llvm/test/CodeGen/X86/avx512vl_vnni-intrinsics.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx512vnni,+avx512vl --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X86
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vnni,+avx512vl --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X64

declare <8 x i32> @llvm.x86.avx512.vpdpbusd.256(<8 x i32>, <8 x i32>, <8 x i32>)

define <8 x i32>@test_int_x86_avx512_vpdpbusd_256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2) {
; CHECK-LABEL: test_int_x86_avx512_vpdpbusd_256:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vpdpbusd %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x75,0x28,0x50,0xc2]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %1 = call <8 x i32> @llvm.x86.avx512.vpdpbusd.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2)
  ret <8 x i32> %1
}

define { <8 x i32>, <8 x i32> } @test_int_x86_avx512_mask_vpdpbusd_256(<8 x i32> %x0, <8 x i32> %x1, ptr %x2p, <8 x i32> %x4, i8 %x3) {
; X86-LABEL: test_int_x86_avx512_mask_vpdpbusd_256:
; X86:       # %bb.0:
; X86-NEXT:    vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    movzbl {{[0-9]+}}(%esp), %ecx # encoding: [0x0f,0xb6,0x4c,0x24,0x08]
; X86-NEXT:    kmovw %ecx, %k1 # encoding: [0xc5,0xf8,0x92,0xc9]
; X86-NEXT:    vpdpbusd (%eax), %ymm1, %ymm0 {%k1} # encoding: [0x62,0xf2,0x75,0x29,0x50,0x00]
; X86-NEXT:    vpdpbusd %ymm2, %ymm1, %ymm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0x50,0xda]
; X86-NEXT:    vmovdqa %ymm3, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xcb]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_vpdpbusd_256:
; X64:       # %bb.0:
; X64-NEXT:    vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpdpbusd (%rdi), %ymm1, %ymm0 {%k1} # encoding: [0x62,0xf2,0x75,0x29,0x50,0x07]
; X64-NEXT:    vpdpbusd %ymm2, %ymm1, %ymm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0x50,0xda]
; X64-NEXT:    vmovdqa %ymm3, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xcb]
; X64-NEXT:    retq # encoding: [0xc3]
  %x2 = load <8 x i32>, ptr %x2p
  %1 = call <8 x i32> @llvm.x86.avx512.vpdpbusd.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2)
  %2 = bitcast i8 %x3 to <8 x i1>
  %3 = select <8 x i1> %2, <8 x i32> %1, <8 x i32> %x0
  %4 = call <8 x i32> @llvm.x86.avx512.vpdpbusd.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x4)
  %5 = bitcast i8 %x3 to <8 x i1>
  %6 = select <8 x i1> %5, <8 x i32> %4, <8 x i32> zeroinitializer
  %res1 = insertvalue { <8 x i32>, <8 x i32> } poison, <8 x i32> %3, 0
  %res2 = insertvalue { <8 x i32>, <8 x i32> }  %res1, <8 x i32> %6, 1
  ret { <8 x i32>, <8 x i32> } %res2
}

declare <4 x i32> @llvm.x86.avx512.vpdpbusd.128(<4 x i32>, <4 x i32>, <4 x i32>)

define <4 x i32>@test_int_x86_avx512_vpdpbusd_128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2) {
; CHECK-LABEL: test_int_x86_avx512_vpdpbusd_128:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vpdpbusd %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf2,0x75,0x08,0x50,0xc2]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %1 = call <4 x i32> @llvm.x86.avx512.vpdpbusd.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2)
  ret <4 x i32> %1
}

