1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py 2; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx512dq --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X86 3; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512dq --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X64 4 5declare <2 x double> @llvm.x86.avx512.mask.vextractf64x2.512(<8 x double>, i32, <2 x double>, i8) 6 7define <2 x double>@test_int_x86_avx512_mask_vextractf64x2_512(<8 x double> %x0, <2 x double> %x2, i8 %x3) { 8; X86-LABEL: test_int_x86_avx512_mask_vextractf64x2_512: 9; X86: # %bb.0: 10; X86-NEXT: vextractf128 $1, %ymm0, %xmm2 # encoding: [0xc4,0xe3,0x7d,0x19,0xc2,0x01] 11; X86-NEXT: kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04] 12; X86-NEXT: vextractf64x2 $1, %zmm0, %xmm1 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x19,0xc1,0x01] 13; X86-NEXT: vaddpd %xmm2, %xmm1, %xmm1 # encoding: [0xc5,0xf1,0x58,0xca] 14; X86-NEXT: vextractf64x2 $1, %zmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0x19,0xc0,0x01] 15; X86-NEXT: vaddpd %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf9,0x58,0xc1] 16; X86-NEXT: vzeroupper # encoding: [0xc5,0xf8,0x77] 17; X86-NEXT: retl # encoding: [0xc3] 18; 19; X64-LABEL: test_int_x86_avx512_mask_vextractf64x2_512: 20; X64: # %bb.0: 21; X64-NEXT: vextractf128 $1, %ymm0, %xmm2 # encoding: [0xc4,0xe3,0x7d,0x19,0xc2,0x01] 22; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf] 23; X64-NEXT: vextractf64x2 $1, %zmm0, %xmm1 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x19,0xc1,0x01] 24; X64-NEXT: vaddpd %xmm2, %xmm1, %xmm1 # encoding: [0xc5,0xf1,0x58,0xca] 25; X64-NEXT: vextractf64x2 $1, %zmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0x19,0xc0,0x01] 26; X64-NEXT: vaddpd %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf9,0x58,0xc1] 27; X64-NEXT: vzeroupper # encoding: [0xc5,0xf8,0x77] 28; X64-NEXT: retq # encoding: [0xc3] 29 %res = call <2 x double> @llvm.x86.avx512.mask.vextractf64x2.512(<8 x double> %x0,i32 1, <2 x double> %x2, i8 %x3) 30 %res2 = call <2 x double> @llvm.x86.avx512.mask.vextractf64x2.512(<8 x double> %x0,i32 1, <2 x double> zeroinitializer, i8 %x3) 31 %res1 = call <2 x double> @llvm.x86.avx512.mask.vextractf64x2.512(<8 x double> %x0,i32 1, <2 x double> zeroinitializer, i8 -1) 32 %res3 = fadd <2 x double> %res, %res1 33 %res4 = fadd <2 x double> %res2, %res3 34 ret <2 x double> %res4 35} 36 37declare <8 x float> @llvm.x86.avx512.mask.vextractf32x8.512(<16 x float>, i32, <8 x float>, i8) 38 39define <8 x float>@test_int_x86_avx512_mask_vextractf32x8(<16 x float> %x0, <8 x float> %x2, i8 %x3) { 40; X86-LABEL: test_int_x86_avx512_mask_vextractf32x8: 41; X86: # %bb.0: 42; X86-NEXT: vextractf64x4 $1, %zmm0, %ymm2 # encoding: [0x62,0xf3,0xfd,0x48,0x1b,0xc2,0x01] 43; X86-NEXT: kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04] 44; X86-NEXT: vextractf32x8 $1, %zmm0, %ymm1 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x1b,0xc1,0x01] 45; X86-NEXT: vaddps %ymm2, %ymm1, %ymm1 # encoding: [0xc5,0xf4,0x58,0xca] 46; X86-NEXT: vextractf32x8 $1, %zmm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x1b,0xc0,0x01] 47; X86-NEXT: vaddps %ymm1, %ymm0, %ymm0 # encoding: [0xc5,0xfc,0x58,0xc1] 48; X86-NEXT: retl # encoding: [0xc3] 49; 50; X64-LABEL: test_int_x86_avx512_mask_vextractf32x8: 51; X64: # %bb.0: 52; X64-NEXT: vextractf64x4 $1, %zmm0, %ymm2 # encoding: [0x62,0xf3,0xfd,0x48,0x1b,0xc2,0x01] 53; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf] 54; X64-NEXT: vextractf32x8 $1, %zmm0, %ymm1 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x1b,0xc1,0x01] 55; X64-NEXT: vaddps %ymm2, %ymm1, %ymm1 # encoding: [0xc5,0xf4,0x58,0xca] 56; X64-NEXT: vextractf32x8 $1, %zmm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x1b,0xc0,0x01] 57; X64-NEXT: vaddps %ymm1, %ymm0, %ymm0 # encoding: [0xc5,0xfc,0x58,0xc1] 58; X64-NEXT: retq # encoding: [0xc3] 59 %res = call <8 x float> @llvm.x86.avx512.mask.vextractf32x8.512(<16 x float> %x0,i32 1, <8 x float> %x2, i8 %x3) 60 %res2 = call <8 x float> @llvm.x86.avx512.mask.vextractf32x8.512(<16 x float> %x0,i32 1, <8 x float> zeroinitializer, i8 %x3) 61 %res1 = call <8 x float> @llvm.x86.avx512.mask.vextractf32x8.512(<16 x float> %x0,i32 1, <8 x float> zeroinitializer, i8 -1) 62 %res3 = fadd <8 x float> %res, %res1 63 %res4 = fadd <8 x float> %res2, %res3 64 ret <8 x float> %res4 65} 66 67declare <16 x float> @llvm.x86.avx512.mask.insertf32x8.512(<16 x