523 lines · plain
1; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py2; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avxvnniint8 --show-mc-encoding | FileCheck %s --check-prefixes=X863; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avxvnniint8 --show-mc-encoding | FileCheck %s --check-prefixes=X644; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx10.2 --show-mc-encoding | FileCheck %s --check-prefixes=AVX10-X865; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx10.2 --show-mc-encoding | FileCheck %s --check-prefixes=AVX10-X646 7 8declare <4 x i32> @llvm.x86.avx2.vpdpbssd.128(<4 x i32>, <16 x i8>, <16 x i8>)9 10define <4 x i32>@test_int_x86_avx2_vpdpbssd_128(<4 x i32> %x0, <16 x i8> %x1, ptr %x2p, <16 x i8> %x4) {11; X86-LABEL: test_int_x86_avx2_vpdpbssd_128:12; X86: # %bb.0:13; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]14; X86-NEXT: vmovaps %xmm0, %xmm3 # encoding: [0xc5,0xf8,0x28,0xd8]15; X86-NEXT: vpdpbssd (%eax), %xmm1, %xmm3 # encoding: [0xc4,0xe2,0x73,0x50,0x18]16; X86-NEXT: vpdpbssd %xmm2, %xmm1, %xmm0 # encoding: [0xc4,0xe2,0x73,0x50,0xc2]17; X86-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # encoding: [0xc5,0xe1,0xfe,0xc0]18; X86-NEXT: retl # encoding: [0xc3]19;20; X64-LABEL: test_int_x86_avx2_vpdpbssd_128:21; X64: # %bb.0:22; X64-NEXT: vmovaps %xmm0, %xmm3 # encoding: [0xc5,0xf8,0x28,0xd8]23; X64-NEXT: vpdpbssd (%rdi), %xmm1, %xmm3 # encoding: [0xc4,0xe2,0x73,0x50,0x1f]24; X64-NEXT: vpdpbssd %xmm2, %xmm1, %xmm0 # encoding: [0xc4,0xe2,0x73,0x50,0xc2]25; X64-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # encoding: [0xc5,0xe1,0xfe,0xc0]26; X64-NEXT: retq # encoding: [0xc3]27;28; AVX10-X86-LABEL: test_int_x86_avx2_vpdpbssd_128:29; AVX10-X86: # %bb.0:30; AVX10-X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]31; AVX10-X86-NEXT: vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]32; AVX10-X86-NEXT: vpdpbssd (%eax), %xmm1, %xmm3 # encoding: [0x62,0xf2,0x77,0x08,0x50,0x18]33; AVX10-X86-NEXT: vpdpbssd %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf2,0x77,0x08,0x50,0xc2]34; AVX10-X86-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe1,0xfe,0xc0]35; AVX10-X86-NEXT: retl # encoding: [0xc3]36;37; AVX10-X64-LABEL: test_int_x86_avx2_vpdpbssd_128:38; AVX10-X64: # %bb.0:39; AVX10-X64-NEXT: vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]40; AVX10-X64-NEXT: vpdpbssd (%rdi), %xmm1, %xmm3 # encoding: [0x62,0xf2,0x77,0x08,0x50,0x1f]41; AVX10-X64-NEXT: vpdpbssd %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf2,0x77,0x08,0x50,0xc2]42; AVX10-X64-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe1,0xfe,0xc0]43; AVX10-X64-NEXT: retq # encoding: [0xc3]44 %x2 = load <16 x i8>, ptr %x2p45 %1 = call <4 x i32> @llvm.x86.avx2.vpdpbssd.128(<4 x i32> %x0, <16 x i8> %x1, <16 x i8> %x2)46 %2 = call <4 x i32> @llvm.x86.avx2.vpdpbssd.128(<4 x i32> %x0, <16 x i8> %x1, <16 x i8> %x4)47 %res = add <4 x i32> %1, %248 ret <4 x i32> %res49}50 51declare <4 x i32> @llvm.x86.avx2.vpdpbssds.128(<4 x i32>, <16 x i8>, <16 x i8>)52 53define <4 x i32>@test_int_x86_avx2_vpdpbssds_128(<4 x i32> %x0, <16 x i8> %x1, ptr %x2p, <16 x i8> %x4) {54; X86-LABEL: test_int_x86_avx2_vpdpbssds_128:55; X86: # %bb.0:56; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]57; X86-NEXT: vmovaps %xmm0, %xmm3 # encoding: [0xc5,0xf8,0x28,0xd8]58; X86-NEXT: vpdpbssds (%eax), %xmm1, %xmm3 # encoding: [0xc4,0xe2,0x73,0x51,0x18]59; X86-NEXT: vpdpbssds %xmm2, %xmm1, %xmm0 # encoding: [0xc4,0xe2,0x73,0x51,0xc2]60; X86-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # encoding: [0xc5,0xe1,0xfe,0xc0]61; X86-NEXT: retl # encoding: [0xc3]62;63; X64-LABEL: test_int_x86_avx2_vpdpbssds_128:64; X64: # %bb.0:65; X64-NEXT: vmovaps %xmm0, %xmm3 # encoding: [0xc5,0xf8,0x28,0xd8]66; X64-NEXT: vpdpbssds (%rdi), %xmm1, %xmm3 # encoding: [0xc4,0xe2,0x73,0x51,0x1f]67; X64-NEXT: vpdpbssds %xmm2, %xmm1, %xmm0 # encoding: [0xc4,0xe2,0x73,0x51,0xc2]68; X64-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # encoding: [0xc5,0xe1,0xfe,0xc0]69; X64-NEXT: retq # encoding: [0xc3]70;71; AVX10-X86-LABEL: test_int_x86_avx2_vpdpbssds_128:72; AVX10-X86: # %bb.0:73; AVX10-X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]74; AVX10-X86-NEXT: vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]75; AVX10-X86-NEXT: vpdpbssds (%eax), %xmm1, %xmm3 # encoding: [0x62,0xf2,0x77,0x08,0x51,0x18]76; AVX10-X86-NEXT: vpdpbssds %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf2,0x77,0x08,0x51,0xc2]77; AVX10-X86-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe1,0xfe,0xc0]78; AVX10-X86-NEXT: retl # encoding: [0xc3]79;80; AVX10-X64-LABEL: test_int_x86_avx2_vpdpbssds_128:81; AVX10-X64: # %bb.0:82; AVX10-X64-NEXT: vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]83; AVX10-X64-NEXT: vpdpbssds (%rdi), %xmm1, %xmm3 # encoding: [0x62,0xf2,0x77,0x08,0x51,0x1f]84; AVX10-X64-NEXT: vpdpbssds %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf2,0x77,0x08,0x51,0xc2]85; AVX10-X64-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe1,0xfe,0xc0]86; AVX10-X64-NEXT: retq # encoding: [0xc3]87 %x2 = load <16 x i8>, ptr %x2p88 %1 = call <4 x i32> @llvm.x86.avx2.vpdpbssds.128(<4 x i32> %x0, <16 x i8> %x1, <16 x i8> %x2)89 %2 = call <4 x i32> @llvm.x86.avx2.vpdpbssds.128(<4 x i32> %x0, <16 x i8> %x1, <16 x i8> %x4)90 %res = add <4 x i32> %1, %291 ret <4 x i32> %res92}93 94declare <8 x i32> @llvm.x86.avx2.vpdpbssd.256(<8 x i32>, <32 x i8>, <32 x i8>)95 96define <8 x i32>@test_int_x86_avx2_vpdpbssd_256(<8 x i32> %x0, <32 x i8> %x1, ptr %x2p, <32 x i8> %x4) {97; X86-LABEL: test_int_x86_avx2_vpdpbssd_256:98; X86: # %bb.0:99; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]100; X86-NEXT: vmovaps %ymm0, %ymm3 # encoding: [0xc5,0xfc,0x28,0xd8]101; X86-NEXT: vpdpbssd (%eax), %ymm1, %ymm3 # encoding: [0xc4,0xe2,0x77,0x50,0x18]102; X86-NEXT: vpdpbssd %ymm2, %ymm1, %ymm0 # encoding: [0xc4,0xe2,0x77,0x50,0xc2]103; X86-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # encoding: [0xc5,0xe5,0xfe,0xc0]104; X86-NEXT: retl # encoding: [0xc3]105;106; X64-LABEL: test_int_x86_avx2_vpdpbssd_256:107; X64: # %bb.0:108; X64-NEXT: vmovaps %ymm0, %ymm3 # encoding: [0xc5,0xfc,0x28,0xd8]109; X64-NEXT: vpdpbssd (%rdi), %ymm1, %ymm3 # encoding: [0xc4,0xe2,0x77,0x50,0x1f]110; X64-NEXT: vpdpbssd %ymm2, %ymm1, %ymm0 # encoding: [0xc4,0xe2,0x77,0x50,0xc2]111; X64-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # encoding: [0xc5,0xe5,0xfe,0xc0]112; X64-NEXT: retq # encoding: [0xc3]113;114; AVX10-X86-LABEL: test_int_x86_avx2_vpdpbssd_256:115; AVX10-X86: # %bb.0:116; AVX10-X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]117; AVX10-X86-NEXT: vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]118; AVX10-X86-NEXT: vpdpbssd (%eax), %ymm1, %ymm3 # encoding: [0x62,0xf2,0x77,0x28,0x50,0x18]119; AVX10-X86-NEXT: vpdpbssd %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x77,0x28,0x50,0xc2]120; AVX10-X86-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xe5,0xfe,0xc0]121; AVX10-X86-NEXT: retl # encoding: [0xc3]122;123; AVX10-X64-LABEL: test_int_x86_avx2_vpdpbssd_256:124; AVX10-X64: # %bb.0:125; AVX10-X64-NEXT: vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]126; AVX10-X64-NEXT: vpdpbssd (%rdi), %ymm1, %ymm3 # encoding: [0x62,0xf2,0x77,0x28,0x50,0x1f]127; AVX10-X64-NEXT: vpdpbssd %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x77,0x28,0x50,0xc2]128; AVX10-X64-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xe5,0xfe,0xc0]129; AVX10-X64-NEXT: retq # encoding: [0xc3]130 %x2 = load <32 x i8>, ptr %x2p131 %1 = call <8 x i32> @llvm.x86.avx2.vpdpbssd.256(<8 x i32> %x0, <32 x i8> %x1, <32 x i8> %x2)132 %2 = call <8 x i32> @llvm.x86.avx2.vpdpbssd.256(<8 x i32> %x0, <32 x i8> %x1, <32 x i8> %x4)133 %res = add <8 x i32> %1, %2134 ret <8 x i32> %res135}136 137declare <8 x i32> @llvm.x86.avx2.vpdpbssds.256(<8 x i32>, <32 x i8>, <32 x i8>)138 139define <8 x i32>@test_int_x86_avx2_vpdpbssds_256(<8 x i32> %x0, <32 x i8> %x1, ptr %x2p, <32 x i8> %x4) {140; X86-LABEL: test_int_x86_avx2_vpdpbssds_256:141; X86: # %bb.0:142; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]143; X86-NEXT: vmovaps %ymm0, %ymm3 # encoding: [0xc5,0xfc,0x28,0xd8]144; X86-NEXT: vpdpbssds (%eax), %ymm1, %ymm3 # encoding: [0xc4,0xe2,0x77,0x51,0x18]145; X86-NEXT: vpdpbssds %ymm2, %ymm1, %ymm0 # encoding: [0xc4,0xe2,0x77,0x51,0xc2]146; X86-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # encoding: [0xc5,0xe5,0xfe,0xc0]147; X86-NEXT: retl # encoding: [0xc3]148;149; X64-LABEL: test_int_x86_avx2_vpdpbssds_256:150; X64: # %bb.0:151; X64-NEXT: vmovaps %ymm0, %ymm3 # encoding: [0xc5,0xfc,0x28,0xd8]152; X64-NEXT: vpdpbssds (%rdi), %ymm1, %ymm3 # encoding: [0xc4,0xe2,0x77,0x51,0x1f]153; X64-NEXT: vpdpbssds %ymm2, %ymm1, %ymm0 # encoding: [0xc4,0xe2,0x77,0x51,0xc2]154; X64-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # encoding: [0xc5,0xe5,0xfe,0xc0]155; X64-NEXT: retq # encoding: [0xc3]156;157; AVX10-X86-LABEL: test_int_x86_avx2_vpdpbssds_256:158; AVX10-X86: # %bb.0:159; AVX10-X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]160; AVX10-X86-NEXT: vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]161; AVX10-X86-NEXT: vpdpbssds (%eax), %ymm1, %ymm3 # encoding: [0x62,0xf2,0x77,0x28,0x51,0x18]162; AVX10-X86-NEXT: vpdpbssds %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x77,0x28,0x51,0xc2]163; AVX10-X86-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xe5,0xfe,0xc0]164; AVX10-X86-NEXT: retl # encoding: [0xc3]165;166; AVX10-X64-LABEL: test_int_x86_avx2_vpdpbssds_256:167; AVX10-X64: # %bb.0:168; AVX10-X64-NEXT: vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]169; AVX10-X64-NEXT: vpdpbssds (%rdi), %ymm1, %ymm3 # encoding: [0x62,0xf2,0x77,0x28,0x51,0x1f]170; AVX10-X64-NEXT: vpdpbssds %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x77,0x28,0x51,0xc2]171; AVX10-X64-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xe5,0xfe,0xc0]172; AVX10-X64-NEXT: retq # encoding: [0xc3]173 %x2 = load <32 x i8>, ptr %x2p174 %1 = call <8 x i32> @llvm.x86.avx2.vpdpbssds.256(<8 x i32> %x0, <32 x i8> %x1, <32 x i8> %x2)175 %2 = call <8 x i32> @llvm.x86.avx2.vpdpbssds.256(<8 x i32> %x0, <32 x i8> %x1, <32 x i8> %x4)176 %res = add <8 x i32> %1, %2177 ret <8 x i32> %res178}179 180declare <4 x i32> @llvm.x86.avx2.vpdpbsud.128(<4 x i32>, <16 x i8>, <16 x i8>)181 182define <4 x i32>@test_int_x86_avx2_vpdpbsud_128(<4 x i32> %x0, <16 x i8> %x1, ptr %x2p, <16 x i8> %x4) {183; X86-LABEL: test_int_x86_avx2_vpdpbsud_128:184; X86: # %bb.0:185; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]186; X86-NEXT: vmovaps %xmm0, %xmm3 # encoding: [0xc5,0xf8,0x28,0xd8]187; X86-NEXT: vpdpbsud (%eax), %xmm1, %xmm3 # encoding: [0xc4,0xe2,0x72,0x50,0x18]188; X86-NEXT: vpdpbsud %xmm2, %xmm1, %xmm0 # encoding: [0xc4,0xe2,0x72,0x50,0xc2]189; X86-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # encoding: [0xc5,0xe1,0xfe,0xc0]190; X86-NEXT: retl # encoding: [0xc3]191;192; X64-LABEL: test_int_x86_avx2_vpdpbsud_128:193; X64: # %bb.0:194; X64-NEXT: vmovaps %xmm0, %xmm3 # encoding: [0xc5,0xf8,0x28,0xd8]195; X64-NEXT: vpdpbsud (%rdi), %xmm1, %xmm3 # encoding: [0xc4,0xe2,0x72,0x50,0x1f]196; X64-NEXT: vpdpbsud %xmm2, %xmm1, %xmm0 # encoding: [0xc4,0xe2,0x72,0x50,0xc2]197; X64-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # encoding: [0xc5,0xe1,0xfe,0xc0]198; X64-NEXT: retq # encoding: [0xc3]199;200; AVX10-X86-LABEL: test_int_x86_avx2_vpdpbsud_128:201; AVX10-X86: # %bb.0:202; AVX10-X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]203; AVX10-X86-NEXT: vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]204; AVX10-X86-NEXT: vpdpbsud (%eax), %xmm1, %xmm3 # encoding: [0x62,0xf2,0x76,0x08,0x50,0x18]205; AVX10-X86-NEXT: vpdpbsud %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf2,0x76,0x08,0x50,0xc2]206; AVX10-X86-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe1,0xfe,0xc0]207; AVX10-X86-NEXT: retl # encoding: [0xc3]208;209; AVX10-X64-LABEL: test_int_x86_avx2_vpdpbsud_128:210; AVX10-X64: # %bb.0:211; AVX10-X64-NEXT: vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]212; AVX10-X64-NEXT: vpdpbsud (%rdi), %xmm1, %xmm3 # encoding: [0x62,0xf2,0x76,0x08,0x50,0x1f]213; AVX10-X64-NEXT: vpdpbsud %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf2,0x76,0x08,0x50,0xc2]214; AVX10-X64-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe1,0xfe,0xc0]215; AVX10-X64-NEXT: retq # encoding: [0xc3]216 %x2 = load <16 x i8>, ptr %x2p217 %1 = call <4 x i32> @llvm.x86.avx2.vpdpbsud.128(<4 x i32> %x0, <16 x i8> %x1, <16 x i8> %x2)218 %2 = call <4 x i32> @llvm.x86.avx2.vpdpbsud.128(<4 x i32> %x0, <16 x i8> %x1, <16 x i8> %x4)219 %res = add <4 x i32> %1, %2220 ret <4 x i32> %res221}222 223declare <4 x i32> @llvm.x86.avx2.vpdpbsuds.128(<4 x i32>, <16 x i8>, <16 x i8>)224 225define <4 x i32>@test_int_x86_avx2_vpdpbsuds_128(<4 x i32> %x0, <16 x i8> %x1, ptr %x2p, <16 x i8> %x4) {226; X86-LABEL: test_int_x86_avx2_vpdpbsuds_128:227; X86: # %bb.0:228; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]229; X86-NEXT: vmovaps %xmm0, %xmm3 # encoding: [0xc5,0xf8,0x28,0xd8]230; X86-NEXT: vpdpbsuds (%eax), %xmm1, %xmm3 # encoding: [0xc4,0xe2,0x72,0x51,0x18]231; X86-NEXT: vpdpbsuds %xmm2, %xmm1, %xmm0 # encoding: [0xc4,0xe2,0x72,0x51,0xc2]232; X86-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # encoding: [0xc5,0xe1,0xfe,0xc0]233; X86-NEXT: retl # encoding: [0xc3]234;235; X64-LABEL: test_int_x86_avx2_vpdpbsuds_128:236; X64: # %bb.0:237; X64-NEXT: vmovaps %xmm0, %xmm3 # encoding: [0xc5,0xf8,0x28,0xd8]238; X64-NEXT: vpdpbsuds (%rdi), %xmm1, %xmm3 # encoding: [0xc4,0xe2,0x72,0x51,0x1f]239; X64-NEXT: vpdpbsuds %xmm2, %xmm1, %xmm0 # encoding: [0xc4,0xe2,0x72,0x51,0xc2]240; X64-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # encoding: [0xc5,0xe1,0xfe,0xc0]241; X64-NEXT: retq # encoding: [0xc3]242;243; AVX10-X86-LABEL: test_int_x86_avx2_vpdpbsuds_128:244; AVX10-X86: # %bb.0:245; AVX10-X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]246; AVX10-X86-NEXT: vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]247; AVX10-X86-NEXT: vpdpbsuds (%eax), %xmm1, %xmm3 # encoding: [0x62,0xf2,0x76,0x08,0x51,0x18]248; AVX10-X86-NEXT: vpdpbsuds %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf2,0x76,0x08,0x51,0xc2]249; AVX10-X86-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe1,0xfe,0xc0]250; AVX10-X86-NEXT: retl # encoding: [0xc3]251;252; AVX10-X64-LABEL: test_int_x86_avx2_vpdpbsuds_128:253; AVX10-X64: # %bb.0:254; AVX10-X64-NEXT: vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]255; AVX10-X64-NEXT: vpdpbsuds (%rdi), %xmm1, %xmm3 # encoding: [0x62,0xf2,0x76,0x08,0x51,0x1f]256; AVX10-X64-NEXT: vpdpbsuds %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf2,0x76,0x08,0x51,0xc2]257; AVX10-X64-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe1,0xfe,0xc0]258; AVX10-X64-NEXT: retq # encoding: [0xc3]259 %x2 = load <16 x i8>, ptr %x2p260 %1 = call <4 x i32> @llvm.x86.avx2.vpdpbsuds.128(<4 