define { <4 x i32>, <4 x i32> } @test_int_x86_avx512_mask_vpdpbusd_128(<4 x i32> %x0, <4 x i32> %x1, ptr %x2p, <4 x i32> %x4, i8 %x3) {
; X86-LABEL: test_int_x86_avx512_mask_vpdpbusd_128:
; X86:       # %bb.0:
; X86-NEXT:    vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    movzbl {{[0-9]+}}(%esp), %ecx # encoding: [0x0f,0xb6,0x4c,0x24,0x08]
; X86-NEXT:    kmovw %ecx, %k1 # encoding: [0xc5,0xf8,0x92,0xc9]
; X86-NEXT:    vpdpbusd (%eax), %xmm1, %xmm0 {%k1} # encoding: [0x62,0xf2,0x75,0x09,0x50,0x00]
; X86-NEXT:    vpdpbusd %xmm2, %xmm1, %xmm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0x50,0xda]
; X86-NEXT:    vmovdqa %xmm3, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xcb]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_vpdpbusd_128:
; X64:       # %bb.0:
; X64-NEXT:    vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpdpbusd (%rdi), %xmm1, %xmm0 {%k1} # encoding: [0x62,0xf2,0x75,0x09,0x50,0x07]
; X64-NEXT:    vpdpbusd %xmm2, %xmm1, %xmm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0x50,0xda]
; X64-NEXT:    vmovdqa %xmm3, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xcb]
; X64-NEXT:    retq # encoding: [0xc3]
  %x2 = load <4 x i32>, ptr %x2p
  %1 = call <4 x i32> @llvm.x86.avx512.vpdpbusd.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2)
  %2 = bitcast i8 %x3 to <8 x i1>
  %extract = shufflevector <8 x i1> %2, <8 x i1> %2, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
  %3 = select <4 x i1> %extract, <4 x i32> %1, <4 x i32> %x0
  %4 = call <4 x i32> @llvm.x86.avx512.vpdpbusd.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x4)
  %5 = bitcast i8 %x3 to <8 x i1>
  %extract1 = shufflevector <8 x i1> %5, <8 x i1> %5, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
  %6 = select <4 x i1> %extract1, <4 x i32> %4, <4 x i32> zeroinitializer
  %res1 = insertvalue { <4 x i32>, <4 x i32> } poison, <4 x i32> %3, 0
  %res2 = insertvalue { <4 x i32>, <4 x i32> }  %res1, <4 x i32> %6, 1
  ret { <4 x i32>, <4 x i32> } %res2
}

declare <8 x i32> @llvm.x86.avx512.vpdpbusds.256(<8 x i32>, <8 x i32>, <8 x i32>)

define <8 x i32>@test_int_x86_avx512_vpdpbusds_256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2) {
; CHECK-LABEL: test_int_x86_avx512_vpdpbusds_256:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vpdpbusds %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x75,0x28,0x51,0xc2]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %1 = call <8 x i32> @llvm.x86.avx512.vpdpbusds.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2)
  ret <8 x i32> %1
}

define { <8 x i32>, <8 x i32> } @test_int_x86_avx512_mask_vpdpbusds_256(<8 x i32> %x0, <8 x i32> %x1, ptr %x2p, <8 x i32> %x4, i8 %x3) {
; X86-LABEL: test_int_x86_avx512_mask_vpdpbusds_256:
; X86:       # %bb.0:
; X86-NEXT:    vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    movzbl {{[0-9]+}}(%esp), %ecx # encoding: [0x0f,0xb6,0x4c,0x24,0x08]
; X86-NEXT:    kmovw %ecx, %k1 # encoding: [0xc5,0xf8,0x92,0xc9]
; X86-NEXT:    vpdpbusds (%eax), %ymm1, %ymm0 {%k1} # encoding: [0x62,0xf2,0x75,0x29,0x51,0x00]
; X86-NEXT:    vpdpbusds %ymm2, %ymm1, %ymm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0x51,0xda]
; X86-NEXT:    vmovdqa %ymm3, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xcb]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_vpdpbusds_256:
; X64:       # %bb.0:
; X64-NEXT:    vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpdpbusds (%rdi), %ymm1, %ymm0 {%k1} # encoding: [0x62,0xf2,0x75,0x29,0x51,0x07]
; X64-NEXT:    vpdpbusds %ymm2, %ymm1, %ymm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0x51,0xda]
; X64-NEXT:    vmovdqa %ymm3, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xcb]
; X64-NEXT:    retq # encoding: [0xc3]
  %x2 = load <8 x i32>, ptr %x2p
  %1 = call <8 x i32> @llvm.x86.avx512.vpdpbusds.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2)
  %2 = bitcast i8 %x3 to <8 x i1>
  %3 = select <8 x i1> %2, <8 x i32> %1, <8 x i32> %x0
  %4 = call <8 x i32> @llvm.x86.avx512.vpdpbusds.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x4)
  %5 = bitcast i8 %x3 to <8 x i1>
  %6 = select <8 x i1> %5, <8 x i32> %4, <8 x i32> zeroinitializer
  %res1 = insertvalue { <8 x i32>, <8 x i32> } poison, <8 x i32> %3, 0
  %res2 = insertvalue { <8 x i32>, <8 x i32> }  %res1, <8 x i32> %6, 1
  ret { <8 x i32>, <8 x i32> } %res2
}

