third_party/llvm-7.0/llvm/test/CodeGen/X86/avx512dqvl-intrinsics-upgrade.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx512dq,+avx512vl --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X86
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512dq,+avx512vl --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X64

define <4 x float> @test_mask_andnot_ps_rr_128(<4 x float> %a, <4 x float> %b) {
; CHECK-LABEL: test_mask_andnot_ps_rr_128:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vandnps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x55,0xc1]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.avx512.mask.andn.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 -1)
  ret <4 x float> %res
}

define <4 x float> @test_mask_andnot_ps_rrk_128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_andnot_ps_rrk_128:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vandnps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x55,0xd1]
; X86-NEXT:    vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rrk_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vandnps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x55,0xd1]
; X64-NEXT:    vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.avx512.mask.andn.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_andnot_ps_rrkz_128(<4 x float> %a, <4 x float> %b, i8 %mask) {
; X86-LABEL: test_mask_andnot_ps_rrkz_128:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vandnps %xmm1, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x55,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rrkz_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vandnps %xmm1, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x55,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.avx512.mask.andn.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_andnot_ps_rm_128(<4 x float> %a, <4 x float>* %ptr_b) {
; X86-LABEL: test_mask_andnot_ps_rm_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vandnps (%eax), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x55,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rm_128:
; X64:       # %bb.0:
; X64-NEXT:    vandnps (%rdi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x55,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <4 x float>, <4 x float>* %ptr_b
  %res = call <4 x float> @llvm.x86.avx512.mask.andn.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 -1)
  ret <4 x float> %res
}

define <4 x float> @test_mask_andnot_ps_rmk_128(<4 x float> %a, <4 x float>* %ptr_b, <4 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_andnot_ps_rmk_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandnps (%eax), %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x55,0x08]
; X86-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rmk_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandnps (%rdi), %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x55,0x0f]
; X64-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <4 x float>, <4 x float>* %ptr_b
  %res = call <4 x float> @llvm.x86.avx512.mask.andn.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_andnot_ps_rmkz_128(<4 x float> %a, <4 x float>* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_andnot_ps_rmkz_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandnps (%eax), %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x55,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rmkz_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandnps (%rdi), %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x55,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <4 x float>, <4 x float>* %ptr_b
  %res = call <4 x float> @llvm.x86.avx512.mask.andn.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_andnot_ps_rmb_128(<4 x float> %a, float* %ptr_b) {
; X86-LABEL: test_mask_andnot_ps_rmb_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vandnps (%eax){1to4}, %xmm0, %xmm0 # encoding: [0x62,0xf1,0x7c,0x18,0x55,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rmb_128:
; X64:       # %bb.0:
; X64-NEXT:    vandnps (%rdi){1to4}, %xmm0, %xmm0 # encoding: [0x62,0xf1,0x7c,0x18,0x55,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <4 x float> undef, float %q, i32 0
  %b = shufflevector <4 x float> %vecinit.i, <4 x float> undef, <4 x i32> zeroinitializer
  %res = call <4 x float> @llvm.x86.avx512.mask.andn.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 -1)
  ret <4 x float> %res
}

define <4 x float> @test_mask_andnot_ps_rmbk_128(<4 x float> %a, float* %ptr_b, <4 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_andnot_ps_rmbk_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandnps (%eax){1to4}, %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x19,0x55,0x08]
; X86-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rmbk_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandnps (%rdi){1to4}, %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x19,0x55,0x0f]
; X64-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <4 x float> undef, float %q, i32 0
  %b = shufflevector <4 x float> %vecinit.i, <4 x float> undef, <4 x i32> zeroinitializer
  %res = call <4 x float> @llvm.x86.avx512.mask.andn.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_andnot_ps_rmbkz_128(<4 x float> %a, float* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_andnot_ps_rmbkz_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandnps (%eax){1to4}, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x99,0x55,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rmbkz_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandnps (%rdi){1to4}, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x99,0x55,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <4 x float> undef, float %q, i32 0
  %b = shufflevector <4 x float> %vecinit.i, <4 x float> undef, <4 x i32> zeroinitializer
  %res = call <4 x float> @llvm.x86.avx512.mask.andn.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 %mask)
  ret <4 x float> %res
}

declare <4 x float> @llvm.x86.avx512.mask.andn.ps.128(<4 x float>, <4 x float>, <4 x float>, i8)

define <8 x float> @test_mask_andnot_ps_rr_256(<8 x float> %a, <8 x float> %b) {
; CHECK-LABEL: test_mask_andnot_ps_rr_256:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vandnps %ymm1, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x55,0xc1]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <8 x float> @llvm.x86.avx512.mask.andn.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 -1)
  ret <8 x float> %res
}

define <8 x float> @test_mask_andnot_ps_rrk_256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_andnot_ps_rrk_256:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vandnps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x55,0xd1]
; X86-NEXT:    vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rrk_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vandnps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x55,0xd1]
; X64-NEXT:    vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x float> @llvm.x86.avx512.mask.andn.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_andnot_ps_rrkz_256(<8 x float> %a, <8 x float> %b, i8 %mask) {
; X86-LABEL: test_mask_andnot_ps_rrkz_256:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vandnps %ymm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x55,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rrkz_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vandnps %ymm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x55,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x float> @llvm.x86.avx512.mask.andn.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_andnot_ps_rm_256(<8 x float> %a, <8 x float>* %ptr_b) {
; X86-LABEL: test_mask_andnot_ps_rm_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vandnps (%eax), %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x55,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rm_256:
; X64:       # %bb.0:
; X64-NEXT:    vandnps (%rdi), %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x55,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <8 x float>, <8 x float>* %ptr_b
  %res = call <8 x float> @llvm.x86.avx512.mask.andn.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 -1)
  ret <8 x float> %res
}

define <8 x float> @test_mask_andnot_ps_rmk_256(<8 x float> %a, <8 x float>* %ptr_b, <8 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_andnot_ps_rmk_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandnps (%eax), %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x55,0x08]
; X86-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rmk_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandnps (%rdi), %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x55,0x0f]
; X64-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <8 x float>, <8 x float>* %ptr_b
  %res = call <8 x float> @llvm.x86.avx512.mask.andn.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_andnot_ps_rmkz_256(<8 x float> %a, <8 x float>* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_andnot_ps_rmkz_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandnps (%eax), %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x55,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rmkz_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandnps (%rdi), %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x55,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <8 x float>, <8 x float>* %ptr_b
  %res = call <8 x float> @llvm.x86.avx512.mask.andn.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_andnot_ps_rmb_256(<8 x float> %a, float* %ptr_b) {
; X86-LABEL: test_mask_andnot_ps_rmb_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vandnps (%eax){1to8}, %ymm0, %ymm0 # encoding: [0x62,0xf1,0x7c,0x38,0x55,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rmb_256:
; X64:       # %bb.0:
; X64-NEXT:    vandnps (%rdi){1to8}, %ymm0, %ymm0 # encoding: [0x62,0xf1,0x7c,0x38,0x55,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <8 x float> undef, float %q, i32 0
  %b = shufflevector <8 x float> %vecinit.i, <8 x float> undef, <8 x i32> zeroinitializer
  %res = call <8 x float> @llvm.x86.avx512.mask.andn.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 -1)
  ret <8 x float> %res
}

define <8 x float> @test_mask_andnot_ps_rmbk_256(<8 x float> %a, float* %ptr_b, <8 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_andnot_ps_rmbk_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandnps (%eax){1to8}, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x39,0x55,0x08]
; X86-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rmbk_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandnps (%rdi){1to8}, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x39,0x55,0x0f]
; X64-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <8 x float> undef, float %q, i32 0
  %b = shufflevector <8 x float> %vecinit.i, <8 x float> undef, <8 x i32> zeroinitializer
  %res = call <8 x float> @llvm.x86.avx512.mask.andn.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_andnot_ps_rmbkz_256(<8 x float> %a, float* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_andnot_ps_rmbkz_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandnps (%eax){1to8}, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xb9,0x55,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rmbkz_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandnps (%rdi){1to8}, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xb9,0x55,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <8 x float> undef, float %q, i32 0
  %b = shufflevector <8 x float> %vecinit.i, <8 x float> undef, <8 x i32> zeroinitializer
  %res = call <8 x float> @llvm.x86.avx512.mask.andn.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 %mask)
  ret <8 x float> %res
}