float>, <8 x float>, i32, <16 x float>, i16) 68 69define <16 x float>@test_int_x86_avx512_mask_insertf32x8_512(<16 x float> %x0, <8 x float> %x1, <16 x float> %x3, i16 %x4) { 70; X86-LABEL: test_int_x86_avx512_mask_insertf32x8_512: 71; X86: # %bb.0: 72; X86-NEXT: vinsertf64x4 $1, %ymm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0xfd,0x48,0x1a,0xd9,0x01] 73; X86-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04] 74; X86-NEXT: vinsertf32x8 $1, %ymm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x1a,0xd1,0x01] 75; X86-NEXT: vinsertf32x8 $1, %ymm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x1a,0xc1,0x01] 76; X86-NEXT: vaddps %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0x6c,0x48,0x58,0xc0] 77; X86-NEXT: vaddps %zmm0, %zmm3, %zmm0 # encoding: [0x62,0xf1,0x64,0x48,0x58,0xc0] 78; X86-NEXT: retl # encoding: [0xc3] 79; 80; X64-LABEL: test_int_x86_avx512_mask_insertf32x8_512: 81; X64: # %bb.0: 82; X64-NEXT: vinsertf64x4 $1, %ymm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0xfd,0x48,0x1a,0xd9,0x01] 83; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf] 84; X64-NEXT: vinsertf32x8 $1, %ymm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x1a,0xd1,0x01] 85; X64-NEXT: vinsertf32x8 $1, %ymm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x1a,0xc1,0x01] 86; X64-NEXT: vaddps %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0x6c,0x48,0x58,0xc0] 87; X64-NEXT: vaddps %zmm0, %zmm3, %zmm0 # encoding: [0x62,0xf1,0x64,0x48,0x58,0xc0] 88; X64-NEXT: retq # encoding: [0xc3] 89 %res = call <16 x float> @llvm.x86.avx512.mask.insertf32x8.512(<16 x float> %x0, <8 x float> %x1, i32 1, <16 x float> %x3, i16 %x4) 90 %res1 = call <16 x float> @llvm.x86.avx512.mask.insertf32x8.512(<16 x float> %x0, <8 x float> %x1, i32 1, <16 x float> zeroinitializer, i16 %x4) 91 %res2 = call <16 x float> @llvm.x86.avx512.mask.insertf32x8.512(<16 x float> %x0, <8 x float> %x1, i32 1, <16 x float> %x3, i16 -1) 92 %res3 = fadd <16 x float> %res, %res1 93 %res4 = fadd <16 x float> %res2, %res3 94 ret <16 x float> %res4 95} 96 97declare <8 x double> @llvm.x86.avx512.mask.insertf64x2.512(<8 x double>, <2 x double>, i32, <8 x double>, i8) 98 99define <8 x double>@test_int_x86_avx512_mask_insertf64x2_512(<8 x double> %x0, <2 x double> %x1,<8 x double> %x3, i8 %x4) { 100; X86-LABEL: test_int_x86_avx512_mask_insertf64x2_512: 101; X86: # %bb.0: 102; X86-NEXT: vinsertf32x4 $1, %xmm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0x7d,0x48,0x18,0xd9,0x01] 103; X86-NEXT: kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04] 104; X86-NEXT: vinsertf64x2 $1, %xmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x18,0xd1,0x01] 105; X86-NEXT: vinsertf64x2 $1, %xmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0x18,0xc1,0x01] 106; X86-NEXT: vaddpd %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0xed,0x48,0x58,0xc0] 107; X86-NEXT: vaddpd %zmm3, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x58,0xc3] 108; X86-NEXT: retl # encoding: [0xc3] 109; 110; X64-LABEL: test_int_x86_avx512_mask_insertf64x2_512: 111; X64: # %bb.0: 112; X64-NEXT: vinsertf32x4 $1, %xmm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0x7d,0x48,0x18,0xd9,0x01] 113; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf] 114; X64-NEXT: vinsertf64x2 $1, %xmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x18,0xd1,0x01] 115; X64-NEXT: vinsertf64x2 $1, %xmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0x18,0xc1,0x01] 116; X64-NEXT: vaddpd %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0xed,0x48,0x58,0xc0] 117; X64-NEXT: vaddpd %zmm3, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x58,0xc3] 118; X64-NEXT: retq # encoding: [0xc3] 119 %res = call <8 x double> @llvm.x86.avx512.mask.insertf64x2.512(<8 x double> %x0, <2 x double> %x1, i32 1, <8 x double> %x3, i8 %x4) 120 %res1 = call <8 x double> @llvm.x86.avx512.mask.insertf64x2.512(<8 x double> %x0, <2 x double> %x1, i32 1, <8 x double> zeroinitializer, i8 %x4) 121 %res2 = call <8 x double> @llvm.x86.avx512.mask.insertf64x2.512(<8 x double> %x0, <2 x double> %x1, i32 1, <8 x double> %x3, i8 -1) 122 %res3 = fadd <8 x double> %res, %res1 123 %res4 = fadd <8 x double> %res3, %res2 124 ret <8 x double> %res4 125} 126 127declare <16 x i32> @llvm.x86.avx512.mask.inserti32x8.512(<16 x i32>, <8 x i32>, i32, <16 x i32>, i16) 128 129define <16 x i32>@test_int_x86_avx512_mask_inserti32x8_512(<16 x i32> %x0, <8 x i32> %x1, <16 x i32> %x3, i16 %x4) { 130; X86-LABEL: test_int_x86_avx512_mask_inserti32x8_512: 131; X86: # %bb.0: 132; X86-NEXT: vinserti64x4 $1, %ymm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0xfd,0x48,0x3a,0xd9,0x01] 133; X86-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04] 134; X86-NEXT: vinserti32x8 $1, %ymm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x3a,0xd1,0x01] 135; X86-NEXT: vinserti32x8 $1, %ymm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x3a,0xc1,0x01] 136; X86-NEXT: vpaddd %zmm3, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7d,0x48,0xfe,0xc3] 137; X86-NEXT: vpaddd %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0x6d,0x48,0xfe,0xc0] 138; X86-NEXT: retl # encoding: [0xc3] 139; 140; X64-LABEL: test_int_x86_avx512_mask_inserti32x8_512: 141; X64: # %bb.0: 142; X64-NEXT: vinserti64x4 $1, %ymm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0xfd,0x48,0x3a,0xd9,0x01] 143; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf] 144; X64-NEXT: vinserti32x8 $1, %ymm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x3a,0xd1,0x01] 145; X64-NEXT: vinserti32x8 $1, %ymm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x3a,0xc1,0x01] 146; X64-NEXT: vpaddd %zmm3, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7d,0x48,0xfe,0xc3] 147; X64-NEXT: vpaddd %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0x6d,0x48,0xfe,0xc0] 148; X64-NEXT: retq # encoding: [0xc3] 149 %res = call <16 x i32> @llvm.x86.avx512.mask.inserti32x8.512(<16 x i32> %x0, <8 x i32> %x1, i32 1, <16 x i32> %x3, i16 %x4) 150 %res1 = call <16 x i32> @llvm.x86.avx512.mask.inserti32x8.512(<16 x i32> %x0, <8 x i32> %x1, i32 1, <16 x i32> zeroinitializer, i16 %x4) 151 %res2 = call <16 x i32> @llvm.x86.avx512.mask.inserti32x8.512(<16 x i32> %x0, <8 x i32> %x1, i32 1, <16 x i32> %x3, i16 -1) 152 %res3 = add <16 x i32> %res, %res1 153 %res4 = add <16 x i32> %res3, %res2 154 ret <16 x i32> %res4 155} 156 157declare <8 x i64> @llvm.x86.avx512.mask.inserti64x2.512(<8 x i64>, <2 x i64>, i32, <8 x i64>, i8) 158 159define <8 x i64>@test_int_x86_avx512_mask_inserti64x2_512(<8 x i64> %x0, <2 x i64> %x1, <8 x i64> %x3, i8 %x4) { 160; X86-LABEL: test_int_x86_avx512_mask_inserti64x2_512: 161; X86: # %bb.0: 162; X86-NEXT: vinserti32x4 $1, %xmm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0x7d,0x48,0x38,0xd9,0x01] 163; X86-NEXT: kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04] 164; X86-NEXT: vinserti64x2 $1, %xmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x38,0xd1,0x01] 165; X86-NEXT: vinserti64x2 $1, %xmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0x38,0xc1,0x01] 166; X86-NEXT: vpaddq %zmm3, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0xd4,0xc3] 167; X86-NEXT: vpaddq %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0xed,0x48,0xd4,0xc0] 168; X86-NEXT: retl # encoding: [0xc3] 169; 170; X64-LABEL: test_int_x86_avx512_mask_inserti64x2_512: 171; X64: # %bb.0: 172; X64-NEXT: vinserti32x4 $1, %xmm1, %zmm0, %zmm3 # encoding: [0x62,0xf3,0x7d,0x48,0x38,0xd9,0x01] 173; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf] 174; X64-NEXT: vinserti64x2 $1, %xmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x38,0xd1,0x01] 175; X64-NEXT: vinserti64x2 $1, %xmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0x38,0xc1,0x01] 176; X64-NEXT: vpaddq %zmm3, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0xd4,0xc3] 177; X64-NEXT: vpaddq %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0xed,0x48,0xd4,0xc0] 178; X64-NEXT: retq # encoding: [0xc3] 179 %res = call <8 x i64> @llvm.x86.avx512.mask.inserti64x2.512(<8 x i64> %x0, <2 x i64> %x1, i32 1, <8 x i64> %x3, i8 %x4) 180 %res1 = call <8 x i64> @llvm.x86.avx512.mask.inserti64x2.512(<8 x i64> %x0, <2 x i64> %x1, i32 1, <8 x i64> zeroinitializer, i8 %x4) 181 %res2 = call <8 x i64> @llvm.x86.avx512.mask.inserti64x2.512(<8 x i64> %x0, <2 x i64> %x1, i32 1, <8 x i64> %x3, i8 -1) 182 %res3 = add <8 x i64> %res, %res1 183 %res4 = add <8 x i64> %res2, %res3 184 ret <8 x i64> %res4 185} 186 187 188declare <16 x i32> @llvm.x86.avx512.cvtmask2d.512(i16) 189 190define <16 x i32>@test_int_x86_avx512_cvtmask2d_512(i16 %x0) { 191; X86-LABEL: test_int_x86_avx512_cvtmask2d_512: 192; X86: # %bb.0: 193; X86-NEXT: kmovw {{[0-9]+}}(%esp), %k0 # encoding: [0xc5,0xf8,0x90,0x44,0x24,0x04] 194; X86-NEXT: vpmovm2d %k0, %zmm0 # encoding: [0x62,0xf2,0x7e,0x48,0x38,0xc0] 195; X86-NEXT: retl # encoding: [0xc3] 196; 197; X64-LABEL: test_int_x86_avx512_cvtmask2d_512: 198; X64: # %bb.0: 199; X64-NEXT: kmovw %edi, %k0 # encoding: [0xc5,0xf8,0x92,0xc7] 200; X64-NEXT: vpmovm2d %k0, %zmm0 # encoding: [0x62,0xf2,0x7e,0x48,0x38,0xc0] 201; X64-NEXT: retq # encoding: [0xc3] 202 %res = call <16 x i32> @llvm.x86.avx512.cvtmask2d.512(i16 %x0) 203 ret <16 x i32> %res 204} 205 206declare <8 x i64> @llvm.x86.avx512.cvtmask2q.512(i8) 207 208define <8 x i64>@test_int_x86_avx512_cvtmask2q_512(i8 %x0) { 209; X86-LABEL: test_int_x86_avx512_cvtmask2q_512: 210; X86: # %bb.0: 211; X86-NEXT: kmovb {{[0-9]+}}(%esp), %k0 # encoding: [0xc5,0xf9,0x90,0x44,0x24,0x04] 212; X86-NEXT: vpmovm2q %k0, %zmm0 # encoding: [0x62,0xf2,0xfe,0x48,0x38,0xc0] 213; X86-NEXT: retl # encoding: [0xc3] 214; 215; X64-LABEL: test_int_x86_avx512_cvtmask2q_512: 216; X64: # %bb.0: 217; X64-NEXT: kmovw %edi, %k0 # encoding: [0xc5,0xf8,0x92,0xc7] 218; X64-NEXT: vpmovm2q %k0, %zmm0 # encoding: [0x62,0xf2,0xfe,0x48,0x38,0xc0] 219; X64-NEXT: retq # encoding: [0xc3] 220 %res = call <8 x i64> @llvm.x86.avx512.cvtmask2q.512(i8 %x0) 221 ret <8 x i64> %res 222} 223 224declare <16 x float> @llvm.x86.avx512.mask.broadcastf32x8.512(<8 x float>, <16 x float>, i16) 225 226define <16 x float>@test_int_x86_avx512_mask_broadcastf32x8_512(<8 x float> %x0, <16 x float> %x2, i16 %mask) { 227; X86-LABEL: test_int_x86_avx512_mask_broadcastf32x8_512: 228; X86: # %bb.0: 229; X86-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0 230; X86-NEXT: vinsertf64x4 $1, %ymm0, %zmm0, %zmm2 # encoding: [0x62,0xf3,0xfd,0x48,0x1a,0xd0,0x01] 231; X86-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04] 232; X86-NEXT: vinsertf32x8 $1, %ymm0, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x1a,0xc8,0x01] 233; X86-NEXT: vaddps %zmm1, %zmm2, %zmm1 # encoding: [0x62,0xf1,0x6c,0x48,0x58,0xc9] 234; X86-NEXT: vinsertf32x8 $1, %ymm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x1a,0xc0,0x01] 235; X86-NEXT: vaddps %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x58,0xc1] 236; X86-NEXT: retl # encoding: [0xc3] 237; 238; X64-LABEL: test_int_x86_avx512_mask_broadcastf32x8_512: 239; X64: # %bb.0: 240; X64-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0 241; X64-NEXT: vinsertf64x4 $1, %ymm0, %zmm0, %zmm2 # encoding: [0x62,0xf3,0xfd,0x48,0x1a,0xd0,0x01] 242; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf] 243; X64-NEXT: vinsertf32x8 $1, %ymm0, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x1a,0xc8,0x01] 244; X64-NEXT: vaddps %zmm1, %zmm2, %zmm1 # encoding: [0x62,0xf1,0x6c,0x48,0x58,0xc9] 245; X64-NEXT: vinsertf32x8 $1, %ymm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x1a,0xc0,0x01] 246; X64-NEXT: vaddps %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x58,0xc1] 247; X64-NEXT: retq # encoding: [0xc3] 248 249 %res1 = call <16 x float> @llvm.x86.avx512.mask.broadcastf32x8.512(<8 x float> %x0, <16 x float> %x2, i16 -1) 250 %res2 = call <16 x float> @llvm.x86.avx512.mask.broadcastf32x8.512(<8 x float> %x0, <16 x float> %x2, i16 %mask) 251 %res3 = call <16 x float> @llvm.x86.avx512.mask.broadcastf32x8.512(<8 x float> %x0, <16 x float> zeroinitializer, i16 %mask) 252 %res4 = fadd <16 x float> %res1, %res2 253 %res5 = fadd <16 x float> %res3, %res4 254 ret <16 x float> %res5 255} 256 257define <16 x float>@test_int_x86_avx512_mask_broadcastf32x8_512_load(<8 x float>* %x0ptr, <16 x float> %x2, i16 %mask) { 258; X86-LABEL: test_int_x86_avx512_mask_broadcastf32x8_512_load: 259; X86: # %bb.0: 260; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04] 261; X86-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08] 262; X86-NEXT: vbroadcastf32x8 (%eax), %zmm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0x1b,0x00] 263; X86-NEXT: # zmm0 {%k1} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7] 264; X86-NEXT: retl # encoding: [0xc3] 265; 266; X64-LABEL: test_int_x86_avx512_mask_broadcastf32x8_512_load: 267; X64: # %bb.0: 268; X64-NEXT: kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce] 269; X64-NEXT: vbroadcastf32x8 (%rdi), %zmm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0x1b,0x07] 270; X64-NEXT: # zmm0 {%k1} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7] 271; X64-NEXT: retq # encoding: [0xc3] 272 273 %x0 = load <8 x float>, <8 x float>* %x0ptr 274 %res = call <16 x float> @llvm.x86.avx512.mask.broadcastf32x8.512(<8 x float> %x0, <16 x float> %x2, i16 %mask) 275 ret <16 x float> %res 276} 277 278declare <8 x double> @llvm.x86.avx512.mask.broadcastf64x2.512(<2 