From: Michael Zuckerman Date: Tue, 4 Apr 2017 13:32:14 +0000 (+0000) Subject: [X86][LLVM] Converting __mm{|256|512}_movm_epi{8|16|32|64} LLVMIR call into generic... X-Git-Url: https://granicus.if.org/sourcecode?a=commitdiff_plain;h=3341db988bece5a1a7adc3c0480bcbc23d514d21;p=llvm [X86][LLVM] Converting __mm{|256|512}_movm_epi{8|16|32|64} LLVMIR call into generic intrinsics. This patch is a part one of two reviews, one for the clang and the other for LLVM. The patch deletes the back-end intrinsics and adds support for them in the auto upgrade. Differential Revision: https://reviews.llvm.org/D31393 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@299432 91177308-0d34-0410-b5e6-96231b3b80d8 --- diff --git a/include/llvm/IR/IntrinsicsX86.td b/include/llvm/IR/IntrinsicsX86.td index 019760cf97b..d3cce634479 100644 --- a/include/llvm/IR/IntrinsicsX86.td +++ b/include/llvm/IR/IntrinsicsX86.td @@ -3870,34 +3870,6 @@ let TargetPrefix = "x86" in { // All intrinsics start with "llvm.x86.". def int_x86_avx512_cvtq2mask_512 : GCCBuiltin<"__builtin_ia32_cvtq2mask512">, Intrinsic<[llvm_i8_ty], [llvm_v8i64_ty], [IntrNoMem]>; - def int_x86_avx512_cvtmask2b_128 : GCCBuiltin<"__builtin_ia32_cvtmask2b128">, - Intrinsic<[llvm_v16i8_ty], [llvm_i16_ty], [IntrNoMem]>; - def int_x86_avx512_cvtmask2b_256 : GCCBuiltin<"__builtin_ia32_cvtmask2b256">, - Intrinsic<[llvm_v32i8_ty], [llvm_i32_ty], [IntrNoMem]>; - def int_x86_avx512_cvtmask2b_512 : GCCBuiltin<"__builtin_ia32_cvtmask2b512">, - Intrinsic<[llvm_v64i8_ty], [llvm_i64_ty], [IntrNoMem]>; - - def int_x86_avx512_cvtmask2w_128 : GCCBuiltin<"__builtin_ia32_cvtmask2w128">, - Intrinsic<[llvm_v8i16_ty], [llvm_i8_ty], [IntrNoMem]>; - def int_x86_avx512_cvtmask2w_256 : GCCBuiltin<"__builtin_ia32_cvtmask2w256">, - Intrinsic<[llvm_v16i16_ty], [llvm_i16_ty], [IntrNoMem]>; - def int_x86_avx512_cvtmask2w_512 : GCCBuiltin<"__builtin_ia32_cvtmask2w512">, - Intrinsic<[llvm_v32i16_ty], [llvm_i32_ty], [IntrNoMem]>; - - def int_x86_avx512_cvtmask2d_128 : GCCBuiltin<"__builtin_ia32_cvtmask2d128">, - Intrinsic<[llvm_v4i32_ty], [llvm_i8_ty], [IntrNoMem]>; - def int_x86_avx512_cvtmask2d_256 : GCCBuiltin<"__builtin_ia32_cvtmask2d256">, - Intrinsic<[llvm_v8i32_ty], [llvm_i8_ty], [IntrNoMem]>; - def int_x86_avx512_cvtmask2d_512 : GCCBuiltin<"__builtin_ia32_cvtmask2d512">, - Intrinsic<[llvm_v16i32_ty], [llvm_i16_ty], [IntrNoMem]>; - - def int_x86_avx512_cvtmask2q_128 : GCCBuiltin<"__builtin_ia32_cvtmask2q128">, - Intrinsic<[llvm_v2i64_ty], [llvm_i8_ty], [IntrNoMem]>; - def int_x86_avx512_cvtmask2q_256 : GCCBuiltin<"__builtin_ia32_cvtmask2q256">, - Intrinsic<[llvm_v4i64_ty], [llvm_i8_ty], [IntrNoMem]>; - def