declare <8 x float> @llvm.x86.avx512.mask.andn.ps.256(<8 x float>, <8 x float>, <8 x float>, i8)

define <16 x float> @test_mask_andnot_ps_rr_512(<16 x float> %a, <16 x float> %b) {
; CHECK-LABEL: test_mask_andnot_ps_rr_512:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vandnps %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x55,0xc1]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <16 x float> @llvm.x86.avx512.mask.andn.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 -1)
  ret <16 x float> %res
}

define <16 x float> @test_mask_andnot_ps_rrk_512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask) {
; X86-LABEL: test_mask_andnot_ps_rrk_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vandnps %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x55,0xd1]
; X86-NEXT:    vmovaps %zmm2, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rrk_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vandnps %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x55,0xd1]
; X64-NEXT:    vmovaps %zmm2, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <16 x float> @llvm.x86.avx512.mask.andn.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_andnot_ps_rrkz_512(<16 x float> %a, <16 x float> %b, i16 %mask) {
; X86-LABEL: test_mask_andnot_ps_rrkz_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vandnps %zmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x55,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rrkz_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vandnps %zmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x55,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <16 x float> @llvm.x86.avx512.mask.andn.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_andnot_ps_rm_512(<16 x float> %a, <16 x float>* %ptr_b) {
; X86-LABEL: test_mask_andnot_ps_rm_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vandnps (%eax), %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x55,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rm_512:
; X64:       # %bb.0:
; X64-NEXT:    vandnps (%rdi), %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x55,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <16 x float>, <16 x float>* %ptr_b
  %res = call <16 x float> @llvm.x86.avx512.mask.andn.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 -1)
  ret <16 x float> %res
}

define <16 x float> @test_mask_andnot_ps_rmk_512(<16 x float> %a, <16 x float>* %ptr_b, <16 x float> %passThru, i16 %mask) {
; X86-LABEL: test_mask_andnot_ps_rmk_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandnps (%eax), %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x55,0x08]
; X86-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rmk_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandnps (%rdi), %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x55,0x0f]
; X64-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <16 x float>, <16 x float>* %ptr_b
  %res = call <16 x float> @llvm.x86.avx512.mask.andn.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_andnot_ps_rmkz_512(<16 x float> %a, <16 x float>* %ptr_b, i16 %mask) {
; X86-LABEL: test_mask_andnot_ps_rmkz_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandnps (%eax), %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x55,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rmkz_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandnps (%rdi), %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x55,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <16 x float>, <16 x float>* %ptr_b
  %res = call <16 x float> @llvm.x86.avx512.mask.andn.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_andnot_ps_rmb_512(<16 x float> %a, float* %ptr_b) {
; X86-LABEL: test_mask_andnot_ps_rmb_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vandnps (%eax){1to16}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x58,0x55,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rmb_512:
; X64:       # %bb.0:
; X64-NEXT:    vandnps (%rdi){1to16}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x58,0x55,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <16 x float> undef, float %q, i32 0
  %b = shufflevector <16 x float> %vecinit.i, <16 x float> undef, <16 x i32> zeroinitializer
  %res = call <16 x float> @llvm.x86.avx512.mask.andn.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 -1)
  ret <16 x float> %res
}

define <16 x float> @test_mask_andnot_ps_rmbk_512(<16 x float> %a, float* %ptr_b, <16 x float> %passThru, i16 %mask) {
; X86-LABEL: test_mask_andnot_ps_rmbk_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandnps (%eax){1to16}, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x59,0x55,0x08]
; X86-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rmbk_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandnps (%rdi){1to16}, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x59,0x55,0x0f]
; X64-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <16 x float> undef, float %q, i32 0
  %b = shufflevector <16 x float> %vecinit.i, <16 x float> undef, <16 x i32> zeroinitializer
  %res = call <16 x float> @llvm.x86.avx512.mask.andn.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_andnot_ps_rmbkz_512(<16 x float> %a, float* %ptr_b, i16 %mask) {
; X86-LABEL: test_mask_andnot_ps_rmbkz_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandnps (%eax){1to16}, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xd9,0x55,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_andnot_ps_rmbkz_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandnps (%rdi){1to16}, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xd9,0x55,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <16 x float> undef, float %q, i32 0
  %b = shufflevector <16 x float> %vecinit.i, <16 x float> undef, <16 x i32> zeroinitializer
  %res = call <16 x float> @llvm.x86.avx512.mask.andn.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 %mask)
  ret <16 x float> %res
}

declare <16 x float> @llvm.x86.avx512.mask.andn.ps.512(<16 x float>, <16 x float>, <16 x float>, i16)

define <4 x float> @test_mask_and_ps_rr_128(<4 x float> %a, <4 x float> %b) {
; CHECK-LABEL: test_mask_and_ps_rr_128:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vandps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x54,0xc1]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.avx512.mask.and.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 -1)
  ret <4 x float> %res
}

define <4 x float> @test_mask_and_ps_rrk_128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_and_ps_rrk_128:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vandps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x54,0xd1]
; X86-NEXT:    vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rrk_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vandps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x54,0xd1]
; X64-NEXT:    vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.avx512.mask.and.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_and_ps_rrkz_128(<4 x float> %a, <4 x float> %b, i8 %mask) {
; X86-LABEL: test_mask_and_ps_rrkz_128:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vandps %xmm1, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x54,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rrkz_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vandps %xmm1, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x54,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.avx512.mask.and.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_and_ps_rm_128(<4 x float> %a, <4 x float>* %ptr_b) {
; X86-LABEL: test_mask_and_ps_rm_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vandps (%eax), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x54,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rm_128:
; X64:       # %bb.0:
; X64-NEXT:    vandps (%rdi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x54,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <4 x float>, <4 x float>* %ptr_b
  %res = call <4 x float> @llvm.x86.avx512.mask.and.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 -1)
  ret <4 x float> %res
}

define <4 x float> @test_mask_and_ps_rmk_128(<4 x float> %a, <4 x float>* %ptr_b, <4 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_and_ps_rmk_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandps (%eax), %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x54,0x08]
; X86-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rmk_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandps (%rdi), %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x54,0x0f]
; X64-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <4 x float>, <4 x float>* %ptr_b
  %res = call <4 x float> @llvm.x86.avx512.mask.and.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_and_ps_rmkz_128(<4 x float> %a, <4 x float>* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_and_ps_rmkz_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandps (%eax), %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x54,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rmkz_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandps (%rdi), %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x54,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <4 x float>, <4 x float>* %ptr_b
  %res = call <4 x float> @llvm.x86.avx512.mask.and.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_and_ps_rmb_128(<4 x float> %a, float* %ptr_b) {
; X86-LABEL: test_mask_and_ps_rmb_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vandps (%eax){1to4}, %xmm0, %xmm0 # encoding: [0x62,0xf1,0x7c,0x18,0x54,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rmb_128:
; X64:       # %bb.0:
; X64-NEXT:    vandps (%rdi){1to4}, %xmm0, %xmm0 # encoding: [0x62,0xf1,0x7c,0x18,0x54,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <4 x float> undef, float %q, i32 0
  %b = shufflevector <4 x float> %vecinit.i, <4 x float> undef, <4 x i32> zeroinitializer
  %res = call <4 x float> @llvm.x86.avx512.mask.and.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 -1)
  ret <4 x float> %res
}

define <4 x float> @test_mask_and_ps_rmbk_128(<4 x float> %a, float* %ptr_b, <4 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_and_ps_rmbk_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandps (%eax){1to4}, %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x19,0x54,0x08]
; X86-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rmbk_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandps (%rdi){1to4}, %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x19,0x54,0x0f]
; X64-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <4 x float> undef, float %q, i32 0
  %b = shufflevector <4 x float> %vecinit.i, <4 x float> undef, <4 x i32> zeroinitializer
  %res = call <4 x float> @llvm.x86.avx512.mask.and.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_and_ps_rmbkz_128(<4 x float> %a, float* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_and_ps_rmbkz_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandps (%eax){1to4}, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x99,0x54,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rmbkz_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandps (%rdi){1to4}, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x99,0x54,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <4 x float> undef, float %q, i32 0
  %b = shufflevector <4 x float> %vecinit.i, <4 x float> undef, <4 x i32> zeroinitializer
  %res = call <4 x float> @llvm.x86.avx512.mask.and.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 %mask)
  ret <4 x float> %res
}