x double>, <8 x double>, i8) 279 280define <8 x double>@test_int_x86_avx512_mask_broadcastf64x2_512(<2 x double> %x0, <8 x double> %x2, i8 %mask) { 281; X86-LABEL: test_int_x86_avx512_mask_broadcastf64x2_512: 282; X86: # %bb.0: 283; X86-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 284; X86-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0 # encoding: [0xc4,0xe3,0x7d,0x18,0xc0,0x01] 285; X86-NEXT: vinsertf64x4 $1, %ymm0, %zmm0, %zmm2 # encoding: [0x62,0xf3,0xfd,0x48,0x1a,0xd0,0x01] 286; X86-NEXT: kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04] 287; X86-NEXT: vinsertf64x4 $1, %ymm0, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x1a,0xc8,0x01] 288; X86-NEXT: vaddpd %zmm1, %zmm2, %zmm1 # encoding: [0x62,0xf1,0xed,0x48,0x58,0xc9] 289; X86-NEXT: vinsertf64x4 $1, %ymm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0x1a,0xc0,0x01] 290; X86-NEXT: vaddpd %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x58,0xc1] 291; X86-NEXT: retl # encoding: [0xc3] 292; 293; X64-LABEL: test_int_x86_avx512_mask_broadcastf64x2_512: 294; X64: # %bb.0: 295; X64-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 296; X64-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0 # encoding: [0xc4,0xe3,0x7d,0x18,0xc0,0x01] 297; X64-NEXT: vinsertf64x4 $1, %ymm0, %zmm0, %zmm2 # encoding: [0x62,0xf3,0xfd,0x48,0x1a,0xd0,0x01] 298; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf] 299; X64-NEXT: vinsertf64x4 $1, %ymm0, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x1a,0xc8,0x01] 300; X64-NEXT: vaddpd %zmm1, %zmm2, %zmm1 # encoding: [0x62,0xf1,0xed,0x48,0x58,0xc9] 301; X64-NEXT: vinsertf64x4 $1, %ymm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0x1a,0xc0,0x01] 302; X64-NEXT: vaddpd %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x58,0xc1] 303; X64-NEXT: retq # encoding: [0xc3] 304 305 %res1 = call <8 x double> @llvm.x86.avx512.mask.broadcastf64x2.512(<2 x double> %x0, <8 x double> %x2, i8 -1) 306 %res2 = call <8 x double> @llvm.x86.avx512.mask.broadcastf64x2.512(<2 x double> %x0, <8 x double> %x2, i8 %mask) 307 %res3 = call <8 x double> @llvm.x86.avx512.mask.broadcastf64x2.512(<2 x double> %x0, <8 x double> zeroinitializer, i8 %mask) 308 %res4 = fadd <8 x double> %res1, %res2 309 %res5 = fadd <8 x double> %res3, %res4 310 ret <8 x double> %res5 311} 312 313define <8 x double>@test_int_x86_avx512_mask_broadcastf64x2_512_load(<2 x double>* %x0ptr, <8 x double> %x2, i8 %mask) { 314; X86-LABEL: test_int_x86_avx512_mask_broadcastf64x2_512_load: 315; X86: # %bb.0: 316; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04] 317; X86-NEXT: kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08] 318; X86-NEXT: vbroadcastf64x2 (%eax), %zmm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x1a,0x00] 319; X86-NEXT: # zmm0 {%k1} = mem[0,1,0,1,0,1,0,1] 320; X86-NEXT: retl # encoding: [0xc3] 321; 322; X64-LABEL: test_int_x86_avx512_mask_broadcastf64x2_512_load: 323; X64: # %bb.0: 324; X64-NEXT: kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce] 325; X64-NEXT: vbroadcastf64x2 (%rdi), %zmm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x1a,0x07] 326; X64-NEXT: # zmm0 {%k1} = mem[0,1,0,1,0,1,0,1] 327; X64-NEXT: retq # encoding: [0xc3] 328 329 %x0 = load <2 x double>, <2 x double>* %x0ptr 330 %res = call <8 x double> @llvm.x86.avx512.mask.broadcastf64x2.512(<2 x double> %x0, <8 x double> %x2, i8 %mask) 331 ret <8 x double> %res 332} 333 334declare <16 x i32> @llvm.x86.avx512.mask.broadcasti32x8.512(<8 x i32>, <16 x i32>, i16) 335 336define <16 x i32>@test_int_x86_avx512_mask_broadcasti32x8_512(<8 x i32> %x0, <16 x i32> %x2, i16 %mask) { 337; X86-LABEL: test_int_x86_avx512_mask_broadcasti32x8_512: 338; X86: # %bb.0: 339; X86-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0 340; X86-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm2 # encoding: [0x62,0xf3,0xfd,0x48,0x3a,0xd0,0x01] 341; X86-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04] 342; X86-NEXT: vinserti32x8 $1, %ymm0, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x3a,0xc8,0x01] 343; X86-NEXT: vinserti32x8 $1, %ymm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x3a,0xc0,0x01] 344; X86-NEXT: vpaddd %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0x75,0x48,0xfe,0xc0] 345; X86-NEXT: vpaddd %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0x6d,0x48,0xfe,0xc0] 346; X86-NEXT: retl # encoding: [0xc3] 347; 348; X64-LABEL: test_int_x86_avx512_mask_broadcasti32x8_512: 349; X64: # %bb.0: 350; X64-NEXT: # kill: def $ymm0 killed $ymm0 def $zmm0 351; X64-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm2 # encoding: [0x62,0xf3,0xfd,0x48,0x3a,0xd0,0x01] 352; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf] 353; X64-NEXT: vinserti32x8 $1, %ymm0, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x3a,0xc8,0x01] 354; X64-NEXT: vinserti32x8 $1, %ymm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x3a,0xc0,0x01] 355; X64-NEXT: vpaddd %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0x75,0x48,0xfe,0xc0] 356; X64-NEXT: vpaddd %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0x6d,0x48,0xfe,0xc0] 357; X64-NEXT: retq # encoding: [0xc3] 358 359 %res1 = call <16 x i32> @llvm.x86.avx512.mask.broadcasti32x8.512(<8 x i32> %x0, <16 x i32> %x2, i16 -1) 360 %res2 = call <16 x i32> @llvm.x86.avx512.mask.broadcasti32x8.512(<8 x i32> %x0, <16 x i32> %x2, i16 %mask) 361 %res3 = call <16 x i32> @llvm.x86.avx512.mask.broadcasti32x8.512(<8 x i32> %x0, <16 x i32> zeroinitializer, i16 %mask) 362 %res4 = add <16 x i32> %res1, %res2 363 %res5 = add <16 x i32> %res3, %res4 364 ret <16 x i32> %res5 365} 366 367define <16 x i32>@test_int_x86_avx512_mask_broadcasti32x8_512_load(<8 x i32>* %x0ptr, <16 x i32> %x2, i16 %mask) { 368; X86-LABEL: test_int_x86_avx512_mask_broadcasti32x8_512_load: 369; X86: # %bb.0: 370; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04] 371; X86-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08] 372; X86-NEXT: vbroadcasti32x8 (%eax), %zmm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0x5b,0x00] 373; X86-NEXT: # zmm0 {%k1} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7] 374; X86-NEXT: retl # encoding: [0xc3] 375; 376; X64-LABEL: test_int_x86_avx512_mask_broadcasti32x8_512_load: 377; X64: # %bb.0: 378; X64-NEXT: kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce] 379; X64-NEXT: vbroadcasti32x8 (%rdi), %zmm0 {%k1} # encoding: [0x62,0xf2,0x7d,0x49,0x5b,0x07] 380; X64-NEXT: # zmm0 {%k1} = mem[0,1,2,3,4,5,6,7,0,1,2,3,4,5,6,7] 381; X64-NEXT: retq # encoding: [0xc3] 382 383 %x0 = load <8 x i32>, <8 x i32>* %x0ptr 384 %res = call <16 x i32> @llvm.x86.avx512.mask.broadcasti32x8.512(<8 x i32> %x0, <16 x i32> %x2, i16 %mask) 385 ret <16 x i32> %res 386} 387 388declare <8 x i64> @llvm.x86.avx512.mask.broadcasti64x2.512(<2 x i64>, <8 x i64>, i8) 389 390define <8 x i64>@test_int_x86_avx512_mask_broadcasti64x2_512(<2 x i64> %x0, <8 x i64> %x2, i8 %mask) { 391; X86-LABEL: test_int_x86_avx512_mask_broadcasti64x2_512: 392; X86: # %bb.0: 393; X86-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 394; X86-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0 # encoding: [0xc4,0xe3,0x7d,0x38,0xc0,0x01] 395; X86-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm2 # encoding: [0x62,0xf3,0xfd,0x48,0x3a,0xd0,0x01] 396; X86-NEXT: kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04] 397; X86-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x3a,0xc8,0x01] 398; X86-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0x3a,0xc0,0x01] 399; X86-NEXT: vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0] 400; X86-NEXT: vpaddq %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0xed,0x48,0xd4,0xc0] 401; X86-NEXT: retl # encoding: [0xc3] 402; 403; X64-LABEL: test_int_x86_avx512_mask_broadcasti64x2_512: 404; X64: # %bb.0: 405; X64-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 406; X64-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0 # encoding: [0xc4,0xe3,0x7d,0x38,0xc0,0x01] 407; X64-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm2 # encoding: [0x62,0xf3,0xfd,0x48,0x3a,0xd0,0x01] 408; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf] 409; X64-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x3a,0xc8,0x01] 410; X64-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xc9,0x3a,0xc0,0x01] 411; X64-NEXT: vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0] 412; X64-NEXT: vpaddq %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0xed,0x48,0xd4,0xc0] 413; X64-NEXT: retq # encoding: [0xc3] 414 415 %res1 = call <8 x i64> @llvm.x86.avx512.mask.broadcasti64x2.512(<2 x i64> %x0, <8 x i64> %x2, i8 -1) 416 %res2 = call <8 x i64> @llvm.x86.avx512.mask.broadcasti64x2.512(<2 x i64> %x0, <8 x i64> %x2, i8 %mask) 417 %res3 = call <8 x i64> @llvm.x86.avx512.mask.broadcasti64x2.512(<2 x i64> %x0, <8 x i64> zeroinitializer, i8 %mask) 418 %res4 = add <8 x i64> %res1, %res2 