From 83fc76e8ebb4a81ff3142035d29275e60afa3ada Mon Sep 17 00:00:00 2001 From: Michael Zuckerman Date: Thu, 7 Jan 2016 14:42:20 +0000 Subject: [PATCH] [AVX512] add PSRAV Intrinsic Differential Revision: http://reviews.llvm.org/D15856 git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@257063 91177308-0d34-0410-b5e6-96231b3b80d8 --- include/llvm/IR/IntrinsicsX86.td | 22 ++++++ lib/Target/X86/X86IntrinsicsInfo.h | 7 ++ test/CodeGen/X86/avx512bw-intrinsics.ll | 20 ++++++ test/CodeGen/X86/avx512bwvl-intrinsics.ll | 41 +++++++++++ test/CodeGen/X86/avx512vl-intrinsics.ll | 85 +++++++++++++++++++++++ 5 files changed, 175 insertions(+) diff --git a/include/llvm/IR/IntrinsicsX86.td b/include/llvm/IR/IntrinsicsX86.td index 2bec5c84df6..66aa1120563 100644 --- a/include/llvm/IR/IntrinsicsX86.td +++ b/include/llvm/IR/IntrinsicsX86.td @@ -2877,6 +2877,28 @@ let TargetPrefix = "x86" in { // All intrinsics start with "llvm.x86.". Intrinsic<[llvm_v8i64_ty], [llvm_v8i64_ty, llvm_i8_ty, llvm_v8i64_ty, llvm_i8_ty], [IntrNoMem]>; + def int_x86_avx512_mask_psrav16_hi : GCCBuiltin<"__builtin_ia32_psrav16hi_mask">, + Intrinsic<[llvm_v16i16_ty], [llvm_v16i16_ty, + llvm_v16i16_ty, llvm_v16i16_ty, llvm_i16_ty], [IntrNoMem]>; + def int_x86_avx512_mask_psrav32_hi : GCCBuiltin<"__builtin_ia32_psrav32hi_mask">, + Intrinsic<[llvm_v32i16_ty], [llvm_v32i16_ty, + llvm_v32i16_ty, llvm_v32i16_ty, llvm_i32_ty], [IntrNoMem]>; + def int_x86_avx512_mask_psrav4_si : GCCBuiltin<"__builtin_ia32_psrav4si_mask">, + Intrinsic<[llvm_v4i32_ty], [llvm_v4i32_ty, + llvm_v4i32_ty, llvm_v4i32_ty, llvm_i8_ty], [IntrNoMem]>; + def int_x86_avx512_mask_psrav8_hi : GCCBuiltin<"__builtin_ia32_psrav8hi_mask">, + Intrinsic<[llvm_v8i16_ty], [llvm_v8i16_ty, + llvm_v8i16_ty, llvm_v8i16_ty, llvm_i8_ty], [IntrNoMem]>; + def int_x86_avx512_mask_psrav8_si : GCCBuiltin<"__builtin_ia32_psrav8si_mask">, + Intrinsic<[llvm_v8i32_ty], [llvm_v8i32_ty, + llvm_v8i32_ty, llvm_v8i32_ty, llvm_i8_ty], [IntrNoMem]>; + def int_x86_avx512_mask_psrav_q_128 : GCCBuiltin<"__builtin_ia32_psravq128_mask">, + Intrinsic<[llvm_v2i64_ty], [llvm_v2i64_ty, + llvm_v2i64_ty, llvm_v2i64_ty, llvm_i8_ty], [IntrNoMem]>; + def int_x86_avx512_mask_psrav_q_256 : GCCBuiltin<"__builtin_ia32_psravq256_mask">, + Intrinsic<[llvm_v4i64_ty], [llvm_v4i64_ty, + llvm_v4i64_ty, llvm_v4i64_ty, llvm_i8_ty], [IntrNoMem]>; + def int_x86_avx512_mask_psrlv16_hi : GCCBuiltin<"__builtin_ia32_psrlv16hi_mask">, Intrinsic<[llvm_v16i16_ty], [llvm_v16i16_ty, llvm_v16i16_ty, llvm_v16i16_ty, llvm_i16_ty], [IntrNoMem]>; diff --git a/lib/Target/X86/X86IntrinsicsInfo.h b/lib/Target/X86/X86IntrinsicsInfo.h index 512ed82f3e5..e460a3fa37e 100644 --- a/lib/Target/X86/X86IntrinsicsInfo.h +++ b/lib/Target/X86/X86IntrinsicsInfo.h @@ -1261,8 +1261,15 @@ static const IntrinsicData IntrinsicsWithoutChain[] = { X86_INTRINSIC_DATA(avx512_mask_psra_wi_512, INTR_TYPE_2OP_MASK, X86ISD::VSRAI, 0), X86_INTRINSIC_DATA(avx512_mask_psrai_d, VSHIFT_MASK, X86ISD::VSRAI, 0), X86_INTRINSIC_DATA(avx512_mask_psrai_q, VSHIFT_MASK, X86ISD::VSRAI, 