declare <4 x float> @llvm.x86.avx512.mask.and.ps.128(<4 x float>, <4 x float>, <4 x float>, i8)

define <8 x float> @test_mask_and_ps_rr_256(<8 x float> %a, <8 x float> %b) {
; CHECK-LABEL: test_mask_and_ps_rr_256:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vandps %ymm1, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x54,0xc1]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <8 x float> @llvm.x86.avx512.mask.and.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 -1)
  ret <8 x float> %res
}

define <8 x float> @test_mask_and_ps_rrk_256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_and_ps_rrk_256:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vandps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x54,0xd1]
; X86-NEXT:    vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rrk_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vandps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x54,0xd1]
; X64-NEXT:    vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x float> @llvm.x86.avx512.mask.and.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_and_ps_rrkz_256(<8 x float> %a, <8 x float> %b, i8 %mask) {
; X86-LABEL: test_mask_and_ps_rrkz_256:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vandps %ymm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x54,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rrkz_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vandps %ymm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x54,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x float> @llvm.x86.avx512.mask.and.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_and_ps_rm_256(<8 x float> %a, <8 x float>* %ptr_b) {
; X86-LABEL: test_mask_and_ps_rm_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vandps (%eax), %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x54,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rm_256:
; X64:       # %bb.0:
; X64-NEXT:    vandps (%rdi), %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x54,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <8 x float>, <8 x float>* %ptr_b
  %res = call <8 x float> @llvm.x86.avx512.mask.and.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 -1)
  ret <8 x float> %res
}

define <8 x float> @test_mask_and_ps_rmk_256(<8 x float> %a, <8 x float>* %ptr_b, <8 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_and_ps_rmk_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandps (%eax), %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x54,0x08]
; X86-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rmk_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandps (%rdi), %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x54,0x0f]
; X64-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <8 x float>, <8 x float>* %ptr_b
  %res = call <8 x float> @llvm.x86.avx512.mask.and.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_and_ps_rmkz_256(<8 x float> %a, <8 x float>* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_and_ps_rmkz_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandps (%eax), %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x54,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rmkz_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandps (%rdi), %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x54,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <8 x float>, <8 x float>* %ptr_b
  %res = call <8 x float> @llvm.x86.avx512.mask.and.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_and_ps_rmb_256(<8 x float> %a, float* %ptr_b) {
; X86-LABEL: test_mask_and_ps_rmb_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vandps (%eax){1to8}, %ymm0, %ymm0 # encoding: [0x62,0xf1,0x7c,0x38,0x54,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rmb_256:
; X64:       # %bb.0:
; X64-NEXT:    vandps (%rdi){1to8}, %ymm0, %ymm0 # encoding: [0x62,0xf1,0x7c,0x38,0x54,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <8 x float> undef, float %q, i32 0
  %b = shufflevector <8 x float> %vecinit.i, <8 x float> undef, <8 x i32> zeroinitializer
  %res = call <8 x float> @llvm.x86.avx512.mask.and.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 -1)
  ret <8 x float> %res
}

define <8 x float> @test_mask_and_ps_rmbk_256(<8 x float> %a, float* %ptr_b, <8 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_and_ps_rmbk_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandps (%eax){1to8}, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x39,0x54,0x08]
; X86-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rmbk_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandps (%rdi){1to8}, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x39,0x54,0x0f]
; X64-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <8 x float> undef, float %q, i32 0
  %b = shufflevector <8 x float> %vecinit.i, <8 x float> undef, <8 x i32> zeroinitializer
  %res = call <8 x float> @llvm.x86.avx512.mask.and.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_and_ps_rmbkz_256(<8 x float> %a, float* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_and_ps_rmbkz_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandps (%eax){1to8}, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xb9,0x54,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rmbkz_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandps (%rdi){1to8}, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xb9,0x54,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <8 x float> undef, float %q, i32 0
  %b = shufflevector <8 x float> %vecinit.i, <8 x float> undef, <8 x i32> zeroinitializer
  %res = call <8 x float> @llvm.x86.avx512.mask.and.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 %mask)
  ret <8 x float> %res
}

declare <8 x float> @llvm.x86.avx512.mask.and.ps.256(<8 x float>, <8 x float>, <8 x float>, i8)

define <16 x float> @test_mask_and_ps_rr_512(<16 x float> %a, <16 x float> %b) {
; CHECK-LABEL: test_mask_and_ps_rr_512:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vandps %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x54,0xc1]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <16 x float> @llvm.x86.avx512.mask.and.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 -1)
  ret <16 x float> %res
}

define <16 x float> @test_mask_and_ps_rrk_512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask) {
; X86-LABEL: test_mask_and_ps_rrk_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vandps %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x54,0xd1]
; X86-NEXT:    vmovaps %zmm2, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rrk_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vandps %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x54,0xd1]
; X64-NEXT:    vmovaps %zmm2, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <16 x float> @llvm.x86.avx512.mask.and.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_and_ps_rrkz_512(<16 x float> %a, <16 x float> %b, i16 %mask) {
; X86-LABEL: test_mask_and_ps_rrkz_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vandps %zmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x54,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rrkz_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vandps %zmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x54,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <16 x float> @llvm.x86.avx512.mask.and.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_and_ps_rm_512(<16 x float> %a, <16 x float>* %ptr_b) {
; X86-LABEL: test_mask_and_ps_rm_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vandps (%eax), %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x54,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rm_512:
; X64:       # %bb.0:
; X64-NEXT:    vandps (%rdi), %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x54,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <16 x float>, <16 x float>* %ptr_b
  %res = call <16 x float> @llvm.x86.avx512.mask.and.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 -1)
  ret <16 x float> %res
}

define <16 x float> @test_mask_and_ps_rmk_512(<16 x float> %a, <16 x float>* %ptr_b, <16 x float> %passThru, i16 %mask) {
; X86-LABEL: test_mask_and_ps_rmk_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandps (%eax), %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x54,0x08]
; X86-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rmk_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandps (%rdi), %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x54,0x0f]
; X64-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <16 x float>, <16 x float>* %ptr_b
  %res = call <16 x float> @llvm.x86.avx512.mask.and.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_and_ps_rmkz_512(<16 x float> %a, <16 x float>* %ptr_b, i16 %mask) {
; X86-LABEL: test_mask_and_ps_rmkz_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandps (%eax), %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x54,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rmkz_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandps (%rdi), %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x54,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <16 x float>, <16 x float>* %ptr_b
  %res = call <16 x float> @llvm.x86.avx512.mask.and.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_and_ps_rmb_512(<16 x float> %a, float* %ptr_b) {
; X86-LABEL: test_mask_and_ps_rmb_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vandps (%eax){1to16}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x58,0x54,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rmb_512:
; X64:       # %bb.0:
; X64-NEXT:    vandps (%rdi){1to16}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x58,0x54,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <16 x float> undef, float %q, i32 0
  %b = shufflevector <16 x float> %vecinit.i, <16 x float> undef, <16 x i32> zeroinitializer
  %res = call <16 x float> @llvm.x86.avx512.mask.and.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 -1)
  ret <16 x float> %res
}

define <16 x float> @test_mask_and_ps_rmbk_512(<16 x float> %a, float* %ptr_b, <16 x float> %passThru, i16 %mask) {
; X86-LABEL: test_mask_and_ps_rmbk_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandps (%eax){1to16}, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x59,0x54,0x08]
; X86-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rmbk_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandps (%rdi){1to16}, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x59,0x54,0x0f]
; X64-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <16 x float> undef, float %q, i32 0
  %b = shufflevector <16 x float> %vecinit.i, <16 x float> undef, <16 x i32> zeroinitializer
  %res = call <16 x float> @llvm.x86.avx512.mask.and.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_and_ps_rmbkz_512(<16 x float> %a, float* %ptr_b, i16 %mask) {
; X86-LABEL: test_mask_and_ps_rmbkz_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vandps (%eax){1to16}, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xd9,0x54,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_and_ps_rmbkz_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vandps (%rdi){1to16}, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xd9,0x54,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <16 x float> undef, float %q, i32 0
  %b = shufflevector <16 x float> %vecinit.i, <16 x float> undef, <16 x i32> zeroinitializer
  %res = call <16 x float> @llvm.x86.avx512.mask.and.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 %mask)
  ret <16 x float> %res
}

