],ymm1[6],ymm0[7],ymm1[7] -; CHECK-NEXT: vpaddd %ymm0, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6d,0x28,0xfe,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <8 x i32> @llvm.x86.avx512.mask.punpckhd.q.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2, i8 %x3) - %res1 = call <8 x i32> @llvm.x86.avx512.mask.punpckhd.q.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2, i8 -1) - %res2 = add <8 x i32> %res, %res1 - ret <8 x i32> %res2 -} - -declare <8 x i32> @llvm.x86.avx512.mask.punpckld.q.256(<8 x i32>, <8 x i32>, <8 x i32>, i8) - -define <8 x i32>@test_int_x86_avx512_mask_punpckld_q_256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpckld_q_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vpunpckldq %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0x7d,0x29,0x62,0xd1] -; CHECK-NEXT: ## ymm2 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5] -; CHECK-NEXT: vpunpckldq %ymm1, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0x7d,0x28,0x62,0xc1] -; CHECK-NEXT: ## ymm0 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[4],ymm1[4],ymm0[5],ymm1[5] -; CHECK-NEXT: vpaddd %ymm0, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0x6d,0x28,0xfe,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <8 x i32> @llvm.x86.avx512.mask.punpckld.q.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2, i8 %x3) - %res1 = call <8 x i32> @llvm.x86.avx512.mask.punpckld.q.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2, i8 -1) - %res2 = add <8 x i32> %res, %res1 - ret <8 x i32> %res2 -} - -declare <2 x i64> @llvm.x86.avx512.mask.punpckhqd.q.128(<2 x i64>, <2 x i64>, <2 x i64>, i8) - -define <2 x i64>@test_int_x86_avx512_mask_punpckhqd_q_128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpckhqd_q_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vpunpckhqdq %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x09,0x6d,0xd1] -; CHECK-NEXT: ## xmm2 = xmm0[1],xmm1[1] -; CHECK-NEXT: vpunpckhqdq %xmm1, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0xfd,0x08,0x6d,0xc1] -; CHECK-NEXT: ## xmm0 = xmm0[1],xmm1[1] -; CHECK-NEXT: vpaddq %xmm0, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0xed,0x08,0xd4,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <2 x i64> @llvm.x86.avx512.mask.punpckhqd.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3) - %res1 = call <2 x i64> @llvm.x86.avx512.mask.punpckhqd.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 -1) - %res2 = add <2 x i64> %res, %res1 - ret <2 x i64> %res2 -} - -declare <2 x i64> @llvm.x86.avx512.mask.punpcklqd.q.128(<2 x i64>, <2 x i64>, <2 x i64>, i8) - -define <2 x i64>@test_int_x86_avx512_mask_punpcklqd_q_128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpcklqd_q_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vpunpcklqdq %xmm1, %xmm0, %xmm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x09,0x6c,0xd1] -; CHECK-NEXT: ## xmm2 = xmm0[0],xmm1[0] -; CHECK-NEXT: vpunpcklqdq %xmm1, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0xfd,0x08,0x6c,0xc1] -; CHECK-NEXT: ## xmm0 = xmm0[0],xmm1[0] -; CHECK-NEXT: vpaddq %xmm0, %xmm2, %xmm0 ## encoding: [0x62,0xf1,0xed,0x08,0xd4,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <2 x i64> @llvm.x86.avx512.mask.punpcklqd.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3) - %res1 = call <2 x i64> @llvm.x86.avx512.mask.punpcklqd.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 -1) - %res2 = add <2 x i64> %res, %res1 - ret <2 x i64> %res2 -} - -declare <4 x i64> @llvm.x86.avx512.mask.punpcklqd.q.256(<4 x i64>, <4 x i64>, <4 x i64>, i8) - -define <4 x i64>@test_int_x86_avx512_mask_punpcklqd_q_256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpcklqd_q_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vpunpcklqdq %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x29,0x6c,0xd1] -; CHECK-NEXT: ## ymm2 = ymm0[0],ymm1[0],ymm0[2],ymm1[2] -; CHECK-NEXT: vpunpcklqdq %ymm1, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0xfd,0x28,0x6c,0xc1] -; CHECK-NEXT: ## ymm0 = ymm0[0],ymm1[0],ymm0[2],ymm1[2] -; CHECK-NEXT: vpaddq %ymm0, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0xed,0x28,0xd4,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <4 x i64> @llvm.x86.avx512.mask.punpcklqd.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3) - %res1 = call <4 x i64> @llvm.x86.avx512.mask.punpcklqd.