0), + X86_INTRINSIC_DATA(avx512_mask_psrav16_hi, INTR_TYPE_2OP_MASK, ISD::SRA, 0), + X86_INTRINSIC_DATA(avx512_mask_psrav32_hi, INTR_TYPE_2OP_MASK, ISD::SRA, 0), + X86_INTRINSIC_DATA(avx512_mask_psrav4_si, INTR_TYPE_2OP_MASK, ISD::SRA, 0), + X86_INTRINSIC_DATA(avx512_mask_psrav8_hi, INTR_TYPE_2OP_MASK, ISD::SRA, 0), + X86_INTRINSIC_DATA(avx512_mask_psrav8_si, INTR_TYPE_2OP_MASK, ISD::SRA, 0), X86_INTRINSIC_DATA(avx512_mask_psrav_d, INTR_TYPE_2OP_MASK, ISD::SRA, 0), X86_INTRINSIC_DATA(avx512_mask_psrav_q, INTR_TYPE_2OP_MASK, ISD::SRA, 0), + X86_INTRINSIC_DATA(avx512_mask_psrav_q_128, INTR_TYPE_2OP_MASK, ISD::SRA, 0), + X86_INTRINSIC_DATA(avx512_mask_psrav_q_256, INTR_TYPE_2OP_MASK, ISD::SRA, 0), X86_INTRINSIC_DATA(avx512_mask_psrl_d, INTR_TYPE_2OP_MASK, X86ISD::VSRL, 0), X86_INTRINSIC_DATA(avx512_mask_psrl_d_128, INTR_TYPE_2OP_MASK, X86ISD::VSRL, 0), X86_INTRINSIC_DATA(avx512_mask_psrl_d_256, INTR_TYPE_2OP_MASK, X86ISD::VSRL, 0), diff --git a/test/CodeGen/X86/avx512bw-intrinsics.ll b/test/CodeGen/X86/avx512bw-intrinsics.ll index 94940ed5aee..ef31b0d1f50 100644 --- a/test/CodeGen/X86/avx512bw-intrinsics.ll +++ b/test/CodeGen/X86/avx512bw-intrinsics.ll @@ -3006,3 +3006,23 @@ define <32 x i16>@test_int_x86_avx512_mask_pshufl_w_512(<32 x i16> %x0, i8 %x1, %res4 = add <32 x i16> %res3, %res2 ret <32 x i16> %res4 } + +declare <32 x i16> @llvm.x86.avx512.mask.psrav32_hi(<32 x i16>, <32 x i16>, <32 x i16>, i32) + +define <32 x i16>@test_int_x86_avx512_mask_psrav32_hi(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_psrav32_hi: +; CHECK: ## BB#0: +; CHECK-NEXT: kmovd %edi, %k1 +; CHECK-NEXT: vpsravw %zmm1, %zmm0, %zmm2 {%k1} +; CHECK-NEXT: vpsravw %zmm1, %zmm0, %zmm3 {%k1} {z} +; CHECK-NEXT: vpsravw %zmm1, %zmm0, %zmm0 +; CHECK-NEXT: vpaddw %zmm3, %zmm2, %zmm1 +; CHECK-NEXT: vpaddw %zmm0, %zmm1, %zmm0 +; CHECK-NEXT: retq + %res = call <32 x i16> @llvm.x86.avx512.mask.psrav32_hi(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 %x3) + %res1 = call <32 x i16> @llvm.x86.avx512.mask.psrav32_hi(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> zeroinitializer, i32 %x3) + %res2 = call <32 x i16> @llvm.x86.avx512.mask.psrav32_hi(<32 x i16> %x0, <32 x i16> %x1, <32 x i16> %x2, i32 -1) + %res3 = add <32 x i16> %res, %res1 + %res4 = add <32 x i16> %res3, %res2 + ret <32 x i16> %res4 +} diff --git a/test/CodeGen/X86/avx512bwvl-intrinsics.ll b/test/CodeGen/X86/avx512bwvl-intrinsics.ll index 3e8514a2505..c0e2e60ec65 100644 --- a/test/CodeGen/X86/avx512bwvl-intrinsics.ll +++ b/test/CodeGen/X86/avx512bwvl-intrinsics.ll @@ -4844,3 +4844,44 @@ define <16 x i16>@test_int_x86_avx512_mask_pshufl_w_256(<16 x i16> %x0, i8 %x1, %res4 = add <16 x i16> %res3, %res2 ret <16 x i16> %res4 } + +declare <16 x i16> @llvm.x86.avx512.mask.psrav16.hi(<16 x i16>, <16 x i16>, <16 x i16>, i16) + +define <16 x i16>@test_int_x86_avx512_mask_psrav16_hi(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_psrav16_hi: +; CHECK: ## BB#0: +; CHECK-NEXT: kmovw %edi, %k1 +; CHECK-NEXT: vpsravw %ymm1, %ymm0, %ymm2 {%k1} +; CHECK-NEXT: vpsravw %ymm1, %ymm0, %ymm3 {%k1} {z} +; CHECK-NEXT: vpsravw %ymm1, %ymm0, %ymm0 +; CHECK-NEXT: vpaddw %ymm3, %ymm2, %ymm1 +; CHECK-NEXT: vpaddw %ymm0, %ymm1, %ymm0 +; CHECK-NEXT: retq + %res = call <16 x i16> @llvm.x86.avx512.mask.psrav16.hi(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 %x3) + %res1 = call <16 x i16> @llvm.x86.avx512.mask.psrav16.hi(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> zeroinitializer, i16 %x3) + %res2 = call <16 x i16> @llvm.x86.avx512.mask.psrav16.hi(<16 x i16> %x0, <16 x i16> %x1, <16 x i16> %x2, i16 -1) + %res3 = add <16 x i16> %res, %res1 + %res4 = add <16 x i16> %res3, %res2 + ret <16 x i16> %res4 +} + +declare <8 x i16> @llvm.x86.avx512.mask.psrav8.hi(<8 x i16>, <8 x i16>, <8 x i16>, i8) + +define <8 x i16>@test_int_x86_avx512_mask_psrav8_hi(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_psrav8_hi: +; CHECK: ## BB#0: +; CHECK-NEXT: movzbl %dil, %eax +; CHECK-NEXT: kmovw %eax, %k1 +; CHECK-NEXT: vpsravw %xmm1, %xmm0, %xmm2 {%k1} +; CHECK-NEXT: vpsravw %xmm1, %xmm0, %xmm3 {%k1} {z} +; CHECK-NEXT: vpsravw %xmm1, %xmm0, %xmm0 +; CHECK-NEXT: vpaddw %xmm3, %xmm2, %xmm1 +; CHECK-NEXT: vpaddw %xmm0, %xmm1, %xmm0 +; CHECK-NEXT: retq + %res = call <8 x i16> @llvm.x86.avx512.mask.psrav8.hi(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 %x3) + %res1 = call <8 x i16> @llvm.x86.avx512.mask.psrav8.hi(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> zeroinitializer, i8 %x3) + %res2 = call <8 x i16> @llvm.x86.avx512.mask.psrav8.hi(<8 x i16> %x0, <8 x i16> %x1, <8 x i16> %x2, i8 -1) + %res3 = add <8 x i16> %res, %res1 + %res4 = add <8 x i16> %res3, %res2 + ret <8 x i16> %res4 +} \ No newline at end of file diff --git a/test/CodeGen/X86/avx512vl-intrinsics.ll b/test/CodeGen/X86/avx512vl-intrinsics.ll index 8ab34bd8c43..6e9b1efdc66 100644 --- a/test/CodeGen/X86/avx512vl-intrinsics.ll +++ b/test/CodeGen/X86/avx512vl-intrinsics.ll @@ -6387,3 +6387,88 @@ define <4 x i64>@test_int_x86_avx512_mask_psll_qi_256(<4 x i64> %x0, i8 %x1, <4 %res4 = add <4 x i64> %res3, %res2 ret <4 x i64> %res4 } + +declare <4 x i32> @llvm.x86.avx512.mask.psrav4.si(<4 x i32>, <4 x i32>, <4 x i32>, i8) + +define <4 x i32>@test_int_x86_avx512_mask_psrav4_si(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_psrav4_si: +; CHECK: ## BB#0: +; CHECK-NEXT: movzbl %dil, %eax +; CHECK-NEXT: kmovw %eax, %k1 +; CHECK-NEXT: vpsravd %xmm1, %xmm0, %xmm2 {%k1} +; CHECK-NEXT: vpsravd %xmm1, %xmm0, %xmm3 {%k1} {z} +; CHECK-NEXT: vpsravd %xmm1, %xmm0, %xmm0 +; CHECK-NEXT: vpaddd %xmm3, %xmm2, %xmm1 +; CHECK-NEXT: vpaddd %xmm0, %xmm1, %xmm0 +; CHECK-NEXT: retq + %res = call <4 x i32> @llvm.x86.avx512.mask.psrav4.si(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 %x3) + %res1 = call <4 x i32> @llvm.x86.avx512.mask.psrav4.si(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> zeroinitializer, i8 %x3) + %res2 = call <4 x i32> @llvm.x86.avx512.mask.psrav4.si(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2, i8 -1) + %res3 = add <4 x i32> %res, %res1 + %res4 = add <4 x i32> %res3, %res2 + ret <4 x i32> %res4 +} + +declare <8 x i32> @llvm.x86.avx512.mask.psrav8.si(<8 x i32>, <8 x i32>, <8 x i32>, i8) + +define <8 x i32>@test_int_x86_avx512_mask_psrav8_si(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_psrav8_si: +; CHECK: ## BB#0: +; CHECK-NEXT: movzbl %dil, %eax +; CHECK-NEXT: kmovw %eax, %k1 +; CHECK-NEXT: vpsravd %ymm1, %ymm0, %ymm2 {%k1} +; CHECK-NEXT: vpsravd %ymm1, %ymm0, %ymm3 {%k1} {z} +; CHECK-NEXT: vpsravd %ymm1, %ymm0, %ymm0 +; CHECK-NEXT: vpaddd %ymm3, %ymm2, %ymm1 +; CHECK-NEXT: vpaddd %ymm0, %ymm1, %ymm0 +; CHECK-NEXT: retq + %res = call <8 x i32> @llvm.x86.avx512.mask.psrav8.si(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2, i8 %x3) + %res1 = call <8 x i32> @llvm.x86.avx512.mask.psrav8.si(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> zeroinitializer, i8 %x3) + %res2 = call <8 x i32> @llvm.x86.avx512.mask.psrav8.si(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2, i8 -1) + %res3 = add <8 x i32> %res, %res1 + %res4 = add <8 x i32> %res3, %res2 + ret <8 x i32> %res4 +} + +declare <2 x i64> @llvm.x86.avx512.mask.psrav.q.128(<2 x i64>, <2 x i64>, <2 x i64>, i8) + +define <2 x i64>@test_int_x86_avx512_mask_psrav_q_128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_psrav_q_128: +; CHECK: ## BB#0: +; CHECK-NEXT: movzbl %dil, %eax +; CHECK-NEXT: kmovw %eax, %k1 +; CHECK-NEXT: vpsravq %xmm1, %xmm0, %xmm2 {%k1} +; CHECK-NEXT: vpsravq %xmm1, %xmm0, %xmm3 {%k1} {z} +; CHECK-NEXT: vpsravq %xmm1, %xmm0, %xmm0 +; CHECK-NEXT: vpaddq %xmm3, %xmm2, %xmm1 +; CHECK-NEXT: vpaddq %xmm0, %xmm1, %xmm0 +; CHECK-NEXT: retq + %res = call <2 x i64> @llvm.x86.avx512.mask.psrav.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.psrav.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> zeroinitializer, i8 %x3) + %res2 = call <2 x i64> @llvm.x86.avx512.mask.psrav.q.128(<2 x i64> %x0, <2 x i64> %x1, <2 x i64> %x2, i8 -1) + %res3 = add <2 x i64> %res, %res1 + %res4 = add <2 x i64> %res3, %res2 + ret <2 x i64> %res4 +} + +declare <4 x i64> @llvm.x86.avx512.mask.psrav.q.256(<4 x i64>, <4 x i64>, <4 x i64>, i8) + +define <4 x i64>@test_int_x86_avx512_mask_psrav_q_256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 %x3) { +; CHECK-LABEL: test_int_x86_avx512_mask_psrav_q_256: +; CHECK: ## BB#0: +; CHECK-NEXT: movzbl %dil, %eax +; CHECK-NEXT: kmovw %eax, %k1 +; CHECK-NEXT: vpsravq %ymm1, %ymm0, %ymm2 {%k1} +; CHECK-NEXT: vpsravq %ymm1, %ymm0, %ymm3 {%k1} {z} +; CHECK-NEXT: vpsravq %ymm1, %ymm0, %ymm0 +; CHECK-NEXT: vpaddq %ymm3, %ymm2, %ymm1 +; CHECK-NEXT: vpaddq %ymm0, %ymm1, %ymm0 +; CHECK-NEXT: retq + %res = call <4 x i64> @llvm.x86.avx512.mask.psrav.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.psrav.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> zeroinitializer, i8 %x3) + %res2 = call <4 x i64> @llvm.x86.avx512.mask.psrav.q.256(<4 x i64> %x0, <4 x i64> %x1, <4 x i64> %x2, i8 -1) + %res3 = add <4 x i64> %res, %res1 + %res4 = add <4 x i64> %res3, %res2 + ret <4 x i64> %res4 +} + -- 2.34.1