declare <16 x float> @llvm.x86.avx512.mask.and.ps.512(<16 x float>, <16 x float>, <16 x float>, i16)

define <4 x float> @test_mask_or_ps_rr_128(<4 x float> %a, <4 x float> %b) {
; CHECK-LABEL: test_mask_or_ps_rr_128:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vorps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x56,0xc1]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.avx512.mask.or.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 -1)
  ret <4 x float> %res
}

define <4 x float> @test_mask_or_ps_rrk_128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_or_ps_rrk_128:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vorps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x56,0xd1]
; X86-NEXT:    vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rrk_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vorps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x56,0xd1]
; X64-NEXT:    vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.avx512.mask.or.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_or_ps_rrkz_128(<4 x float> %a, <4 x float> %b, i8 %mask) {
; X86-LABEL: test_mask_or_ps_rrkz_128:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vorps %xmm1, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x56,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rrkz_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vorps %xmm1, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x56,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.avx512.mask.or.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_or_ps_rm_128(<4 x float> %a, <4 x float>* %ptr_b) {
; X86-LABEL: test_mask_or_ps_rm_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vorps (%eax), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x56,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rm_128:
; X64:       # %bb.0:
; X64-NEXT:    vorps (%rdi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x56,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <4 x float>, <4 x float>* %ptr_b
  %res = call <4 x float> @llvm.x86.avx512.mask.or.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 -1)
  ret <4 x float> %res
}

define <4 x float> @test_mask_or_ps_rmk_128(<4 x float> %a, <4 x float>* %ptr_b, <4 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_or_ps_rmk_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vorps (%eax), %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x56,0x08]
; X86-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rmk_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vorps (%rdi), %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x56,0x0f]
; X64-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <4 x float>, <4 x float>* %ptr_b
  %res = call <4 x float> @llvm.x86.avx512.mask.or.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_or_ps_rmkz_128(<4 x float> %a, <4 x float>* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_or_ps_rmkz_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vorps (%eax), %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x56,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rmkz_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vorps (%rdi), %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x56,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <4 x float>, <4 x float>* %ptr_b
  %res = call <4 x float> @llvm.x86.avx512.mask.or.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_or_ps_rmb_128(<4 x float> %a, float* %ptr_b) {
; X86-LABEL: test_mask_or_ps_rmb_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vorps (%eax){1to4}, %xmm0, %xmm0 # encoding: [0x62,0xf1,0x7c,0x18,0x56,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rmb_128:
; X64:       # %bb.0:
; X64-NEXT:    vorps (%rdi){1to4}, %xmm0, %xmm0 # encoding: [0x62,0xf1,0x7c,0x18,0x56,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <4 x float> undef, float %q, i32 0
  %b = shufflevector <4 x float> %vecinit.i, <4 x float> undef, <4 x i32> zeroinitializer
  %res = call <4 x float> @llvm.x86.avx512.mask.or.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 -1)
  ret <4 x float> %res
}

define <4 x float> @test_mask_or_ps_rmbk_128(<4 x float> %a, float* %ptr_b, <4 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_or_ps_rmbk_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vorps (%eax){1to4}, %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x19,0x56,0x08]
; X86-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rmbk_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vorps (%rdi){1to4}, %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x19,0x56,0x0f]
; X64-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <4 x float> undef, float %q, i32 0
  %b = shufflevector <4 x float> %vecinit.i, <4 x float> undef, <4 x i32> zeroinitializer
  %res = call <4 x float> @llvm.x86.avx512.mask.or.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_or_ps_rmbkz_128(<4 x float> %a, float* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_or_ps_rmbkz_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vorps (%eax){1to4}, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x99,0x56,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rmbkz_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vorps (%rdi){1to4}, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x99,0x56,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <4 x float> undef, float %q, i32 0
  %b = shufflevector <4 x float> %vecinit.i, <4 x float> undef, <4 x i32> zeroinitializer
  %res = call <4 x float> @llvm.x86.avx512.mask.or.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 %mask)
  ret <4 x float> %res
}

declare <4 x float> @llvm.x86.avx512.mask.or.ps.128(<4 x float>, <4 x float>, <4 x float>, i8)

define <8 x float> @test_mask_or_ps_rr_256(<8 x float> %a, <8 x float> %b) {
; CHECK-LABEL: test_mask_or_ps_rr_256:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vorps %ymm1, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x56,0xc1]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <8 x float> @llvm.x86.avx512.mask.or.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 -1)
  ret <8 x float> %res
}

define <8 x float> @test_mask_or_ps_rrk_256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_or_ps_rrk_256:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vorps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x56,0xd1]
; X86-NEXT:    vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rrk_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vorps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x56,0xd1]
; X64-NEXT:    vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x float> @llvm.x86.avx512.mask.or.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_or_ps_rrkz_256(<8 x float> %a, <8 x float> %b, i8 %mask) {
; X86-LABEL: test_mask_or_ps_rrkz_256:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vorps %ymm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x56,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rrkz_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vorps %ymm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x56,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x float> @llvm.x86.avx512.mask.or.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_or_ps_rm_256(<8 x float> %a, <8 x float>* %ptr_b) {
; X86-LABEL: test_mask_or_ps_rm_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vorps (%eax), %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x56,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rm_256:
; X64:       # %bb.0:
; X64-NEXT:    vorps (%rdi), %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x56,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <8 x float>, <8 x float>* %ptr_b
  %res = call <8 x float> @llvm.x86.avx512.mask.or.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 -1)
  ret <8 x float> %res
}

define <8 x float> @test_mask_or_ps_rmk_256(<8 x float> %a, <8 x float>* %ptr_b, <8 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_or_ps_rmk_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vorps (%eax), %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x56,0x08]
; X86-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rmk_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vorps (%rdi), %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x56,0x0f]
; X64-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <8 x float>, <8 x float>* %ptr_b
  %res = call <8 x float> @llvm.x86.avx512.mask.or.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_or_ps_rmkz_256(<8 x float> %a, <8 x float>* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_or_ps_rmkz_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vorps (%eax), %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x56,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rmkz_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vorps (%rdi), %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x56,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <8 x float>, <8 x float>* %ptr_b
  %res = call <8 x float> @llvm.x86.avx512.mask.or.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_or_ps_rmb_256(<8 x float> %a, float* %ptr_b) {
; X86-LABEL: test_mask_or_ps_rmb_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vorps (%eax){1to8}, %ymm0, %ymm0 # encoding: [0x62,0xf1,0x7c,0x38,0x56,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rmb_256:
; X64:       # %bb.0:
; X64-NEXT:    vorps (%rdi){1to8}, %ymm0, %ymm0 # encoding: [0x62,0xf1,0x7c,0x38,0x56,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <8 x float> undef, float %q, i32 0
  %b = shufflevector <8 x float> %vecinit.i, <8 x float> undef, <8 x i32> zeroinitializer
  %res = call <8 x float> @llvm.x86.avx512.mask.or.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 -1)
  ret <8 x float> %res
}

define <8 x float> @test_mask_or_ps_rmbk_256(<8 x float> %a, float* %ptr_b, <8 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_or_ps_rmbk_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vorps (%eax){1to8}, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x39,0x56,0x08]
; X86-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rmbk_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vorps (%rdi){1to8}, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x39,0x56,0x0f]
; X64-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <8 x float> undef, float %q, i32 0
  %b = shufflevector <8 x float> %vecinit.i, <8 x float> undef, <8 x i32> zeroinitializer
  %res = call <8 x float> @llvm.x86.avx512.mask.or.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_or_ps_rmbkz_256(<8 x float> %a, float* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_or_ps_rmbkz_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vorps (%eax){1to8}, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xb9,0x56,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rmbkz_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vorps (%rdi){1to8}, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xb9,0x56,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <8 x float> undef, float %q, i32 0
  %b = shufflevector <8 x float> %vecinit.i, <8 x float> undef, <8 x i32> zeroinitializer
  %res = call <8 x float> @llvm.x86.avx512.mask.or.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 %mask)
  ret <8 x float> %res
}

declare <8 x float> @llvm.x86.avx512.mask.or.ps.256(<8 x float>, <8 x float>, <8 x float>, i8)