declare <4 x i32> @llvm.x86.avx512.vpdpbusds.128(<4 x i32>, <4 x i32>, <4 x i32>)

define <4 x i32>@test_int_x86_avx512_vpdpbusds_128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2) {
; CHECK-LABEL: test_int_x86_avx512_vpdpbusds_128:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vpdpbusds %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf2,0x75,0x08,0x51,0xc2]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %1 = call <4 x i32> @llvm.x86.avx512.vpdpbusds.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2)
  ret <4 x i32> %1
}

define { <4 x i32>, <4 x i32> } @test_int_x86_avx512_mask_vpdpbusds_128(<4 x i32> %x0, <4 x i32> %x1, ptr %x2p, <4 x i32> %x4, i8 %x3) {
; X86-LABEL: test_int_x86_avx512_mask_vpdpbusds_128:
; X86:       # %bb.0:
; X86-NEXT:    vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    movzbl {{[0-9]+}}(%esp), %ecx # encoding: [0x0f,0xb6,0x4c,0x24,0x08]
; X86-NEXT:    kmovw %ecx, %k1 # encoding: [0xc5,0xf8,0x92,0xc9]
; X86-NEXT:    vpdpbusds (%eax), %xmm1, %xmm0 {%k1} # encoding: [0x62,0xf2,0x75,0x09,0x51,0x00]
; X86-NEXT:    vpdpbusds %xmm2, %xmm1, %xmm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0x51,0xda]
; X86-NEXT:    vmovdqa %xmm3, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xcb]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_vpdpbusds_128:
; X64:       # %bb.0:
; X64-NEXT:    vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpdpbusds (%rdi), %xmm1, %xmm0 {%k1} # encoding: [0x62,0xf2,0x75,0x09,0x51,0x07]
; X64-NEXT:    vpdpbusds %xmm2, %xmm1, %xmm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0x51,0xda]
; X64-NEXT:    vmovdqa %xmm3, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xcb]
; X64-NEXT:    retq # encoding: [0xc3]
  %x2 = load <4 x i32>, ptr %x2p
  %1 = call <4 x i32> @llvm.x86.avx512.vpdpbusds.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2)
  %2 = bitcast i8 %x3 to <8 x i1>
  %extract = shufflevector <8 x i1> %2, <8 x i1> %2, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
  %3 = select <4 x i1> %extract, <4 x i32> %1, <4 x i32> %x0
  %4 = call <4 x i32> @llvm.x86.avx512.vpdpbusds.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x4)
  %5 = bitcast i8 %x3 to <8 x i1>
  %extract1 = shufflevector <8 x i1> %5, <8 x i1> %5, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
  %6 = select <4 x i1> %extract1, <4 x i32> %4, <4 x i32> zeroinitializer
  %res1 = insertvalue { <4 x i32>, <4 x i32> } poison, <4 x i32> %3, 0
  %res2 = insertvalue { <4 x i32>, <4 x i32> }  %res1, <4 x i32> %6, 1
  ret { <4 x i32>, <4 x i32> } %res2
}

declare <8 x i32> @llvm.x86.avx512.vpdpwssd.256(<8 x i32>, <8 x i32>, <8 x i32>)

define <8 x i32>@test_int_x86_avx512_vpdpwssd_256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2) {
; CHECK-LABEL: test_int_x86_avx512_vpdpwssd_256:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vpdpwssd %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x75,0x28,0x52,0xc2]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %1 = call <8 x i32> @llvm.x86.avx512.vpdpwssd.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2)
  ret <8 x i32> %1
}