x i32> %x0, <16 x i8> %x1, <16 x i8> %x2)261 %2 = call <4 x i32> @llvm.x86.avx2.vpdpbsuds.128(<4 x i32> %x0, <16 x i8> %x1, <16 x i8> %x4)262 %res = add <4 x i32> %1, %2263 ret <4 x i32> %res264}265 266declare <8 x i32> @llvm.x86.avx2.vpdpbsud.256(<8 x i32>, <32 x i8>, <32 x i8>)267 268define <8 x i32>@test_int_x86_avx2_vpdpbsud_256(<8 x i32> %x0, <32 x i8> %x1, ptr %x2p, <32 x i8> %x4) {269; X86-LABEL: test_int_x86_avx2_vpdpbsud_256:270; X86: # %bb.0:271; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]272; X86-NEXT: vmovaps %ymm0, %ymm3 # encoding: [0xc5,0xfc,0x28,0xd8]273; X86-NEXT: vpdpbsud (%eax), %ymm1, %ymm3 # encoding: [0xc4,0xe2,0x76,0x50,0x18]274; X86-NEXT: vpdpbsud %ymm2, %ymm1, %ymm0 # encoding: [0xc4,0xe2,0x76,0x50,0xc2]275; X86-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # encoding: [0xc5,0xe5,0xfe,0xc0]276; X86-NEXT: retl # encoding: [0xc3]277;278; X64-LABEL: test_int_x86_avx2_vpdpbsud_256:279; X64: # %bb.0:280; X64-NEXT: vmovaps %ymm0, %ymm3 # encoding: [0xc5,0xfc,0x28,0xd8]281; X64-NEXT: vpdpbsud (%rdi), %ymm1, %ymm3 # encoding: [0xc4,0xe2,0x76,0x50,0x1f]282; X64-NEXT: vpdpbsud %ymm2, %ymm1, %ymm0 # encoding: [0xc4,0xe2,0x76,0x50,0xc2]283; X64-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # encoding: [0xc5,0xe5,0xfe,0xc0]284; X64-NEXT: retq # encoding: [0xc3]285;286; AVX10-X86-LABEL: test_int_x86_avx2_vpdpbsud_256:287; AVX10-X86: # %bb.0:288; AVX10-X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]289; AVX10-X86-NEXT: vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]290; AVX10-X86-NEXT: vpdpbsud (%eax), %ymm1, %ymm3 # encoding: [0x62,0xf2,0x76,0x28,0x50,0x18]291; AVX10-X86-NEXT: vpdpbsud %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x76,0x28,0x50,0xc2]292; AVX10-X86-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xe5,0xfe,0xc0]293; AVX10-X86-NEXT: retl # encoding: [0xc3]294;295; AVX10-X64-LABEL: test_int_x86_avx2_vpdpbsud_256:296; AVX10-X64: # %bb.0:297; AVX10-X64-NEXT: vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]298; AVX10-X64-NEXT: vpdpbsud (%rdi), %ymm1, %ymm3 # encoding: [0x62,0xf2,0x76,0x28,0x50,0x1f]299; AVX10-X64-NEXT: vpdpbsud %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x76,0x28,0x50,0xc2]300; AVX10-X64-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xe5,0xfe,0xc0]301; AVX10-X64-NEXT: retq # encoding: [0xc3]302 %x2 = load <32 x i8>, ptr %x2p303 %1 = call <8 x i32> @llvm.x86.avx2.vpdpbsud.256(<8 x i32> %x0, <32 x i8> %x1, <32 x i8> %x2)304 %2 = call <8 x i32> @llvm.x86.avx2.vpdpbsud.256(<8 x i32> %x0, <32 x i8> %x1, <32 x i8> %x4)305 %res = add <8 x i32> %1, %2306 ret <8 x i32> %res307}308 309declare <8 x i32> @llvm.x86.avx2.vpdpbsuds.256(<8 x i32>, <32 x i8>, <32 x i8>)310 311define <8 x i32>@test_int_x86_avx2_vpdpbsuds_256(<8 x i32> %x0, <32 x i8> %x1, ptr %x2p, <32 x i8> %x4) {312; X86-LABEL: test_int_x86_avx2_vpdpbsuds_256:313; X86: # %bb.0:314; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]315; X86-NEXT: vmovaps %ymm0, %ymm3 # encoding: [0xc5,0xfc,0x28,0xd8]316; X86-NEXT: vpdpbsuds (%eax), %ymm1, %ymm3 # encoding: [0xc4,0xe2,0x76,0x51,0x18]317; X86-NEXT: vpdpbsuds %ymm2, %ymm1, %ymm0 # encoding: [0xc4,0xe2,0x76,0x51,0xc2]318; X86-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # encoding: [0xc5,0xe5,0xfe,0xc0]319; X86-NEXT: retl # encoding: [0xc3]320;321; X64-LABEL: test_int_x86_avx2_vpdpbsuds_256:322; X64: # %bb.0:323; X64-NEXT: vmovaps %ymm0, %ymm3 # encoding: [0xc5,0xfc,0x28,0xd8]324; X64-NEXT: vpdpbsuds (%rdi), %ymm1, %ymm3 # encoding: [0xc4,0xe2,0x76,0x51,0x1f]325; X64-NEXT: vpdpbsuds %ymm2, %ymm1, %ymm0 # encoding: [0xc4,0xe2,0x76,0x51,0xc2]326; X64-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # encoding: [0xc5,0xe5,0xfe,0xc0]327; X64-NEXT: retq # encoding: [0xc3]328;329; AVX10-X86-LABEL: test_int_x86_avx2_vpdpbsuds_256:330; AVX10-X86: # %bb.0:331; AVX10-X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]332; AVX10-X86-NEXT: vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]333; AVX10-X86-NEXT: vpdpbsuds (%eax), %ymm1, %ymm3 # encoding: [0x62,0xf2,0x76,0x28,0x51,0x18]334; AVX10-X86-NEXT: vpdpbsuds %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x76,0x28,0x51,0xc2]335; AVX10-X86-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xe5,0xfe,0xc0]336; AVX10-X86-NEXT: retl # encoding: [0xc3]337;338; AVX10-X64-LABEL: test_int_x86_avx2_vpdpbsuds_256:339; AVX10-X64: # %bb.0:340; AVX10-X64-NEXT: vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]341; AVX10-X64-NEXT: vpdpbsuds (%rdi), %ymm1, %ymm3 # encoding: [0x62,0xf2,0x76,0x28,0x51,0x1f]342; AVX10-X64-NEXT: vpdpbsuds %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x76,0x28,0x51,0xc2]343; AVX10-X64-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xe5,0xfe,0xc0]344; AVX10-X64-NEXT: retq # encoding: [0xc3]345 %x2 = load <32 x i8>, ptr %x2p346 %1 = call <8 x i32> @llvm.x86.avx2.vpdpbsuds.256(<8 x i32> %x0, <32 x i8> %x1, <32 x i8> %x2)347 %2 = call <8 x i32> @llvm.x86.avx2.vpdpbsuds.256(<8 x i32> %x0, <32 x i8> %x1, <32 x i8> %x4)348 %res = add <8 x i32> %1, %2349 ret <8 x i32> %res350}351 352declare <4 x i32> @llvm.x86.avx2.vpdpbuud.128(<4 x i32>, <16 x i8>, <16 x i8>)353 354define <4 x i32>@test_int_x86_avx2_vpdpbuud_128(<4 x i32> %x0, <16 x i8> %x1, ptr %x2p, <16 x i8> %x4) {355; X86-LABEL: test_int_x86_avx2_vpdpbuud_128:356; X86: # %bb.0:357; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]358; X86-NEXT: vmovaps %xmm0, %xmm3 # encoding: [0xc5,0xf8,0x28,0xd8]359; X86-NEXT: vpdpbuud (%eax), %xmm1, %xmm3 # encoding: [0xc4,0xe2,0x70,0x50,0x18]360; X86-NEXT: vpdpbuud %xmm2, %xmm1, %xmm0 # encoding: [0xc4,0xe2,0x70,0x50,0xc2]361; X86-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # encoding: [0xc5,0xe1,0xfe,0xc0]362; X86-NEXT: retl # encoding: [0xc3]363;364; X64-LABEL: test_int_x86_avx2_vpdpbuud_128:365; X64: # %bb.0:366; X64-NEXT: vmovaps %xmm0, %xmm3 # encoding: [0xc5,0xf8,0x28,0xd8]367; X64-NEXT: vpdpbuud (%rdi), %xmm1, %xmm3 # encoding: [0xc4,0xe2,0x70,0x50,0x1f]368; X64-NEXT: vpdpbuud %xmm2, %xmm1, %xmm0 # encoding: [0xc4,0xe2,0x70,0x50,0xc2]369; X64-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # encoding: [0xc5,0xe1,0xfe,0xc0]370; X64-NEXT: retq # encoding: [0xc3]371;372; AVX10-X86-LABEL: test_int_x86_avx2_vpdpbuud_128:373; AVX10-X86: # %bb.0:374; AVX10-X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]375; AVX10-X86-NEXT: vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]376; AVX10-X86-NEXT: vpdpbuud (%eax), %xmm1, %xmm3 # encoding: [0x62,0xf2,0x74,0x08,0x50,0x18]377; AVX10-X86-NEXT: vpdpbuud %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf2,0x74,0x08,0x50,0xc2]378; AVX10-X86-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe1,0xfe,0xc0]379; AVX10-X86-NEXT: retl # encoding: [0xc3]380;381; AVX10-X64-LABEL: test_int_x86_avx2_vpdpbuud_128:382; AVX10-X64: # %bb.0:383; AVX10-X64-NEXT: vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]384; AVX10-X64-NEXT: vpdpbuud (%rdi), %xmm1, %xmm3 # encoding: [0x62,0xf2,0x74,0x08,0x50,0x1f]385; AVX10-X64-NEXT: vpdpbuud %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf2,0x74,0x08,0x50,0xc2]386; AVX10-X64-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe1,0xfe,0xc0]387; AVX10-X64-NEXT: retq # encoding: [0xc3]388 %x2 = load <16 x i8>, ptr %x2p389 %1 = call <4 x i32> @llvm.x86.avx2.vpdpbuud.128(<4 x i32> %x0, <16 x i8> %x1, <16 x i8> %x2)390 %2 = call <4 x i32> @llvm.x86.avx2.vpdpbuud.128(<4 x i32> %x0, <16 x i8> %x1, <16 