define <16 x float> @test_mask_or_ps_rr_512(<16 x float> %a, <16 x float> %b) {
; CHECK-LABEL: test_mask_or_ps_rr_512:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vorps %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x56,0xc1]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <16 x float> @llvm.x86.avx512.mask.or.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 -1)
  ret <16 x float> %res
}

define <16 x float> @test_mask_or_ps_rrk_512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask) {
; X86-LABEL: test_mask_or_ps_rrk_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vorps %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x56,0xd1]
; X86-NEXT:    vmovaps %zmm2, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rrk_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vorps %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x56,0xd1]
; X64-NEXT:    vmovaps %zmm2, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <16 x float> @llvm.x86.avx512.mask.or.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_or_ps_rrkz_512(<16 x float> %a, <16 x float> %b, i16 %mask) {
; X86-LABEL: test_mask_or_ps_rrkz_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vorps %zmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x56,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rrkz_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vorps %zmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x56,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <16 x float> @llvm.x86.avx512.mask.or.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_or_ps_rm_512(<16 x float> %a, <16 x float>* %ptr_b) {
; X86-LABEL: test_mask_or_ps_rm_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vorps (%eax), %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x56,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rm_512:
; X64:       # %bb.0:
; X64-NEXT:    vorps (%rdi), %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x56,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <16 x float>, <16 x float>* %ptr_b
  %res = call <16 x float> @llvm.x86.avx512.mask.or.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 -1)
  ret <16 x float> %res
}

define <16 x float> @test_mask_or_ps_rmk_512(<16 x float> %a, <16 x float>* %ptr_b, <16 x float> %passThru, i16 %mask) {
; X86-LABEL: test_mask_or_ps_rmk_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vorps (%eax), %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x56,0x08]
; X86-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rmk_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vorps (%rdi), %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x56,0x0f]
; X64-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <16 x float>, <16 x float>* %ptr_b
  %res = call <16 x float> @llvm.x86.avx512.mask.or.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_or_ps_rmkz_512(<16 x float> %a, <16 x float>* %ptr_b, i16 %mask) {
; X86-LABEL: test_mask_or_ps_rmkz_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vorps (%eax), %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x56,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rmkz_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vorps (%rdi), %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x56,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <16 x float>, <16 x float>* %ptr_b
  %res = call <16 x float> @llvm.x86.avx512.mask.or.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_or_ps_rmb_512(<16 x float> %a, float* %ptr_b) {
; X86-LABEL: test_mask_or_ps_rmb_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vorps (%eax){1to16}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x58,0x56,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rmb_512:
; X64:       # %bb.0:
; X64-NEXT:    vorps (%rdi){1to16}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x58,0x56,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <16 x float> undef, float %q, i32 0
  %b = shufflevector <16 x float> %vecinit.i, <16 x float> undef, <16 x i32> zeroinitializer
  %res = call <16 x float> @llvm.x86.avx512.mask.or.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 -1)
  ret <16 x float> %res
}

define <16 x float> @test_mask_or_ps_rmbk_512(<16 x float> %a, float* %ptr_b, <16 x float> %passThru, i16 %mask) {
; X86-LABEL: test_mask_or_ps_rmbk_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vorps (%eax){1to16}, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x59,0x56,0x08]
; X86-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rmbk_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vorps (%rdi){1to16}, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x59,0x56,0x0f]
; X64-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <16 x float> undef, float %q, i32 0
  %b = shufflevector <16 x float> %vecinit.i, <16 x float> undef, <16 x i32> zeroinitializer
  %res = call <16 x float> @llvm.x86.avx512.mask.or.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_or_ps_rmbkz_512(<16 x float> %a, float* %ptr_b, i16 %mask) {
; X86-LABEL: test_mask_or_ps_rmbkz_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vorps (%eax){1to16}, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xd9,0x56,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_or_ps_rmbkz_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vorps (%rdi){1to16}, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xd9,0x56,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <16 x float> undef, float %q, i32 0
  %b = shufflevector <16 x float> %vecinit.i, <16 x float> undef, <16 x i32> zeroinitializer
  %res = call <16 x float> @llvm.x86.avx512.mask.or.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 %mask)
  ret <16 x float> %res
}

declare <16 x float> @llvm.x86.avx512.mask.or.ps.512(<16 x float>, <16 x float>, <16 x float>, i16)

define <4 x float> @test_mask_xor_ps_rr_128(<4 x float> %a, <4 x float> %b) {
; CHECK-LABEL: test_mask_xor_ps_rr_128:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vxorps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x57,0xc1]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.avx512.mask.xor.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 -1)
  ret <4 x float> %res
}

define <4 x float> @test_mask_xor_ps_rrk_128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_xor_ps_rrk_128:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vxorps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x57,0xd1]
; X86-NEXT:    vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rrk_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vxorps %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x57,0xd1]
; X64-NEXT:    vmovaps %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.avx512.mask.xor.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_xor_ps_rrkz_128(<4 x float> %a, <4 x float> %b, i8 %mask) {
; X86-LABEL: test_mask_xor_ps_rrkz_128:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vxorps %xmm1, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x57,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rrkz_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vxorps %xmm1, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x57,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.avx512.mask.xor.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_xor_ps_rm_128(<4 x float> %a, <4 x float>* %ptr_b) {
; X86-LABEL: test_mask_xor_ps_rm_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vxorps (%eax), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x57,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rm_128:
; X64:       # %bb.0:
; X64-NEXT:    vxorps (%rdi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x57,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <4 x float>, <4 x float>* %ptr_b
  %res = call <4 x float> @llvm.x86.avx512.mask.xor.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 -1)
  ret <4 x float> %res
}

define <4 x float> @test_mask_xor_ps_rmk_128(<4 x float> %a, <4 x float>* %ptr_b, <4 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_xor_ps_rmk_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vxorps (%eax), %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x57,0x08]
; X86-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rmk_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vxorps (%rdi), %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x09,0x57,0x0f]
; X64-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <4 x float>, <4 x float>* %ptr_b
  %res = call <4 x float> @llvm.x86.avx512.mask.xor.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_xor_ps_rmkz_128(<4 x float> %a, <4 x float>* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_xor_ps_rmkz_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vxorps (%eax), %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x57,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rmkz_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vxorps (%rdi), %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x89,0x57,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <4 x float>, <4 x float>* %ptr_b
  %res = call <4 x float> @llvm.x86.avx512.mask.xor.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_xor_ps_rmb_128(<4 x float> %a, float* %ptr_b) {
; X86-LABEL: test_mask_xor_ps_rmb_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vxorps (%eax){1to4}, %xmm0, %xmm0 # encoding: [0x62,0xf1,0x7c,0x18,0x57,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rmb_128:
; X64:       # %bb.0:
; X64-NEXT:    vxorps (%rdi){1to4}, %xmm0, %xmm0 # encoding: [0x62,0xf1,0x7c,0x18,0x57,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <4 x float> undef, float %q, i32 0
  %b = shufflevector <4 x float> %vecinit.i, <4 x float> undef, <4 x i32> zeroinitializer
  %res = call <4 x float> @llvm.x86.avx512.mask.xor.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 -1)
  ret <4 x float> %res
}

define <4 x float> @test_mask_xor_ps_rmbk_128(<4 x float> %a, float* %ptr_b, <4 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_xor_ps_rmbk_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vxorps (%eax){1to4}, %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x19,0x57,0x08]
; X86-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rmbk_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vxorps (%rdi){1to4}, %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x19,0x57,0x0f]
; X64-NEXT:    vmovaps %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <4 x float> undef, float %q, i32 0
  %b = shufflevector <4 x float> %vecinit.i, <4 x float> undef, <4 x i32> zeroinitializer
  %res = call <4 x float> @llvm.x86.avx512.mask.xor.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> %passThru, i8 %mask)
  ret <4 x float> %res
}

define <4 x float> @test_mask_xor_ps_rmbkz_128(<4 x float> %a, float* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_xor_ps_rmbkz_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vxorps (%eax){1to4}, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x99,0x57,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rmbkz_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vxorps (%rdi){1to4}, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0x99,0x57,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <4 x float> undef, float %q, i32 0
  %b = shufflevector <4 x float> %vecinit.i, <4 x float> undef, <4 x i32> zeroinitializer
  %res = call <4 x float> @llvm.x86.avx512.mask.xor.ps.128(<4 x float> %a, <4 x float> %b, <4 x float> zeroinitializer, i8 %mask)
  ret <4 x float> %res
}