define { <8 x i32>, <8 x i32> } @test_int_x86_avx512_mask_vpdpwssd_256(<8 x i32> %x0, <8 x i32> %x1, ptr %x2p, <8 x i32> %x4, i8 %x3) {
; X86-LABEL: test_int_x86_avx512_mask_vpdpwssd_256:
; X86:       # %bb.0:
; X86-NEXT:    vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    movzbl {{[0-9]+}}(%esp), %ecx # encoding: [0x0f,0xb6,0x4c,0x24,0x08]
; X86-NEXT:    kmovw %ecx, %k1 # encoding: [0xc5,0xf8,0x92,0xc9]
; X86-NEXT:    vpdpwssd (%eax), %ymm1, %ymm0 {%k1} # encoding: [0x62,0xf2,0x75,0x29,0x52,0x00]
; X86-NEXT:    vpdpwssd %ymm2, %ymm1, %ymm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0x52,0xda]
; X86-NEXT:    vmovdqa %ymm3, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xcb]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_vpdpwssd_256:
; X64:       # %bb.0:
; X64-NEXT:    vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpdpwssd (%rdi), %ymm1, %ymm0 {%k1} # encoding: [0x62,0xf2,0x75,0x29,0x52,0x07]
; X64-NEXT:    vpdpwssd %ymm2, %ymm1, %ymm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0x52,0xda]
; X64-NEXT:    vmovdqa %ymm3, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xcb]
; X64-NEXT:    retq # encoding: [0xc3]
  %x2 = load <8 x i32>, ptr %x2p
  %1 = call <8 x i32> @llvm.x86.avx512.vpdpwssd.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2)
  %2 = bitcast i8 %x3 to <8 x i1>
  %3 = select <8 x i1> %2, <8 x i32> %1, <8 x i32> %x0
  %4 = call <8 x i32> @llvm.x86.avx512.vpdpwssd.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x4)
  %5 = bitcast i8 %x3 to <8 x i1>
  %6 = select <8 x i1> %5, <8 x i32> %4, <8 x i32> zeroinitializer
  %res1 = insertvalue { <8 x i32>, <8 x i32> } poison, <8 x i32> %3, 0
  %res2 = insertvalue { <8 x i32>, <8 x i32> }  %res1, <8 x i32> %6, 1
  ret { <8 x i32>, <8 x i32> } %res2
}

declare <4 x i32> @llvm.x86.avx512.vpdpwssd.128(<4 x i32>, <4 x i32>, <4 x i32>)

define <4 x i32>@test_int_x86_avx512_vpdpwssd_128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2) {
; CHECK-LABEL: test_int_x86_avx512_vpdpwssd_128:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vpdpwssd %xmm2, %xmm1, %xmm0 # encoding: [0x62,0xf2,0x75,0x08,0x52,0xc2]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %1 = call <4 x i32> @llvm.x86.avx512.vpdpwssd.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2)
  ret <4 x i32> %1
}