x i8> %x4)391 %res = add <4 x i32> %1, %2392 ret <4 x i32> %res393}394 395declare <4 x i32> @llvm.x86.avx2.vpdpbuuds.128(<4 x i32>, <16 x i8>, <16 x i8>)396 397define <4 x i32>@test_int_x86_avx2_vpdpbuuds_128(<4 x i32> %x0, <16 x i8> %x1, ptr %x2p, <16 x i8> %x4) {398; X86-LABEL: test_int_x86_avx2_vpdpbuuds_128:399; X86: # %bb.0:400; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]401; X86-NEXT: vmovaps %xmm0, %xmm3 # encoding: [0xc5,0xf8,0x28,0xd8]402; X86-NEXT: vpdpbuuds (%eax), %xmm1, %xmm3 # encoding: [0xc4,0xe2,0x70,0x51,0x18]403; X86-NEXT: vpdpbuuds %xmm2, %xmm1, %xmm0 # encoding: [0xc4,0xe2,0x70,0x51,0xc2]404; X86-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # encoding: [0xc5,0xe1,0xfe,0xc0]405; X86-NEXT: retl # encoding: [0xc3]406;407; X64-LABEL: test_int_x86_avx2_vpdpbuuds_128:408; X64: # %bb.0:409; X64-NEXT: vmovaps %xmm0, %xmm3 # encoding: [0xc5,0xf8,0x28,0xd8]410; X64-NEXT: vpdpbuuds (%rdi), %xmm1, %xmm3 # encoding: [0xc4,0xe2,0x70,0x51,0x1f]411; X64-NEXT: vpdpbuuds %xmm2, %xmm1, %xmm0 # encoding: [0xc4,0xe2,0x70,0x51,0xc2]412; X64-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # encoding: [0xc5,0xe1,0xfe,0xc0]413; X64-NEXT: retq # encoding: [0xc3]414;415; AVX10-X86-LABEL: test_int_x86_avx2_vpdpbuuds_128:416; AVX10-X86: # %bb.0:417; AVX10-X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]418; AVX10-X86-NEXT: vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]419; AVX10-X86-NEXT: vpdpbuuds (%eax), %xmm1, %xmm3 # encoding: [0x62,0xf2,0x74,0x08,0x51,0x18]420; AVX10-X86-NEXT: vpdpbuuds %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf2,0x74,0x08,0x51,0xc2]421; AVX10-X86-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe1,0xfe,0xc0]422; AVX10-X86-NEXT: retl # encoding: [0xc3]423;424; AVX10-X64-LABEL: test_int_x86_avx2_vpdpbuuds_128:425; AVX10-X64: # %bb.0:426; AVX10-X64-NEXT: vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]427; AVX10-X64-NEXT: vpdpbuuds (%rdi), %xmm1, %xmm3 # encoding: [0x62,0xf2,0x74,0x08,0x51,0x1f]428; AVX10-X64-NEXT: vpdpbuuds %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf2,0x74,0x08,0x51,0xc2]429; AVX10-X64-NEXT: vpaddd %xmm0, %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe1,0xfe,0xc0]430; AVX10-X64-NEXT: retq # encoding: [0xc3]431 %x2 = load <16 x i8>, ptr %x2p432 %1 = call <4 x i32> @llvm.x86.avx2.vpdpbuuds.128(<4 x i32> %x0, <16 x i8> %x1, <16 x i8> %x2)433 %2 = call <4 x i32> @llvm.x86.avx2.vpdpbuuds.128(<4 x i32> %x0, <16 x i8> %x1, <16 x i8> %x4)434 %res = add <4 x i32> %1, %2435 ret <4 x i32> %res436}437 438declare <8 x i32> @llvm.x86.avx2.vpdpbuud.256(<8 x i32>, <32 x i8>, <32 x i8>)439 440define <8 x i32>@test_int_x86_avx2_vpdpbuud_256(<8 x i32> %x0, <32 x i8> %x1, ptr %x2p, <32 x i8> %x4) {441; X86-LABEL: test_int_x86_avx2_vpdpbuud_256:442; X86: # %bb.0:443; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]444; X86-NEXT: vmovaps %ymm0, %ymm3 # encoding: [0xc5,0xfc,0x28,0xd8]445; X86-NEXT: vpdpbuud (%eax), %ymm1, %ymm3 # encoding: [0xc4,0xe2,0x74,0x50,0x18]446; X86-NEXT: vpdpbuud %ymm2, %ymm1, %ymm0 # encoding: [0xc4,0xe2,0x74,0x50,0xc2]447; X86-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # encoding: [0xc5,0xe5,0xfe,0xc0]448; X86-NEXT: retl # encoding: [0xc3]449;450; X64-LABEL: test_int_x86_avx2_vpdpbuud_256:451; X64: # %bb.0:452; X64-NEXT: vmovaps %ymm0, %ymm3 # encoding: [0xc5,0xfc,0x28,0xd8]453; X64-NEXT: vpdpbuud (%rdi), %ymm1, %ymm3 # encoding: [0xc4,0xe2,0x74,0x50,0x1f]454; X64-NEXT: vpdpbuud %ymm2, %ymm1, %ymm0 # encoding: [0xc4,0xe2,0x74,0x50,0xc2]455; X64-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # encoding: [0xc5,0xe5,0xfe,0xc0]456; X64-NEXT: retq # encoding: [0xc3]457;458; AVX10-X86-LABEL: test_int_x86_avx2_vpdpbuud_256:459; AVX10-X86: # %bb.0:460; AVX10-X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]461; AVX10-X86-NEXT: vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]462; AVX10-X86-NEXT: vpdpbuud (%eax), %ymm1, %ymm3 # encoding: [0x62,0xf2,0x74,0x28,0x50,0x18]463; AVX10-X86-NEXT: vpdpbuud %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x74,0x28,0x50,0xc2]464; AVX10-X86-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xe5,0xfe,0xc0]465; AVX10-X86-NEXT: retl # encoding: [0xc3]466;467; AVX10-X64-LABEL: test_int_x86_avx2_vpdpbuud_256:468; AVX10-X64: # %bb.0:469; AVX10-X64-NEXT: vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]470; AVX10-X64-NEXT: vpdpbuud (%rdi), %ymm1, %ymm3 # encoding: [0x62,0xf2,0x74,0x28,0x50,0x1f]471; AVX10-X64-NEXT: vpdpbuud %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x74,0x28,0x50,0xc2]472; AVX10-X64-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xe5,0xfe,0xc0]473; AVX10-X64-NEXT: retq # encoding: [0xc3]474 %x2 = load <32 x i8>, ptr %x2p475 %1 = call <8 x i32> @llvm.x86.avx2.vpdpbuud.256(<8 x i32> %x0, <32 x i8> %x1, <32 x i8> %x2)476 %2 = call <8 x i32> @llvm.x86.avx2.vpdpbuud.256(<8 x i32> %x0, <32 x i8> %x1, <32 x i8> %x4)477 %res = add <8 x i32> %1, %2478 ret <8 x i32> %res479}480 481declare <8 x i32> @llvm.x86.avx2.vpdpbuuds.256(<8 x i32>, <32 x i8>, <32 x i8>)482 483define <8 x i32>@test_int_x86_avx2_vpdpbuuds_256(<8 x i32> %x0, <32 x i8> %x1, ptr %x2p, <32 x i8> %x4) {484; X86-LABEL: test_int_x86_avx2_vpdpbuuds_256:485; X86: # %bb.0:486; X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]487; X86-NEXT: vmovaps %ymm0, %ymm3 # encoding: [0xc5,0xfc,0x28,0xd8]488; X86-NEXT: vpdpbuuds (%eax), %ymm1, %ymm3 # encoding: [0xc4,0xe2,0x74,0x51,0x18]489; X86-NEXT: vpdpbuuds %ymm2, %ymm1, %ymm0 # encoding: [0xc4,0xe2,0x74,0x51,0xc2]490; X86-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # encoding: [0xc5,0xe5,0xfe,0xc0]491; X86-NEXT: retl # encoding: [0xc3]492;493; X64-LABEL: test_int_x86_avx2_vpdpbuuds_256:494; X64: # %bb.0:495; X64-NEXT: vmovaps %ymm0, %ymm3 # encoding: [0xc5,0xfc,0x28,0xd8]496; X64-NEXT: vpdpbuuds (%rdi), %ymm1, %ymm3 # encoding: [0xc4,0xe2,0x74,0x51,0x1f]497; X64-NEXT: vpdpbuuds %ymm2, %ymm1, %ymm0 # encoding: [0xc4,0xe2,0x74,0x51,0xc2]498; X64-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # encoding: [0xc5,0xe5,0xfe,0xc0]499; X64-NEXT: retq # encoding: [0xc3]500;501; AVX10-X86-LABEL: test_int_x86_avx2_vpdpbuuds_256:502; AVX10-X86: # %bb.0:503; AVX10-X86-NEXT: movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]504; AVX10-X86-NEXT: vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]505; AVX10-X86-NEXT: vpdpbuuds (%eax), %ymm1, %ymm3 # encoding: [0x62,0xf2,0x74,0x28,0x51,0x18]506; AVX10-X86-NEXT: vpdpbuuds %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x74,0x28,0x51,0xc2]507; AVX10-X86-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xe5,0xfe,0xc0]508; AVX10-X86-NEXT: retl # encoding: [0xc3]509;510; AVX10-X64-LABEL: test_int_x86_avx2_vpdpbuuds_256:511; AVX10-X64: # %bb.0:512; AVX10-X64-NEXT: vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]513; AVX10-X64-NEXT: vpdpbuuds (%rdi), %ymm1, %ymm3 # encoding: [0x62,0xf2,0x74,0x28,0x51,0x1f]514; AVX10-X64-NEXT: vpdpbuuds %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x74,0x28,0x51,0xc2]515; AVX10-X64-NEXT: vpaddd %ymm0, %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xe5,0xfe,0xc0]516; AVX10-X64-NEXT: retq # encoding: [0xc3]517 %x2 = load <32 x i8>, ptr %x2p518 %1 = call <8 x i32> @llvm.x86.avx2.vpdpbuuds.256(<8 x i32> %x0, <32 x i8> %x1, <32 x i8> %x2)519 %2 = call <8 x i32> @llvm.x86.avx2.vpdpbuuds.256(<8 x i32> %x0, <32 x i8> %x1, <32 x i8> %x4)520 %res = add <8 x i32> %1, %2521 ret <8 x i32> %res522}523