419 %res5 = add <8 x i64> %res3, %res4 420 ret <8 x i64> %res5 421} 422 423define <8 x i64>@test_int_x86_avx512_mask_broadcasti64x2_512_load(<2 x i64>* %x0ptr, <8 x i64> %x2, i8 %mask) { 424; X86-LABEL: test_int_x86_avx512_mask_broadcasti64x2_512_load: 425; X86: # %bb.0: 426; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04] 427; X86-NEXT: kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08] 428; X86-NEXT: vbroadcasti64x2 (%eax), %zmm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x5a,0x00] 429; X86-NEXT: # zmm0 {%k1} = mem[0,1,0,1,0,1,0,1] 430; X86-NEXT: retl # encoding: [0xc3] 431; 432; X64-LABEL: test_int_x86_avx512_mask_broadcasti64x2_512_load: 433; X64: # %bb.0: 434; X64-NEXT: kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce] 435; X64-NEXT: vbroadcasti64x2 (%rdi), %zmm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x5a,0x07] 436; X64-NEXT: # zmm0 {%k1} = mem[0,1,0,1,0,1,0,1] 437; X64-NEXT: retq # encoding: [0xc3] 438 439 %x0 = load <2 x i64>, <2 x i64>* %x0ptr 440 %res = call <8 x i64> @llvm.x86.avx512.mask.broadcasti64x2.512(<2 x i64> %x0, <8 x i64> %x2, i8 %mask) 441 ret <8 x i64> %res 442} 443 444declare <16 x float> @llvm.x86.avx512.mask.broadcastf32x2.512(<4 x float>, <16 x float>, i16) 445 446define <16 x float>@test_int_x86_avx512_mask_broadcastf32x2_512(<4 x float> %x0, <16 x float> %x2, i16 %x3) { 447; X86-LABEL: test_int_x86_avx512_mask_broadcastf32x2_512: 448; X86: # %bb.0: 449; X86-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 450; X86-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0 # encoding: [0xc4,0xe3,0x7d,0x18,0xc0,0x01] 451; X86-NEXT: vinsertf64x4 $1, %ymm0, %zmm0, %zmm2 # encoding: [0x62,0xf3,0xfd,0x48,0x1a,0xd0,0x01] 452; X86-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04] 453; X86-NEXT: vinsertf32x8 $1, %ymm0, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x1a,0xc8,0x01] 454; X86-NEXT: vinsertf32x8 $1, %ymm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x1a,0xc0,0x01] 455; X86-NEXT: vaddps %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0x74,0x48,0x58,0xc0] 456; X86-NEXT: vaddps %zmm2, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x58,0xc2] 457; X86-NEXT: retl # encoding: [0xc3] 458; 459; X64-LABEL: test_int_x86_avx512_mask_broadcastf32x2_512: 460; X64: # %bb.0: 461; X64-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 462; X64-NEXT: vinsertf128 $1, %xmm0, %ymm0, %ymm0 # encoding: [0xc4,0xe3,0x7d,0x18,0xc0,0x01] 463; X64-NEXT: vinsertf64x4 $1, %ymm0, %zmm0, %zmm2 # encoding: [0x62,0xf3,0xfd,0x48,0x1a,0xd0,0x01] 464; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf] 465; X64-NEXT: vinsertf32x8 $1, %ymm0, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x1a,0xc8,0x01] 466; X64-NEXT: vinsertf32x8 $1, %ymm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x1a,0xc0,0x01] 467; X64-NEXT: vaddps %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0x74,0x48,0x58,0xc0] 468; X64-NEXT: vaddps %zmm2, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x58,0xc2] 469; X64-NEXT: retq # encoding: [0xc3] 470 %res = call <16 x float> @llvm.x86.avx512.mask.broadcastf32x2.512(<4 x float> %x0, <16 x float> %x2, i16 %x3) 471 %res1 = call <16 x float> @llvm.x86.avx512.mask.broadcastf32x2.512(<4 x float> %x0, <16 x float> zeroinitializer, i16 %x3) 472 %res2 = call <16 x float> @llvm.x86.avx512.mask.broadcastf32x2.512(<4 x float> %x0, <16 x float> %x2, i16 -1) 473 %res3 = fadd <16 x float> %res, %res1 474 %res4 = fadd <16 x float> %res3, %res2 475 ret <16 x float> %res4 476} 477 478declare <16 x i32> @llvm.x86.avx512.mask.broadcasti32x2.512(<4 x i32>, <16 x i32>, i16) 479 480define <16 x i32>@test_int_x86_avx512_mask_broadcasti32x2_512(<4 x i32> %x0, <16 x i32> %x2, i16 %x3) { 481; X86-LABEL: test_int_x86_avx512_mask_broadcasti32x2_512: 482; X86: # %bb.0: 483; X86-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 484; X86-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0 # encoding: [0xc4,0xe3,0x7d,0x38,0xc0,0x01] 485; X86-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm2 # encoding: [0x62,0xf3,0xfd,0x48,0x3a,0xd0,0x01] 486; X86-NEXT: kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04] 487; X86-NEXT: vinserti32x8 $1, %ymm0, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x3a,0xc8,0x01] 488; X86-NEXT: vinserti32x8 $1, %ymm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x3a,0xc0,0x01] 489; X86-NEXT: vpaddd %zmm2, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7d,0x48,0xfe,0xc2] 