declare <4 x float> @llvm.x86.avx512.mask.xor.ps.128(<4 x float>, <4 x float>, <4 x float>, i8)

define <8 x float> @test_mask_xor_ps_rr_256(<8 x float> %a, <8 x float> %b) {
; CHECK-LABEL: test_mask_xor_ps_rr_256:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vxorps %ymm1, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x57,0xc1]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <8 x float> @llvm.x86.avx512.mask.xor.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 -1)
  ret <8 x float> %res
}

define <8 x float> @test_mask_xor_ps_rrk_256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_xor_ps_rrk_256:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vxorps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x57,0xd1]
; X86-NEXT:    vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rrk_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vxorps %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x57,0xd1]
; X64-NEXT:    vmovaps %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x float> @llvm.x86.avx512.mask.xor.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_xor_ps_rrkz_256(<8 x float> %a, <8 x float> %b, i8 %mask) {
; X86-LABEL: test_mask_xor_ps_rrkz_256:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vxorps %ymm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x57,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rrkz_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vxorps %ymm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x57,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x float> @llvm.x86.avx512.mask.xor.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_xor_ps_rm_256(<8 x float> %a, <8 x float>* %ptr_b) {
; X86-LABEL: test_mask_xor_ps_rm_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vxorps (%eax), %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x57,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rm_256:
; X64:       # %bb.0:
; X64-NEXT:    vxorps (%rdi), %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x57,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <8 x float>, <8 x float>* %ptr_b
  %res = call <8 x float> @llvm.x86.avx512.mask.xor.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 -1)
  ret <8 x float> %res
}

define <8 x float> @test_mask_xor_ps_rmk_256(<8 x float> %a, <8 x float>* %ptr_b, <8 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_xor_ps_rmk_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vxorps (%eax), %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x57,0x08]
; X86-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rmk_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vxorps (%rdi), %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x29,0x57,0x0f]
; X64-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <8 x float>, <8 x float>* %ptr_b
  %res = call <8 x float> @llvm.x86.avx512.mask.xor.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_xor_ps_rmkz_256(<8 x float> %a, <8 x float>* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_xor_ps_rmkz_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vxorps (%eax), %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x57,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rmkz_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vxorps (%rdi), %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xa9,0x57,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <8 x float>, <8 x float>* %ptr_b
  %res = call <8 x float> @llvm.x86.avx512.mask.xor.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_xor_ps_rmb_256(<8 x float> %a, float* %ptr_b) {
; X86-LABEL: test_mask_xor_ps_rmb_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vxorps (%eax){1to8}, %ymm0, %ymm0 # encoding: [0x62,0xf1,0x7c,0x38,0x57,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rmb_256:
; X64:       # %bb.0:
; X64-NEXT:    vxorps (%rdi){1to8}, %ymm0, %ymm0 # encoding: [0x62,0xf1,0x7c,0x38,0x57,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <8 x float> undef, float %q, i32 0
  %b = shufflevector <8 x float> %vecinit.i, <8 x float> undef, <8 x i32> zeroinitializer
  %res = call <8 x float> @llvm.x86.avx512.mask.xor.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 -1)
  ret <8 x float> %res
}

define <8 x float> @test_mask_xor_ps_rmbk_256(<8 x float> %a, float* %ptr_b, <8 x float> %passThru, i8 %mask) {
; X86-LABEL: test_mask_xor_ps_rmbk_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vxorps (%eax){1to8}, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x39,0x57,0x08]
; X86-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rmbk_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vxorps (%rdi){1to8}, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x39,0x57,0x0f]
; X64-NEXT:    vmovaps %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <8 x float> undef, float %q, i32 0
  %b = shufflevector <8 x float> %vecinit.i, <8 x float> undef, <8 x i32> zeroinitializer
  %res = call <8 x float> @llvm.x86.avx512.mask.xor.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> %passThru, i8 %mask)
  ret <8 x float> %res
}

define <8 x float> @test_mask_xor_ps_rmbkz_256(<8 x float> %a, float* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_xor_ps_rmbkz_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vxorps (%eax){1to8}, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xb9,0x57,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rmbkz_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vxorps (%rdi){1to8}, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xb9,0x57,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <8 x float> undef, float %q, i32 0
  %b = shufflevector <8 x float> %vecinit.i, <8 x float> undef, <8 x i32> zeroinitializer
  %res = call <8 x float> @llvm.x86.avx512.mask.xor.ps.256(<8 x float> %a, <8 x float> %b, <8 x float> zeroinitializer, i8 %mask)
  ret <8 x float> %res
}

declare <8 x float> @llvm.x86.avx512.mask.xor.ps.256(<8 x float>, <8 x float>, <8 x float>, i8)

define <16 x float> @test_mask_xor_ps_rr_512(<16 x float> %a, <16 x float> %b) {
; CHECK-LABEL: test_mask_xor_ps_rr_512:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vxorps %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x57,0xc1]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <16 x float> @llvm.x86.avx512.mask.xor.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 -1)
  ret <16 x float> %res
}

define <16 x float> @test_mask_xor_ps_rrk_512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask) {
; X86-LABEL: test_mask_xor_ps_rrk_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vxorps %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x57,0xd1]
; X86-NEXT:    vmovaps %zmm2, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rrk_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vxorps %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x57,0xd1]
; X64-NEXT:    vmovaps %zmm2, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <16 x float> @llvm.x86.avx512.mask.xor.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_xor_ps_rrkz_512(<16 x float> %a, <16 x float> %b, i16 %mask) {
; X86-LABEL: test_mask_xor_ps_rrkz_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vxorps %zmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x57,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rrkz_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vxorps %zmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x57,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <16 x float> @llvm.x86.avx512.mask.xor.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_xor_ps_rm_512(<16 x float> %a, <16 x float>* %ptr_b) {
; X86-LABEL: test_mask_xor_ps_rm_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vxorps (%eax), %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x57,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rm_512:
; X64:       # %bb.0:
; X64-NEXT:    vxorps (%rdi), %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x57,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <16 x float>, <16 x float>* %ptr_b
  %res = call <16 x float> @llvm.x86.avx512.mask.xor.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 -1)
  ret <16 x float> %res
}

define <16 x float> @test_mask_xor_ps_rmk_512(<16 x float> %a, <16 x float>* %ptr_b, <16 x float> %passThru, i16 %mask) {
; X86-LABEL: test_mask_xor_ps_rmk_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vxorps (%eax), %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x57,0x08]
; X86-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rmk_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vxorps (%rdi), %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x49,0x57,0x0f]
; X64-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <16 x float>, <16 x float>* %ptr_b
  %res = call <16 x float> @llvm.x86.avx512.mask.xor.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_xor_ps_rmkz_512(<16 x float> %a, <16 x float>* %ptr_b, i16 %mask) {
; X86-LABEL: test_mask_xor_ps_rmkz_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vxorps (%eax), %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x57,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rmkz_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vxorps (%rdi), %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xc9,0x57,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <16 x float>, <16 x float>* %ptr_b
  %res = call <16 x float> @llvm.x86.avx512.mask.xor.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_xor_ps_rmb_512(<16 x float> %a, float* %ptr_b) {
; X86-LABEL: test_mask_xor_ps_rmb_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vxorps (%eax){1to16}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x58,0x57,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rmb_512:
; X64:       # %bb.0:
; X64-NEXT:    vxorps (%rdi){1to16}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0x7c,0x58,0x57,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <16 x float> undef, float %q, i32 0
  %b = shufflevector <16 x float> %vecinit.i, <16 x float> undef, <16 x i32> zeroinitializer
  %res = call <16 x float> @llvm.x86.avx512.mask.xor.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 -1)
  ret <16 x float> %res
}

define <16 x float> @test_mask_xor_ps_rmbk_512(<16 x float> %a, float* %ptr_b, <16 x float> %passThru, i16 %mask) {
; X86-LABEL: test_mask_xor_ps_rmbk_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vxorps (%eax){1to16}, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x59,0x57,0x08]
; X86-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rmbk_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vxorps (%rdi){1to16}, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7c,0x59,0x57,0x0f]
; X64-NEXT:    vmovaps %zmm1, %zmm0 # encoding: [0x62,0xf1,0x7c,0x48,0x28,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <16 x float> undef, float %q, i32 0
  %b = shufflevector <16 x float> %vecinit.i, <16 x float> undef, <16 x i32> zeroinitializer
  %res = call <16 x float> @llvm.x86.avx512.mask.xor.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask)
  ret <16 x float> %res
}