int_x86_avx512_cvtmask2q_512 : GCCBuiltin<"__builtin_ia32_cvtmask2q512">, - Intrinsic<[llvm_v8i64_ty], [llvm_i8_ty], [IntrNoMem]>; - } // Pack ops. diff --git a/lib/IR/AutoUpgrade.cpp b/lib/IR/AutoUpgrade.cpp index 4922f724bcf..e73a49f7e10 100644 --- a/lib/IR/AutoUpgrade.cpp +++ b/lib/IR/AutoUpgrade.cpp @@ -234,6 +234,7 @@ static bool ShouldUpgradeX86Intrinsic(Function *F, StringRef Name) { Name == "xop.vpcmov" || // Added in 3.8 Name == "xop.vpcmov.256" || // Added in 5.0 Name.startswith("avx512.mask.move.s") || // Added in 4.0 + Name.startswith("avx512.cvtmask2") || // Added in 5.0 (Name.startswith("xop.vpcom") && // Added in 3.2 F->arg_size() == 2)) return true; @@ -800,6 +801,15 @@ static Value* upgradeMaskedMove(IRBuilder<> &Builder, CallInst &CI) { return Builder.CreateInsertElement(A, Select, (uint64_t)0); } + +static Value* UpgradeMaskToInt(IRBuilder<> &Builder, CallInst &CI) { + Value* Op = CI.getArgOperand(0); + Type* ReturnOp = CI.getType(); + unsigned NumElts = CI.getType()->getVectorNumElements(); + Value *Mask = getX86MaskVec(Builder, Op, NumElts); + return Builder.CreateSExt(Mask, ReturnOp, "vpmovm2"); +} + /// Upgrade a call to an old intrinsic. All argument and return casting must be /// provided to seamlessly integrate with existing context. void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) { @@ -1836,6 +1846,8 @@ void llvm::UpgradeIntrinsicCall(CallInst *CI, Function *NewFn) { Rep = UpgradeX86MaskedShift(Builder, *CI, IID); } else if (IsX86 && Name.startswith("avx512.mask.move.s")) { Rep = upgradeMaskedMove(Builder, *CI); + } else if (IsX86 && Name.startswith("avx512.cvtmask2")) { + Rep = UpgradeMaskToInt(Builder, *CI); } else if (IsX86 && Name.startswith("avx512.mask.vpermilvar.")) { Intrinsic::ID IID; if (Name.endswith("ps.128")) diff --git a/lib/Target/X86/X86IntrinsicsInfo.h b/lib/Target/X86/X86IntrinsicsInfo.h index 82c5ac5f1df..2a40399ba57 100644 --- a/lib/Target/X86/X86IntrinsicsInfo.h +++ b/lib/Target/X86/X86IntrinsicsInfo.h @@ -440,18 +440,6 @@ static const IntrinsicData IntrinsicsWithoutChain[] = { X86_INTRINSIC_DATA(avx512_cvtd2mask_128, CONVERT_TO_MASK, X86ISD::CVT2MASK, 0), X86_INTRINSIC_DATA(avx512_cvtd2mask_256, CONVERT_TO_MASK, X86ISD::CVT2MASK, 0), X86_INTRINSIC_DATA(avx512_cvtd2mask_512, CONVERT_TO_MASK, X86ISD::CVT2MASK, 0), - X86_INTRINSIC_DATA(avx512_cvtmask2b_128, CONVERT_MASK_TO_VEC, X86ISD::VSEXT, 0), - X86_INTRINSIC_DATA(avx512_cvtmask2b_256, CONVERT_MASK_TO_VEC, X86ISD::VSEXT, 0), - X86_INTRINSIC_DATA(avx512_cvtmask2b_512, CONVERT_MASK_TO_VEC, X86ISD::VSEXT, 0), - X86_INTRINSIC_DATA(avx512_cvtmask2d_128, CONVERT_MASK_TO_VEC, X86ISD::VSEXT, 0), - X86_INTRINSIC_DATA(avx512_cvtmask2d_256, CONVERT_MASK_TO_VEC, X86ISD::VSEXT, 