490; X86-NEXT: vpaddd %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0x75,0x48,0xfe,0xc0] 491; X86-NEXT: retl # encoding: [0xc3] 492; 493; X64-LABEL: test_int_x86_avx512_mask_broadcasti32x2_512: 494; X64: # %bb.0: 495; X64-NEXT: # kill: def $xmm0 killed $xmm0 def $ymm0 496; X64-NEXT: vinserti128 $1, %xmm0, %ymm0, %ymm0 # encoding: [0xc4,0xe3,0x7d,0x38,0xc0,0x01] 497; X64-NEXT: vinserti64x4 $1, %ymm0, %zmm0, %zmm2 # encoding: [0x62,0xf3,0xfd,0x48,0x3a,0xd0,0x01] 498; X64-NEXT: kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf] 499; X64-NEXT: vinserti32x8 $1, %ymm0, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x3a,0xc8,0x01] 500; X64-NEXT: vinserti32x8 $1, %ymm0, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xc9,0x3a,0xc0,0x01] 501; X64-NEXT: vpaddd %zmm2, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7d,0x48,0xfe,0xc2] 502; X64-NEXT: vpaddd %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0x75,0x48,0xfe,0xc0] 503; X64-NEXT: retq # encoding: [0xc3] 504 %res = call <16 x i32> @llvm.x86.avx512.mask.broadcasti32x2.512(<4 x i32> %x0, <16 x i32> %x2, i16 %x3) 505 %res1 = call <16 x i32> @llvm.x86.avx512.mask.broadcasti32x2.512(<4 x i32> %x0, <16 x i32> zeroinitializer, i16 %x3) 506 %res2 = call <16 x i32> @llvm.x86.avx512.mask.broadcasti32x2.512(<4 x i32> %x0, <16 x i32> %x2, i16 -1) 507 %res3 = add <16 x i32> %res, %res1 508 %res4 = add <16 x i32> %res3, %res2 509 ret <16 x i32> %res4 510} 511 512declare i16 @llvm.x86.avx512.cvtd2mask.512(<16 x i32>) 513 514define i16@test_int_x86_avx512_cvtd2mask_512(<16 x i32> %x0) { 515; CHECK-LABEL: test_int_x86_avx512_cvtd2mask_512: 516; CHECK: # %bb.0: 517; CHECK-NEXT: vpmovd2m %zmm0, %k0 # encoding: [0x62,0xf2,0x7e,0x48,0x39,0xc0] 518; CHECK-NEXT: kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0] 519; CHECK-NEXT: # kill: def $ax killed $ax killed $eax 520; CHECK-NEXT: vzeroupper # encoding: [0xc5,0xf8,0x77] 521; CHECK-NEXT: ret{{[l|q]}} # encoding: [0xc3] 522 %res = call i16 @llvm.x86.avx512.cvtd2mask.512(<16 x i32> %x0) 523 ret i16 %res 524} 525 526declare i8 @llvm.x86.avx512.cvtq2mask.512(<8 x i64>) 527 528define i8@test_int_x86_avx512_cvtq2mask_512(<8 x i64> %x0) { 529; CHECK-LABEL: test_int_x86_avx512_cvtq2mask_512: 530; CHECK: # %bb.0: 531; CHECK-NEXT: vpmovq2m %zmm0, %k0 # encoding: [0x62,0xf2,0xfe,0x48,0x39,0xc0] 532; CHECK-NEXT: kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0] 533; CHECK-NEXT: # kill: def $al killed $al killed $eax 534; CHECK-NEXT: vzeroupper # encoding: [0xc5,0xf8,0x77] 535; CHECK-NEXT: ret{{[l|q]}} # encoding: [0xc3] 536 %res = call i8 @llvm.x86.avx512.cvtq2mask.512(<8 x i64> %x0) 537 ret i8 %res 538} 539 540declare i8 @llvm.x86.avx512.mask.fpclass.pd.512(<8 x double>, i32, i8) 541 542define i8 @test_int_x86_avx512_mask_fpclass_pd_512(<8 x double> %x0) { 543; CHECK-LABEL: test_int_x86_avx512_mask_fpclass_pd_512: 544; CHECK: # %bb.0: 545; CHECK-NEXT: vfpclasspd $2, %zmm0, %k1 # encoding: [0x62,0xf3,0xfd,0x48,0x66,0xc8,0x02] 546; CHECK-NEXT: vfpclasspd $4, %zmm0, %k0 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x66,0xc0,0x04] 547; CHECK-NEXT: kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0] 548; CHECK-NEXT: # kill: def $al killed $al killed $eax 549; CHECK-NEXT: vzeroupper # encoding: [0xc5,0xf8,0x77] 550; CHECK-NEXT: ret{{[l|q]}} # encoding: [0xc3] 551 %res = call i8 @llvm.x86.avx512.mask.fpclass.pd.512(<8 x double> %x0, i32 4, i8 -1) 552 %res1 = call i8 @llvm.x86.avx512.mask.fpclass.pd.512(<8 x double> %x0, i32 2, i8 %res) 553 ret i8 %res1 554} 555declare i16 @llvm.x86.avx512.mask.fpclass.ps.512(<16 x float>, i32, i16) 556 557define i16@test_int_x86_avx512_mask_fpclass_ps_512(<16 x float> %x0) { 558; CHECK-LABEL: test_int_x86_avx512_mask_fpclass_ps_512: 559; CHECK: # %bb.0: 560; CHECK-NEXT: vfpclassps $2, %zmm0, %k1 # encoding: [0x62,0xf3,0x7d,0x48,0x66,0xc8,0x02] 561; CHECK-NEXT: vfpclassps $4, %zmm0, %k0 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x66,0xc0,0x04] 562; CHECK-NEXT: kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0] 563; CHECK-NEXT: # kill: def $ax killed $ax killed $eax 564; CHECK-NEXT: vzeroupper # encoding: [0xc5,0xf8,0x77] 565; CHECK-NEXT: ret{{[l|q]}} # encoding: [0xc3] 566 %res = call i16 @llvm.x86.avx512.mask.fpclass.ps.512(<16 x float> %x0, i32 4, i16 -1) 567 %res1 = call i16 @llvm.x86.avx512.mask.fpclass.ps.512(<16 x float> %x0, i32 2, i16 %res) 568 ret i16 %res1 569} 570