define { <4 x i32>, <4 x i32> } @test_int_x86_avx512_mask_vpdpwssd_128(<4 x i32> %x0, <4 x i32> %x1, ptr %x2p, <4 x i32> %x4, i8 %x3) {
; X86-LABEL: test_int_x86_avx512_mask_vpdpwssd_128:
; X86:       # %bb.0:
; X86-NEXT:    vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    movzbl {{[0-9]+}}(%esp), %ecx # encoding: [0x0f,0xb6,0x4c,0x24,0x08]
; X86-NEXT:    kmovw %ecx, %k1 # encoding: [0xc5,0xf8,0x92,0xc9]
; X86-NEXT:    vpdpwssd (%eax), %xmm1, %xmm0 {%k1} # encoding: [0x62,0xf2,0x75,0x09,0x52,0x00]
; X86-NEXT:    vpdpwssd %xmm2, %xmm1, %xmm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0x52,0xda]
; X86-NEXT:    vmovdqa %xmm3, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xcb]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_vpdpwssd_128:
; X64:       # %bb.0:
; X64-NEXT:    vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpdpwssd (%rdi), %xmm1, %xmm0 {%k1} # encoding: [0x62,0xf2,0x75,0x09,0x52,0x07]
; X64-NEXT:    vpdpwssd %xmm2, %xmm1, %xmm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0x52,0xda]
; X64-NEXT:    vmovdqa %xmm3, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xcb]
; X64-NEXT:    retq # encoding: [0xc3]
  %x2 = load <4 x i32>, ptr %x2p
  %1 = call <4 x i32> @llvm.x86.avx512.vpdpwssd.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2)
  %2 = bitcast i8 %x3 to <8 x i1>
  %extract = shufflevector <8 x i1> %2, <8 x i1> %2, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
  %3 = select <4 x i1> %extract, <4 x i32> %1, <4 x i32> %x0
  %4 = call <4 x i32> @llvm.x86.avx512.vpdpwssd.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x4)
  %5 = bitcast i8 %x3 to <8 x i1>
  %extract1 = shufflevector <8 x i1> %5, <8 x i1> %5, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
  %6 = select <4 x i1> %extract1, <4 x i32> %4, <4 x i32> zeroinitializer
  %res1 = insertvalue { <4 x i32>, <4 x i32> } poison, <4 x i32> %3, 0
  %res2 = insertvalue { <4 x i32>, <4 x i32> }  %res1, <4 x i32> %6, 1
  ret { <4 x i32>, <4 x i32> } %res2
}

declare <8 x i32> @llvm.x86.avx512.vpdpwssds.256(<8 x i32>, <8 x i32>, <8 x i32>)

define <8 x i32>@test_int_x86_avx512_vpdpwssds_256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2) {
; CHECK-LABEL: test_int_x86_avx512_vpdpwssds_256:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vpdpwssds %ymm2, %ymm1, %ymm0 # encoding: [0x62,0xf2,0x75,0x28,0x53,0xc2]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %1 = call <8 x i32> @llvm.x86.avx512.vpdpwssds.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2)
  ret <8 x i32> %1
}

define { <8 x i32>, <8 x i32> } @test_int_x86_avx512_mask_vpdpwssds_256(<8 x i32> %x0, <8 x i32> %x1, ptr %x2p, <8 x i32> %x4, i8 %x3) {
; X86-LABEL: test_int_x86_avx512_mask_vpdpwssds_256:
; X86:       # %bb.0:
; X86-NEXT:    vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    movzbl {{[0-9]+}}(%esp), %ecx # encoding: [0x0f,0xb6,0x4c,0x24,0x08]
; X86-NEXT:    kmovw %ecx, %k1 # encoding: [0xc5,0xf8,0x92,0xc9]
; X86-NEXT:    vpdpwssds (%eax), %ymm1, %ymm0 {%k1} # encoding: [0x62,0xf2,0x75,0x29,0x53,0x00]
; X86-NEXT:    vpdpwssds %ymm2, %ymm1, %ymm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0x53,0xda]
; X86-NEXT:    vmovdqa %ymm3, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xcb]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_vpdpwssds_256:
; X64:       # %bb.0:
; X64-NEXT:    vmovdqa %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xd8]
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpdpwssds (%rdi), %ymm1, %ymm0 {%k1} # encoding: [0x62,0xf2,0x75,0x29,0x53,0x07]
; X64-NEXT:    vpdpwssds %ymm2, %ymm1, %ymm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0xa9,0x53,0xda]
; X64-NEXT:    vmovdqa %ymm3, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xcb]
; X64-NEXT:    retq # encoding: [0xc3]
  %x2 = load <8 x i32>, ptr %x2p
  %1 = call <8 x i32> @llvm.x86.avx512.vpdpwssds.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x2)
  %2 = bitcast i8 %x3 to <8 x i1>
  %3 = select <8 x i1> %2, <8 x i32> %1, <8 x i32> %x0
  %4 = call <8 x i32> @llvm.x86.avx512.vpdpwssds.256(<8 x i32> %x0, <8 x i32> %x1, <8 x i32> %x4)
  %5 = bitcast i8 %x3 to <8 x i1>
  %6 = select <8 x i1> %5, <8 x i32> %4, <8 x i32> zeroinitializer
  %res1 = insertvalue { <8 x i32>, <8 x i32> } poison, <8 x i32> %3, 0
  %res2 = insertvalue { <8 x i32>, <8 x i32> }  %res1, <8 x i32> %6, 1
  ret { <8 x i32>, <8 x i32> } %res2
}