0), - X86_INTRINSIC_DATA(avx512_cvtmask2d_512, CONVERT_MASK_TO_VEC, X86ISD::VSEXT, 0), - X86_INTRINSIC_DATA(avx512_cvtmask2q_128, CONVERT_MASK_TO_VEC, X86ISD::VSEXT, 0), - X86_INTRINSIC_DATA(avx512_cvtmask2q_256, CONVERT_MASK_TO_VEC, X86ISD::VSEXT, 0), - X86_INTRINSIC_DATA(avx512_cvtmask2q_512, CONVERT_MASK_TO_VEC, X86ISD::VSEXT, 0), - X86_INTRINSIC_DATA(avx512_cvtmask2w_128, CONVERT_MASK_TO_VEC, X86ISD::VSEXT, 0), - X86_INTRINSIC_DATA(avx512_cvtmask2w_256, CONVERT_MASK_TO_VEC, X86ISD::VSEXT, 0), - X86_INTRINSIC_DATA(avx512_cvtmask2w_512, CONVERT_MASK_TO_VEC, X86ISD::VSEXT, 0), X86_INTRINSIC_DATA(avx512_cvtq2mask_128, CONVERT_TO_MASK, X86ISD::CVT2MASK, 0), X86_INTRINSIC_DATA(avx512_cvtq2mask_256, CONVERT_TO_MASK, X86ISD::CVT2MASK, 0), X86_INTRINSIC_DATA(avx512_cvtq2mask_512, CONVERT_TO_MASK, X86ISD::CVT2MASK, 0), diff --git a/test/CodeGen/X86/avx512bw-intrinsics-upgrade.ll b/test/CodeGen/X86/avx512bw-intrinsics-upgrade.ll index 0cda27a8349..9b4e73a18fc 100644 --- a/test/CodeGen/X86/avx512bw-intrinsics-upgrade.ll +++ b/test/CodeGen/X86/avx512bw-intrinsics-upgrade.ll @@ -992,6 +992,42 @@ define <64 x i8>@test_int_x86_avx512_mask_pshuf_b_512(<64 x i8> %x0, <64 x i8> % ret <64 x i8> %res2 } + +declare <64 x i8> @llvm.x86.avx512.cvtmask2b.512(i64) + +define <64 x i8>@test_int_x86_avx512_cvtmask2b_512(i64 %x0) { +; AVX512BW-LABEL: test_int_x86_avx512_cvtmask2b_512: +; AVX512BW: ## BB#0: +; AVX512BW-NEXT: kmovq %rdi, %k0 +; AVX512BW-NEXT: vpmovm2b %k0, %zmm0 +; AVX512BW-NEXT: retq +; +; AVX512F-32-LABEL: test_int_x86_avx512_cvtmask2b_512: +; AVX512F-32: # BB#0: +; AVX512F-32-NEXT: kmovq {{[0-9]+}}(%esp), %k0 +; AVX512F-32-NEXT: vpmovm2b %k0, %zmm0 +; AVX512F-32-NEXT: retl + %res = call <64 x i8> @llvm.x86.avx512.cvtmask2b.512(i64 %x0) + ret <64 x i8> %res +} + +declare <32 x i16> @llvm.x86.avx512.cvtmask2w.512(i32) + +define <32 x i16>@test_int_x86_avx512_cvtmask2w_512(i32 %x0) { +; AVX512BW-LABEL: test_int_x86_avx512_cvtmask2w_512: +; AVX512BW: ## BB#0: +; AVX512BW-NEXT: kmovd %edi, %k0 +; AVX512BW-NEXT: vpmovm2w %k0, %zmm0 +; AVX512BW-NEXT: retq +; +; AVX512F-32-LABEL: test_int_x86_avx512_cvtmask2w_512: +; AVX512F-32: # BB#0: +; AVX512F-32-NEXT: kmovd {{[0-9]+}}(%esp), %k0 +; AVX512F-32-NEXT: vpmovm2w %k0, %zmm0 +; AVX512F-32-NEXT: retl + %res = call <32 x i16> @llvm.x86.avx512.cvtmask2w.512(i32 %x0) + ret <32 x i16> %res +} define <32 x i16> @test_mask_packs_epi32_rr_512(<16 x i32> %a, <16 x i32> %b) { ; AVX512BW-LABEL: test_mask_packs_epi32_rr_512: ; AVX512BW: ## BB#0: diff --git a/test/CodeGen/X86/avx512bw-intrinsics.ll b/test/CodeGen/X86/avx512bw-intrinsics.ll index d9b0fd74e71..3337f42eb14 