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 -1) - %res2 = add <4 x i64> %res, %res1 - ret <4 x i64> %res2 -} - -declare <4 x i64> @llvm.x86.avx512.mask.punpckhqd.q.256(<4 x i64>, <4 x i64>, <4 x i64>, i8) - -define <4 x i64>@test_int_x86_avx512_mask_punpckhqd_q_256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3) { -; CHECK-LABEL: test_int_x86_avx512_mask_punpckhqd_q_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k1 ## encoding: [0xc5,0xf8,0x92,0xcf] -; CHECK-NEXT: vpunpckhqdq %ymm1, %ymm0, %ymm2 {%k1} ## encoding: [0x62,0xf1,0xfd,0x29,0x6d,0xd1] -; CHECK-NEXT: ## ymm2 = ymm0[1],ymm1[1],ymm0[3],ymm1[3] -; CHECK-NEXT: vpunpckhqdq %ymm1, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0xfd,0x28,0x6d,0xc1] -; CHECK-NEXT: ## ymm0 = ymm0[1],ymm1[1],ymm0[3],ymm1[3] -; CHECK-NEXT: vpaddq %ymm0, %ymm2, %ymm0 ## encoding: [0x62,0xf1,0xed,0x28,0xd4,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <4 x i64> @llvm.x86.avx512.mask.punpckhqd.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3) - %res1 = call <4 x i64> @llvm.x86.avx512.mask.punpckhqd.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 -1) - %res2 = add <4 x i64> %res, %res1 - ret <4 x i64> %res2 -} - declare <16 x i8> @llvm.x86.avx512.mask.pmov.qb.128(<2 x i64>, <16 x i8>, i8) define <16 x i8>@test_int_x86_avx512_mask_pmov_qb_128(<2 x i64> %x0, <16 x i8> %x1, i8 %x2) { @@ -7644,9 +7356,9 @@ define <8 x i32>@test_int_x86_avx512_mask_psrav8_si_const() { ; CHECK: ## BB#0: ; CHECK-NEXT: vmovdqa32 {{.*#+}} ymm0 = [2,9,4294967284,23,4294967270,37,4294967256,51] ; CHECK-NEXT: ## encoding: [0x62,0xf1,0x7d,0x28,0x6f,0x05,A,A,A,A] -; CHECK-NEXT: ## fixup A - offset: 6, value: LCPI494_0-4, kind: reloc_riprel_4byte +; CHECK-NEXT: ## fixup A - offset: 6, value: LCPI478_0-4, kind: reloc_riprel_4byte ; CHECK-NEXT: vpsravd {{.*}}(%rip), %ymm0, %ymm0 ## encoding: [0x62,0xf2,0x7d,0x28,0x46,0x05,A,A,A,A] -; CHECK-NEXT: ## fixup A - offset: 6, value: LCPI494_1-4, kind: reloc_riprel_4byte +; CHECK-NEXT: ## fixup A - offset: 6, value: LCPI478_1-4, kind: reloc_riprel_4byte ; CHECK-NEXT: retq ## encoding: [0xc3] %res = call <8 x i32> @llvm.x86.avx512.mask.psrav8.si(<8 x i32> , <8 x i32> , <8 x i32> zeroinitializer, i8 -1) ret <8 x i32> %res @@ -7677,9 +7389,9 @@ define <2 x i64>@test_int_x86_avx512_mask_psrav_q_128_const(i8 %x3) { ; CHECK: ## BB#0: ; CHECK-NEXT: vmovdqa64 {{.*#+}} xmm0 = [2,18446744073709551607] ; CHECK-NEXT: ## encoding: [0x62,0xf1,0xfd,0x08,0x6f,0x05,A,A,A,A] -; CHECK-NEXT: ## fixup A - offset: 6, value: LCPI496_0-4, kind: reloc_riprel_4byte +; CHECK-NEXT: ## fixup A - offset: 6, value: LCPI480_0-4, kind: reloc_riprel_4byte ; CHECK-NEXT: vpsravq {{.*}}(%rip), %xmm0, %xmm0 ## encoding: [0x62,0xf2,0xfd,0x08,0x46,0x05,A,A,A,A] -; CHECK-NEXT: ## fixup A - offset: 6, value: LCPI496_1-4, kind: reloc_riprel_4byte +; CHECK-NEXT: ## fixup A - offset: 6, value: LCPI480_1-4, kind: reloc_riprel_4byte ; CHECK-NEXT: retq ## encoding: [0xc3] %res = call <2 x i64> @llvm.x86.avx512.mask.psrav.q.128(<2 x i64> , <2 x i64> , <2 x i64> zeroinitializer, i8 -1) ret <2 x i64> %res diff --git a/test/CodeGen/X86/vector-shuffle-512-v32.ll b/test/CodeGen/X86/vector-shuffle-512-v32.ll index a4455e76d5b..bcc4ad2d141 100644 --- a/test/CodeGen/X86/vector-shuffle-512-v32.ll +++ b/test/CodeGen/X86/vector-shuffle-512-v32.ll @@ -45,8 +45,7 @@ define <32 x i16> @shuffle_v32i16_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_1 define <32 x i16> @shuffle_v16i32_0_32_1_33_2_34_3_35_8_40_9_41_u_u_u_u(<32 x i16> %a, <32 x i16> %b) { ; ALL-LABEL: shuffle_v16i32_0_32_1_33_2_34_3_35_8_40_9_41_u_u_u_u: ; ALL: # BB#0: -; ALL-NEXT: vmovdqu16 {{.*#+}} zmm2 = <0,32,1,33,2,34,3,35,8,40,9,41,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u> -; ALL-NEXT: vpermt2w %zmm1, %zmm2, %zmm0 +; ALL-NEXT: vpunpcklwd {{.*#+}} zmm0 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27] ; ALL-NEXT: retq %c = shufflevector <32 x i16> %a, <32 x i16> %b, <32 x i32> ret <32 x i16> %c @@ -55,8 +54,7 @@ define <32 x i16> @shuffle_v16i32_0_32_1_33_2_34_3_35_8_40_9_41_u_u_u_u(<32 x i1 define <32 x i16> @shuffle_v16i32_4_36_5_37_6_38_7_39_12_44_13_45_u_u_u_u(<32 x i16> %a, <32 x i16> %b) { ; ALL-LABEL: shuffle_v16i32_4_36_5_37_6_38_7_39_12_44_13_45_u_u_u_u: ; ALL: # BB#0: -; ALL-NEXT: vmovdqu16 {{.*#+}} zmm2 = <4,36,5,37,6,38,7,39,12,44,13,45,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u> -; ALL-NEXT: vpermt2w %zmm1, %zmm2, %zmm0 +; ALL-NEXT: vpunpckhwd {{.*#+}} zmm0 = zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31] ; ALL-NEXT: retq %c = shufflevector <32 x i16> %a, <32 x i16> %b, <32 x i32> ret <32 x i16> %c