declare <4 x i32> @llvm.x86.avx512.vpdpwssds.128(<4 x i32>, <4 x i32>, <4 x i32>)

define <4 x i32>@test_int_x86_avx512_vpdpwssds_128(<4 x i32> %x0, <4 x i32> %x1, ptr %x2p) {
; X86-LABEL: test_int_x86_avx512_vpdpwssds_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vpdpwssds (%eax), %xmm1, %xmm0 # encoding: [0x62,0xf2,0x75,0x08,0x53,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_vpdpwssds_128:
; X64:       # %bb.0:
; X64-NEXT:    vpdpwssds (%rdi), %xmm1, %xmm0 # encoding: [0x62,0xf2,0x75,0x08,0x53,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %x2 = load <4 x i32>, ptr %x2p
  %1 = call <4 x i32> @llvm.x86.avx512.vpdpwssds.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2)
  ret <4 x i32> %1
}

define { <4 x i32>, <4 x i32> } @test_int_x86_avx512_mask_vpdpwssds_128(<4 x i32> %x0, <4 x i32> %x1, ptr %x2p, <4 x i32> %x4, i8 %x3) {
; X86-LABEL: test_int_x86_avx512_mask_vpdpwssds_128:
; X86:       # %bb.0:
; X86-NEXT:    vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    movzbl {{[0-9]+}}(%esp), %ecx # encoding: [0x0f,0xb6,0x4c,0x24,0x08]
; X86-NEXT:    kmovw %ecx, %k1 # encoding: [0xc5,0xf8,0x92,0xc9]
; X86-NEXT:    vpdpwssds (%eax), %xmm1, %xmm0 {%k1} # encoding: [0x62,0xf2,0x75,0x09,0x53,0x00]
; X86-NEXT:    vpdpwssds %xmm2, %xmm1, %xmm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0x53,0xda]
; X86-NEXT:    vmovdqa %xmm3, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xcb]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_vpdpwssds_128:
; X64:       # %bb.0:
; X64-NEXT:    vmovdqa %xmm0, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xd8]
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpdpwssds (%rdi), %xmm1, %xmm0 {%k1} # encoding: [0x62,0xf2,0x75,0x09,0x53,0x07]
; X64-NEXT:    vpdpwssds %xmm2, %xmm1, %xmm3 {%k1} {z} # encoding: [0x62,0xf2,0x75,0x89,0x53,0xda]
; X64-NEXT:    vmovdqa %xmm3, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xcb]
; X64-NEXT:    retq # encoding: [0xc3]
  %x2 = load <4 x i32>, ptr %x2p
  %1 = call <4 x i32> @llvm.x86.avx512.vpdpwssds.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x2)
  %2 = bitcast i8 %x3 to <8 x i1>
  %extract = shufflevector <8 x i1> %2, <8 x i1> %2, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
  %3 = select <4 x i1> %extract, <4 x i32> %1, <4 x i32> %x0
  %4 = call <4 x i32> @llvm.x86.avx512.vpdpwssds.128(<4 x i32> %x0, <4 x i32> %x1, <4 x i32> %x4)
  %5 = bitcast i8 %x3 to <8 x i1>
  %extract1 = shufflevector <8 x i1> %5, <8 x i1> %5, <4 x i32> <i32 0, i32 1, i32 2, i32 3>
  %6 = select <4 x i1> %extract1, <4 x i32> %4, <4 x i32> zeroinitializer
  %res1 = insertvalue { <4 x i32>, <4 x i32> } poison, <4 x i32> %3, 0
  %res2 = insertvalue { <4 x i32>, <4 x i32> }  %res1, <4 x i32> %6, 1
  ret { <4 x i32>, <4 x i32> } %res2
}