100644 --- a/test/CodeGen/X86/avx512bw-intrinsics.ll +++ b/test/CodeGen/X86/avx512bw-intrinsics.ll @@ -2295,44 +2295,6 @@ define i32@test_int_x86_avx512_cvtw2mask_512(<32 x i16> %x0) { ret i32 %res } -declare <64 x i8> @llvm.x86.avx512.cvtmask2b.512(i64) - -define <64 x i8>@test_int_x86_avx512_cvtmask2b_512(i64 %x0) { -; AVX512BW-LABEL: test_int_x86_avx512_cvtmask2b_512: -; AVX512BW: ## BB#0: -; AVX512BW-NEXT: kmovq %rdi, %k0 -; AVX512BW-NEXT: vpmovm2b %k0, %zmm0 -; AVX512BW-NEXT: retq -; -; AVX512F-32-LABEL: test_int_x86_avx512_cvtmask2b_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, %k0 -; AVX512F-32-NEXT: vpmovm2b %k0, %zmm0 -; AVX512F-32-NEXT: retl - %res = call <64 x i8> @llvm.x86.avx512.cvtmask2b.512(i64 %x0) - ret <64 x i8> %res -} - -declare <32 x i16> @llvm.x86.avx512.cvtmask2w.512(i32) - -define <32 x i16>@test_int_x86_avx512_cvtmask2w_512(i32 %x0) { -; AVX512BW-LABEL: test_int_x86_avx512_cvtmask2w_512: -; AVX512BW: ## BB#0: -; AVX512BW-NEXT: kmovd %edi, %k0 -; AVX512BW-NEXT: vpmovm2w %k0, %zmm0 -; AVX512BW-NEXT: retq -; -; AVX512F-32-LABEL: test_int_x86_avx512_cvtmask2w_512: -; AVX512F-32: # BB#0: -; AVX512F-32-NEXT: kmovd {{[0-9]+}}(%esp), %k0 -; AVX512F-32-NEXT: vpmovm2w %k0, %zmm0 -; AVX512F-32-NEXT: retl - %res = call <32 x i16> @llvm.x86.avx512.cvtmask2w.512(i32 %x0) - ret <32 x i16> %res -} - declare <32 x i16> @llvm.x86.avx512.mask.psrlv32hi(<32 x i16>, <32 x i16>, <32 x i16>, i32) define <32 x i16>@test_int_x86_avx512_mask_psrlv32hi(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 %x3) { diff --git a/test/CodeGen/X86/avx512bwvl-intrinsics-upgrade.ll b/test/CodeGen/X86/avx512bwvl-intrinsics-upgrade.ll index 886621191cb..7df07b0413e 100644 --- a/test/CodeGen/X86/avx512bwvl-intrinsics-upgrade.ll +++ b/test/CodeGen/X86/avx512bwvl-intrinsics-upgrade.ll @@ -1867,6 +1867,54 @@ define <4 x i64>@test_int_x86_avx512_mask_pmovsxd_q_256(<4 x i32> %x0, <4 x i64> ret <4 x i64> %res4 } + +declare <16 x i8> @llvm.x86.avx512.cvtmask2b.128(i16) + +define <16 x i8>@test_int_x86_avx512_cvtmask2b_128(i16 %x0) { +; CHECK-LABEL: test_int_x86_avx512_cvtmask2b_128: +; CHECK: ## BB#0: +; CHECK-NEXT: kmovd %edi, %k0 ## encoding: [0xc5,0xfb,0x92,0xc7] +; CHECK-NEXT: vpmovm2b %k0, %xmm0 ## encoding: [0x62,0xf2,0x7e,0x08,0x28,0xc0] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <16 x i8> @llvm.x86.avx512.cvtmask2b.128(i16 %x0) + ret <16 x i8> %res +} + +declare <32 x i8> @llvm.x86.avx512.cvtmask2b.256(i32) + +define <32 x i8>@test_int_x86_avx512_cvtmask2b_256(i32 %x0) { +; CHECK-LABEL: test_int_x86_avx512_cvtmask2b_256: +; CHECK: ## BB#0: +; CHECK-NEXT: kmovd %edi, %k0 ## encoding: [0xc5,0xfb,0x92,0xc7] +; CHECK-NEXT: vpmovm2b %k0, %ymm0 ## encoding: [0x62,0xf2,0x7e,0x28,0x28,0xc0] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <32 x i8> @llvm.x86.avx512.cvtmask2b.256(i32 %x0) + ret <32 x i8> %res +} + +declare <8 x i16> @llvm.x86.avx512.cvtmask2w.128(i8) + +define <8 x i16>@test_int_x86_avx512_cvtmask2w_128(i8 %x0) { +; CHECK-LABEL: test_int_x86_avx512_cvtmask2w_128: +; CHECK: ## BB#0: +; CHECK-NEXT: kmovd %edi, %k0 ## encoding: [0xc5,0xfb,0x92,0xc7] +; CHECK-NEXT: vpmovm2w %k0, %xmm0 ## encoding: [0x62,0xf2,0xfe,0x08,0x28,0xc0] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <8 x i16> @llvm.x86.avx512.cvtmask2w.128(i8 %x0) + ret <8 x i16> %res +} + +declare <16 x i16> @llvm.x86.avx512.cvtmask2w.256(i16) + +define <16 x i16>@test_int_x86_avx512_cvtmask2w_256(i16 %x0) { +; CHECK-LABEL: test_int_x86_avx512_cvtmask2w_256: +; CHECK: ## BB#0: +; CHECK-NEXT: kmovd %edi, %k0 ## encoding: [0xc5,0xfb,0x92,0xc7] +; CHECK-NEXT: vpmovm2w %k0, %ymm0 ## encoding: [0x62,0xf2,0xfe,0x28,0x28,0xc0] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <16 x i16> @llvm.x86.avx512.cvtmask2w.256(i16 %x0) + ret <16 x i16> %res +} define <8 x i16> @test_mask_packs_epi32_rr_128(<4 x i32> %a, <4 x i32> %b) { ; CHECK-LABEL: test_mask_packs_epi32_rr_128: ; CHECK: ## BB#0: diff --git a/test/CodeGen/X86/avx512bwvl-intrinsics.ll b/test/CodeGen/X86/avx512bwvl-intrinsics.ll index d02eb2b55e5..1d0a3be0694 100644 --- a/test/CodeGen/X86/avx512bwvl-intrinsics.ll +++ b/test/CodeGen/X86/avx512bwvl-intrinsics.ll @@ -3367,54 +3367,6 @@ define i16@test_int_x86_avx512_cvtw2mask_256(<16 x i16> %x0) { ret i16 %res } -declare <16 x i8> @llvm.x86.avx512.cvtmask2b.128(i16) - -define <16 x i8>@test_int_x86_avx512_cvtmask2b_128(i16 %x0) { -; CHECK-LABEL: test_int_x86_avx512_cvtmask2b_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovd %edi, %k0 ## encoding: [0xc5,0xfb,0x92,0xc7] -; CHECK-NEXT: vpmovm2b %k0, %xmm0 ## encoding: [0x62,0xf2,0x7e,0x08,0x28,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <16 x i8> @llvm.x86.avx512.cvtmask2b.128(i16 %x0) - ret <16 x i8> %res -} - -declare <32 x i8> @llvm.x86.avx512.cvtmask2b.256(i32) - -define <32 x i8>@test_int_x86_avx512_cvtmask2b_256(i32 %x0) { -; CHECK-LABEL: test_int_x86_avx512_cvtmask2b_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovd %edi, %k0 ## encoding: [0xc5,0xfb,0x92,0xc7] -; CHECK-NEXT: vpmovm2b %k0, %ymm0 ## encoding: [0x62,0xf2,0x7e,0x28,0x28,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <32 x i8> @llvm.x86.avx512.cvtmask2b.256(i32 %x0) - ret <32 x i8> %res -} - -declare <8 x i16> @llvm.x86.avx512.cvtmask2w.128(i8) - -define <8 x i16>@test_int_x86_avx512_cvtmask2w_128(i8 %x0) { -; CHECK-LABEL: test_int_x86_avx512_cvtmask2w_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovd %edi, %k0 ## encoding: [0xc5,0xfb,0x92,0xc7] -; CHECK-NEXT: vpmovm2w %k0, %xmm0 ## encoding: [0x62,0xf2,0xfe,0x08,0x28,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <8 x i16> @llvm.x86.avx512.cvtmask2w.128(i8 %x0) - ret <8 x i16> %res -} - -declare <16 x i16> @llvm.x86.avx512.cvtmask2w.256(i16) - -define <16 x i16>@test_int_x86_avx512_cvtmask2w_256(i16 %x0) { -; CHECK-LABEL: test_int_x86_avx512_cvtmask2w_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovd %edi, %k0 ## encoding: [0xc5,0xfb,0x92,0xc7] -; CHECK-NEXT: vpmovm2w %k0, %ymm0 ## encoding: [0x62,0xf2,0xfe,0x28,0x28,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <16 x i16> @llvm.x86.avx512.cvtmask2w.256(i16 %x0) - ret <16 x i16> %res -} - declare <16 x i16> @llvm.x86.avx512.mask.psrlv16.hi(<16 x i16>, <16 x i16>, <16 x i16>, i16) define <16 x i16>@test_int_x86_avx512_mask_psrlv16_hi(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 %x3) { diff --git a/test/CodeGen/X86/avx512dq-intrinsics-upgrade.ll b/test/CodeGen/X86/avx512dq-intrinsics-upgrade.ll index e06b6306920..c5478dad422 100644 --- a/test/CodeGen/X86/avx512dq-intrinsics-upgrade.ll +++ b/test/CodeGen/X86/avx512dq-intrinsics-upgrade.ll @@ -131,3 +131,28 @@ define <8 x i64>@test_int_x86_avx512_mask_inserti64x2_512(<8 x i64> %x0, <2 x i6 %res4 = add <8 x i64> %res2, %res3 ret <8 x i64> %res4 } + + +declare <16 x i32> @llvm.x86.avx512.cvtmask2d.512(i16) + +define <16 x i32>@test_int_x86_avx512_cvtmask2d_512(i16 %x0) { +; CHECK-LABEL: test_int_x86_avx512_cvtmask2d_512: +; CHECK: ## BB#0: +; CHECK-NEXT: kmovw %edi, %k0 +; CHECK-NEXT: vpmovm2d %k0, %zmm0 +; CHECK-NEXT: retq + %res = call <16 x i32> @llvm.x86.avx512.cvtmask2d.512(i16 %x0) + ret <16 x i32> %res +} + +declare <8 x i64> @llvm.x86.avx512.cvtmask2q.512(i8) + +define <8 x i64>@test_int_x86_avx512_cvtmask2q_512(i8 %x0) { +; CHECK-LABEL: test_int_x86_avx512_cvtmask2q_512: +; CHECK: ## BB#0: +; CHECK-NEXT: kmovw %edi, %k0 +; CHECK-NEXT: vpmovm2q %k0, %zmm0 +; CHECK-NEXT: retq + %res = call <8 x i64> @llvm.x86.avx512.cvtmask2q.512(i8 %x0) + ret <8 x i64> %res +} diff --git a/test/CodeGen/X86/avx512dq-intrinsics.ll b/test/CodeGen/X86/avx512dq-intrinsics.ll index 16df2f7ea1a..000390404b5 100644 --- a/test/CodeGen/X86/avx512dq-intrinsics.ll +++ b/test/CodeGen/X86/avx512dq-intrinsics.ll @@ -472,30 +472,6 @@ define i8@test_int_x86_avx512_cvtq2mask_512(<8 x i64> %x0) { ret i8 %res } -declare <16 x i32> @llvm.x86.avx512.cvtmask2d.512(i16) - -define <16 x i32>@test_int_x86_avx512_cvtmask2d_512(i16 %x0) { -; CHECK-LABEL: test_int_x86_avx512_cvtmask2d_512: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k0 -; CHECK-NEXT: vpmovm2d %k0, %zmm0 -; CHECK-NEXT: retq - %res = call <16 x i32> @llvm.x86.avx512.cvtmask2d.512(i16 %x0) - ret <16 x i32> %res -} - -declare <8 x i64> @llvm.x86.avx512.cvtmask2q.512(i8) - -define <8 x i64>@test_int_x86_avx512_cvtmask2q_512(i8 %x0) { -; CHECK-LABEL: test_int_x86_avx512_cvtmask2q_512: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k0 -; CHECK-NEXT: vpmovm2q %k0, %zmm0 -; CHECK-NEXT: retq - %res = call <8 x i64> @llvm.x86.avx512.cvtmask2q.512(i8 %x0) - ret <8 x i64> %res -} - declare <16 x float> @llvm.x86.avx512.mask.broadcastf32x8.512(<8 x float>, <16 x float>, i16) define <16 x float>@test_int_x86_avx512_mask_broadcastf32x8_512(<8 x float> %x0, <16 x float> %x2, i16 %mask) { diff --git a/test/CodeGen/X86/avx512dqvl-intrinsics-upgrade.ll b/test/CodeGen/X86/avx512dqvl-intrinsics-upgrade.ll index b6df958c9f1..52a84deebf5 100644 --- a/test/CodeGen/X86/avx512dqvl-intrinsics-upgrade.ll +++ b/test/CodeGen/X86/avx512dqvl-intrinsics-upgrade.ll @@ -1619,3 +1619,51 @@ define <4 x i64>@test_int_x86_avx512_mask_inserti64x2_256(<4 x i64> %x0, <2 x i6 %res4 = add <4 x i64> %res3, %res2 ret <4 x i64> %res4 } + +declare <4 x i32> @llvm.x86.avx512.cvtmask2d.128(i8) + +define <4 x i32>@test_int_x86_avx512_cvtmask2d_128(i8 %x0) { +; CHECK-LABEL: test_int_x86_avx512_cvtmask2d_128: +; CHECK: ## BB#0: +; CHECK-NEXT: kmovw %edi, %k0 ## encoding: [0xc5,0xf8,0x92,0xc7] +; CHECK-NEXT: vpmovm2d %k0, %xmm0 ## encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <4 x i32> @llvm.x86.avx512.cvtmask2d.128(i8 %x0) + ret <4 x i32> %res +} + +declare <8 x i32> @llvm.x86.avx512.cvtmask2d.256(i8) + +define <8 x i32>@test_int_x86_avx512_cvtmask2d_256(i8 %x0) { +; CHECK-LABEL: test_int_x86_avx512_cvtmask2d_256: +; CHECK: ## BB#0: +; CHECK-NEXT: kmovw %edi, %k0 ## encoding: [0xc5,0xf8,0x92,0xc7] +; CHECK-NEXT: vpmovm2d %k0, %ymm0 ## encoding: [0x62,0xf2,0x7e,0x28,0x38,0xc0] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <8 x i32> @llvm.x86.avx512.cvtmask2d.256(i8 %x0) + ret <8 x i32> %res +} + +declare <2 x i64> @llvm.x86.avx512.cvtmask2q.128(i8) + +define <2 x i64>@test_int_x86_avx512_cvtmask2q_128(i8 %x0) { +; CHECK-LABEL: test_int_x86_avx512_cvtmask2q_128: +; CHECK: ## BB#0: +; CHECK-NEXT: kmovw %edi, %k0 ## encoding: [0xc5,0xf8,0x92,0xc7] +; CHECK-NEXT: vpmovm2q %k0, %xmm0 ## encoding: [0x62,0xf2,0xfe,0x08,0x38,0xc0] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <2 x i64> @llvm.x86.avx512.cvtmask2q.128(i8 %x0) + ret <2 x i64> %res +} + +declare <4 x i64> @llvm.x86.avx512.cvtmask2q.256(i8) + +define <4 x i64>@test_int_x86_avx512_cvtmask2q_256(i8 %x0) { +; CHECK-LABEL: test_int_x86_avx512_cvtmask2q_256: +; CHECK: ## BB#0: +; CHECK-NEXT: kmovw %edi, %k0 ## encoding: [0xc5,0xf8,0x92,0xc7] +; CHECK-NEXT: vpmovm2q %k0, %ymm0 ## encoding: [0x62,0xf2,0xfe,0x28,0x38,0xc0] +; CHECK-NEXT: retq ## encoding: [0xc3] + %res = call <4 x i64> @llvm.x86.avx512.cvtmask2q.256(i8 %x0) + ret <4 x i64> %res +} diff --git a/test/CodeGen/X86/avx512dqvl-intrinsics.ll b/test/CodeGen/X86/avx512dqvl-intrinsics.ll index 04d3602c305..ad9ea93c203 100644 --- a/test/CodeGen/X86/avx512dqvl-intrinsics.ll +++ b/test/CodeGen/X86/avx512dqvl-intrinsics.ll @@ -744,54 +744,6 @@ define i8@test_int_x86_avx512_cvtq2mask_256(<4 x i64> %x0) { ret i8 %res } -declare <4 x i32> @llvm.x86.avx512.cvtmask2d.128(i8) - -define <4 x i32>@test_int_x86_avx512_cvtmask2d_128(i8 %x0) { -; CHECK-LABEL: test_int_x86_avx512_cvtmask2d_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k0 ## encoding: [0xc5,0xf8,0x92,0xc7] -; CHECK-NEXT: vpmovm2d %k0, %xmm0 ## encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <4 x i32> @llvm.x86.avx512.cvtmask2d.128(i8 %x0) - ret <4 x i32> %res -} - -declare <8 x i32> @llvm.x86.avx512.cvtmask2d.256(i8) - -define <8 x i32>@test_int_x86_avx512_cvtmask2d_256(i8 %x0) { -; CHECK-LABEL: test_int_x86_avx512_cvtmask2d_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k0 ## encoding: [0xc5,0xf8,0x92,0xc7] -; CHECK-NEXT: vpmovm2d %k0, %ymm0 ## encoding: [0x62,0xf2,0x7e,0x28,0x38,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <8 x i32> @llvm.x86.avx512.cvtmask2d.256(i8 %x0) - ret <8 x i32> %res -} - -declare <2 x i64> @llvm.x86.avx512.cvtmask2q.128(i8) - -define <2 x i64>@test_int_x86_avx512_cvtmask2q_128(i8 %x0) { -; CHECK-LABEL: test_int_x86_avx512_cvtmask2q_128: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k0 ## encoding: [0xc5,0xf8,0x92,0xc7] -; CHECK-NEXT: vpmovm2q %k0, %xmm0 ## encoding: [0x62,0xf2,0xfe,0x08,0x38,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <2 x i64> @llvm.x86.avx512.cvtmask2q.128(i8 %x0) - ret <2 x i64> %res -} - -declare <4 x i64> @llvm.x86.avx512.cvtmask2q.256(i8) - -define <4 x i64>@test_int_x86_avx512_cvtmask2q_256(i8 %x0) { -; CHECK-LABEL: test_int_x86_avx512_cvtmask2q_256: -; CHECK: ## BB#0: -; CHECK-NEXT: kmovw %edi, %k0 ## encoding: [0xc5,0xf8,0x92,0xc7] -; CHECK-NEXT: vpmovm2q %k0, %ymm0 ## encoding: [0x62,0xf2,0xfe,0x28,0x38,0xc0] -; CHECK-NEXT: retq ## encoding: [0xc3] - %res = call <4 x i64> @llvm.x86.avx512.cvtmask2q.256(i8 %x0) - ret <4 x i64> %res -} - declare <4 x double> @llvm.x86.avx512.mask.broadcastf64x2.256(<2 x double>, <4 x double>, i8) define <4 x double>@test_int_x86_avx512_mask_broadcastf64x2_256(<2 x double> %x0, <4 x double> %x2, i8 %mask) {