define <16 x float> @test_mask_xor_ps_rmbkz_512(<16 x float> %a, float* %ptr_b, i16 %mask) {
; X86-LABEL: test_mask_xor_ps_rmbkz_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vxorps (%eax){1to16}, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xd9,0x57,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_xor_ps_rmbkz_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vxorps (%rdi){1to16}, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf1,0x7c,0xd9,0x57,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load float, float* %ptr_b
  %vecinit.i = insertelement <16 x float> undef, float %q, i32 0
  %b = shufflevector <16 x float> %vecinit.i, <16 x float> undef, <16 x i32> zeroinitializer
  %res = call <16 x float> @llvm.x86.avx512.mask.xor.ps.512(<16 x float> %a, <16 x float> %b, <16 x float> zeroinitializer, i16 %mask)
  ret <16 x float> %res
}

declare <16 x float> @llvm.x86.avx512.mask.xor.ps.512(<16 x float>, <16 x float>, <16 x float>, i16)

define <8 x i64> @test_mask_mullo_epi64_rr_512(<8 x i64> %a, <8 x i64> %b) {
; CHECK-LABEL: test_mask_mullo_epi64_rr_512:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vpmullq %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf2,0xfd,0x48,0x40,0xc1]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <8 x i64> @llvm.x86.avx512.mask.pmull.q.512(<8 x i64> %a, <8 x i64> %b, <8 x i64> zeroinitializer, i8 -1)
  ret <8 x i64> %res
}

define <8 x i64> @test_mask_mullo_epi64_rrk_512(<8 x i64> %a, <8 x i64> %b, <8 x i64> %passThru, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rrk_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vpmullq %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x40,0xd1]
; X86-NEXT:    vmovdqa64 %zmm2, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rrk_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vpmullq %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x40,0xd1]
; X64-NEXT:    vmovdqa64 %zmm2, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x i64> @llvm.x86.avx512.mask.pmull.q.512(<8 x i64> %a, <8 x i64> %b, <8 x i64> %passThru, i8 %mask)
  ret <8 x i64> %res
}

define <8 x i64> @test_mask_mullo_epi64_rrkz_512(<8 x i64> %a, <8 x i64> %b, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rrkz_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vpmullq %zmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xc9,0x40,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rrkz_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vpmullq %zmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xc9,0x40,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x i64> @llvm.x86.avx512.mask.pmull.q.512(<8 x i64> %a, <8 x i64> %b, <8 x i64> zeroinitializer, i8 %mask)
  ret <8 x i64> %res
}

define <8 x i64> @test_mask_mullo_epi64_rm_512(<8 x i64> %a, <8 x i64>* %ptr_b) {
; X86-LABEL: test_mask_mullo_epi64_rm_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vpmullq (%eax), %zmm0, %zmm0 # encoding: [0x62,0xf2,0xfd,0x48,0x40,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rm_512:
; X64:       # %bb.0:
; X64-NEXT:    vpmullq (%rdi), %zmm0, %zmm0 # encoding: [0x62,0xf2,0xfd,0x48,0x40,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <8 x i64>, <8 x i64>* %ptr_b
  %res = call <8 x i64> @llvm.x86.avx512.mask.pmull.q.512(<8 x i64> %a, <8 x i64> %b, <8 x i64> zeroinitializer, i8 -1)
  ret <8 x i64> %res
}

define <8 x i64> @test_mask_mullo_epi64_rmk_512(<8 x i64> %a, <8 x i64>* %ptr_b, <8 x i64> %passThru, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rmk_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vpmullq (%eax), %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x40,0x08]
; X86-NEXT:    vmovdqa64 %zmm1, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rmk_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpmullq (%rdi), %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x40,0x0f]
; X64-NEXT:    vmovdqa64 %zmm1, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <8 x i64>, <8 x i64>* %ptr_b
  %res = call <8 x i64> @llvm.x86.avx512.mask.pmull.q.512(<8 x i64> %a, <8 x i64> %b, <8 x i64> %passThru, i8 %mask)
  ret <8 x i64> %res
}

define <8 x i64> @test_mask_mullo_epi64_rmkz_512(<8 x i64> %a, <8 x i64>* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rmkz_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vpmullq (%eax), %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xc9,0x40,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rmkz_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpmullq (%rdi), %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xc9,0x40,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <8 x i64>, <8 x i64>* %ptr_b
  %res = call <8 x i64> @llvm.x86.avx512.mask.pmull.q.512(<8 x i64> %a, <8 x i64> %b, <8 x i64> zeroinitializer, i8 %mask)
  ret <8 x i64> %res
}

define <8 x i64> @test_mask_mullo_epi64_rmb_512(<8 x i64> %a, i64* %ptr_b) {
; X86-LABEL: test_mask_mullo_epi64_rmb_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vmovq (%eax), %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x7e,0x08]
; X86-NEXT:    # xmm1 = mem[0],zero
; X86-NEXT:    vpbroadcastq %xmm1, %zmm1 # encoding: [0x62,0xf2,0xfd,0x48,0x59,0xc9]
; X86-NEXT:    vpmullq %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf2,0xfd,0x48,0x40,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rmb_512:
; X64:       # %bb.0:
; X64-NEXT:    vpmullq (%rdi){1to8}, %zmm0, %zmm0 # encoding: [0x62,0xf2,0xfd,0x58,0x40,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load i64, i64* %ptr_b
  %vecinit.i = insertelement <8 x i64> undef, i64 %q, i32 0
  %b = shufflevector <8 x i64> %vecinit.i, <8 x i64> undef, <8 x i32> zeroinitializer
  %res = call <8 x i64> @llvm.x86.avx512.mask.pmull.q.512(<8 x i64> %a, <8 x i64> %b, <8 x i64> zeroinitializer, i8 -1)
  ret <8 x i64> %res
}

define <8 x i64> @test_mask_mullo_epi64_rmbk_512(<8 x i64> %a, i64* %ptr_b, <8 x i64> %passThru, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rmbk_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vmovq (%eax), %xmm2 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x7e,0x10]
; X86-NEXT:    # xmm2 = mem[0],zero
; X86-NEXT:    vpbroadcastq %xmm2, %zmm2 # encoding: [0x62,0xf2,0xfd,0x48,0x59,0xd2]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vpmullq %zmm2, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf2,0xfd,0x49,0x40,0xca]
; X86-NEXT:    vmovdqa64 %zmm1, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rmbk_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpmullq (%rdi){1to8}, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf2,0xfd,0x59,0x40,0x0f]
; X64-NEXT:    vmovdqa64 %zmm1, %zmm0 # encoding: [0x62,0xf1,0xfd,0x48,0x6f,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load i64, i64* %ptr_b
  %vecinit.i = insertelement <8 x i64> undef, i64 %q, i32 0
  %b = shufflevector <8 x i64> %vecinit.i, <8 x i64> undef, <8 x i32> zeroinitializer
  %res = call <8 x i64> @llvm.x86.avx512.mask.pmull.q.512(<8 x i64> %a, <8 x i64> %b, <8 x i64> %passThru, i8 %mask)
  ret <8 x i64> %res
}

define <8 x i64> @test_mask_mullo_epi64_rmbkz_512(<8 x i64> %a, i64* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rmbkz_512:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vmovq (%eax), %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x7e,0x08]
; X86-NEXT:    # xmm1 = mem[0],zero
; X86-NEXT:    vpbroadcastq %xmm1, %zmm1 # encoding: [0x62,0xf2,0xfd,0x48,0x59,0xc9]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vpmullq %zmm1, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xc9,0x40,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rmbkz_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpmullq (%rdi){1to8}, %zmm0, %zmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xd9,0x40,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load i64, i64* %ptr_b
  %vecinit.i = insertelement <8 x i64> undef, i64 %q, i32 0
  %b = shufflevector <8 x i64> %vecinit.i, <8 x i64> undef, <8 x i32> zeroinitializer
  %res = call <8 x i64> @llvm.x86.avx512.mask.pmull.q.512(<8 x i64> %a, <8 x i64> %b, <8 x i64> zeroinitializer, i8 %mask)
  ret <8 x i64> %res
}
declare <8 x i64> @llvm.x86.avx512.mask.pmull.q.512(<8 x i64>, <8 x i64>, <8 x i64>, i8)

define <4 x i64> @test_mask_mullo_epi64_rr_256(<4 x i64> %a, <4 x i64> %b) {
; CHECK-LABEL: test_mask_mullo_epi64_rr_256:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vpmullq %ymm1, %ymm0, %ymm0 # encoding: [0x62,0xf2,0xfd,0x28,0x40,0xc1]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x i64> @llvm.x86.avx512.mask.pmull.q.256(<4 x i64> %a, <4 x i64> %b, <4 x i64> zeroinitializer, i8 -1)
  ret <4 x i64> %res
}

define <4 x i64> @test_mask_mullo_epi64_rrk_256(<4 x i64> %a, <4 x i64> %b, <4 x i64> %passThru, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rrk_256:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vpmullq %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x40,0xd1]
; X86-NEXT:    vmovdqa %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rrk_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vpmullq %ymm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x40,0xd1]
; X64-NEXT:    vmovdqa %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x i64> @llvm.x86.avx512.mask.pmull.q.256(<4 x i64> %a, <4 x i64> %b, <4 x i64> %passThru, i8 %mask)
  ret <4 x i64> %res
}

define <4 x i64> @test_mask_mullo_epi64_rrkz_256(<4 x i64> %a, <4 x i64> %b, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rrkz_256:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vpmullq %ymm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xa9,0x40,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rrkz_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vpmullq %ymm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xa9,0x40,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x i64> @llvm.x86.avx512.mask.pmull.q.256(<4 x i64> %a, <4 x i64> %b, <4 x i64> zeroinitializer, i8 %mask)
  ret <4 x i64> %res
}

define <4 x i64> @test_mask_mullo_epi64_rm_256(<4 x i64> %a, <4 x i64>* %ptr_b) {
; X86-LABEL: test_mask_mullo_epi64_rm_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vpmullq (%eax), %ymm0, %ymm0 # encoding: [0x62,0xf2,0xfd,0x28,0x40,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rm_256:
; X64:       # %bb.0:
; X64-NEXT:    vpmullq (%rdi), %ymm0, %ymm0 # encoding: [0x62,0xf2,0xfd,0x28,0x40,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <4 x i64>, <4 x i64>* %ptr_b
  %res = call <4 x i64> @llvm.x86.avx512.mask.pmull.q.256(<4 x i64> %a, <4 x i64> %b, <4 x i64> zeroinitializer, i8 -1)
  ret <4 x i64> %res
}

define <4 x i64> @test_mask_mullo_epi64_rmk_256(<4 x i64> %a, <4 x i64>* %ptr_b, <4 x i64> %passThru, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rmk_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vpmullq (%eax), %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x40,0x08]
; X86-NEXT:    vmovdqa %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rmk_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpmullq (%rdi), %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x40,0x0f]
; X64-NEXT:    vmovdqa %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <4 x i64>, <4 x i64>* %ptr_b
  %res = call <4 x i64> @llvm.x86.avx512.mask.pmull.q.256(<4 x i64> %a, <4 x i64> %b, <4 x i64> %passThru, i8 %mask)
  ret <4 x i64> %res
}

define <4 x i64> @test_mask_mullo_epi64_rmkz_256(<4 x i64> %a, <4 x i64>* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rmkz_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vpmullq (%eax), %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xa9,0x40,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rmkz_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpmullq (%rdi), %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xa9,0x40,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <4 x i64>, <4 x i64>* %ptr_b
  %res = call <4 x i64> @llvm.x86.avx512.mask.pmull.q.256(<4 x i64> %a, <4 x i64> %b, <4 x i64> zeroinitializer, i8 %mask)
  ret <4 x i64> %res
}

define <4 x i64> @test_mask_mullo_epi64_rmb_256(<4 x i64> %a, i64* %ptr_b) {
; X86-LABEL: test_mask_mullo_epi64_rmb_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vmovq (%eax), %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x7e,0x08]
; X86-NEXT:    # xmm1 = mem[0],zero
; X86-NEXT:    vpbroadcastq %xmm1, %ymm1 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x7d,0x59,0xc9]
; X86-NEXT:    vpmullq %ymm1, %ymm0, %ymm0 # encoding: [0x62,0xf2,0xfd,0x28,0x40,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rmb_256:
; X64:       # %bb.0:
; X64-NEXT:    vpmullq (%rdi){1to4}, %ymm0, %ymm0 # encoding: [0x62,0xf2,0xfd,0x38,0x40,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load i64, i64* %ptr_b
  %vecinit.i = insertelement <4 x i64> undef, i64 %q, i32 0
  %b = shufflevector <4 x i64> %vecinit.i, <4 x i64> undef, <4 x i32> zeroinitializer
  %res = call <4 x i64> @llvm.x86.avx512.mask.pmull.q.256(<4 x i64> %a, <4 x i64> %b, <4 x i64> zeroinitializer, i8 -1)
  ret <4 x i64> %res
}

define <4 x i64> @test_mask_mullo_epi64_rmbk_256(<4 x i64> %a, i64* %ptr_b, <4 x i64> %passThru, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rmbk_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vmovq (%eax), %xmm2 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x7e,0x10]
; X86-NEXT:    # xmm2 = mem[0],zero
; X86-NEXT:    vpbroadcastq %xmm2, %ymm2 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x7d,0x59,0xd2]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vpmullq %ymm2, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x40,0xca]
; X86-NEXT:    vmovdqa %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rmbk_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpmullq (%rdi){1to4}, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf2,0xfd,0x39,0x40,0x0f]
; X64-NEXT:    vmovdqa %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x6f,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load i64, i64* %ptr_b
  %vecinit.i = insertelement <4 x i64> undef, i64 %q, i32 0
  %b = shufflevector <4 x i64> %vecinit.i, <4 x i64> undef, <4 x i32> zeroinitializer
  %res = call <4 x i64> @llvm.x86.avx512.mask.pmull.q.256(<4 x i64> %a, <4 x i64> %b, <4 x i64> %passThru, i8 %mask)
  ret <4 x i64> %res
}

define <4 x i64> @test_mask_mullo_epi64_rmbkz_256(<4 x i64> %a, i64* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rmbkz_256:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vmovq (%eax), %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x7e,0x08]
; X86-NEXT:    # xmm1 = mem[0],zero
; X86-NEXT:    vpbroadcastq %xmm1, %ymm1 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x7d,0x59,0xc9]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vpmullq %ymm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xa9,0x40,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rmbkz_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpmullq (%rdi){1to4}, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0xb9,0x40,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load i64, i64* %ptr_b
  %vecinit.i = insertelement <4 x i64> undef, i64 %q, i32 0
  %b = shufflevector <4 x i64> %vecinit.i, <4 x i64> undef, <4 x i32> zeroinitializer
  %res = call <4 x i64> @llvm.x86.avx512.mask.pmull.q.256(<4 x i64> %a, <4 x i64> %b, <4 x i64> zeroinitializer, i8 %mask)
  ret <4 x i64> %res
}

declare <4 x i64> @llvm.x86.avx512.mask.pmull.q.256(<4 x i64>, <4 x i64>, <4 x i64>, i8)

define <2 x i64> @test_mask_mullo_epi64_rr_128(<2 x i64> %a, <2 x i64> %b) {
; CHECK-LABEL: test_mask_mullo_epi64_rr_128:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vpmullq %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf2,0xfd,0x08,0x40,0xc1]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <2 x i64> @llvm.x86.avx512.mask.pmull.q.128(<2 x i64> %a, <2 x i64> %b, <2 x i64> zeroinitializer, i8 -1)
  ret <2 x i64> %res
}

define <2 x i64> @test_mask_mullo_epi64_rrk_128(<2 x i64> %a, <2 x i64> %b, <2 x i64> %passThru, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rrk_128:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vpmullq %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0x40,0xd1]
; X86-NEXT:    vmovdqa %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rrk_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vpmullq %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0x40,0xd1]
; X64-NEXT:    vmovdqa %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <2 x i64> @llvm.x86.avx512.mask.pmull.q.128(<2 x i64> %a, <2 x i64> %b, <2 x i64> %passThru, i8 %mask)
  ret <2 x i64> %res
}

define <2 x i64> @test_mask_mullo_epi64_rrkz_128(<2 x i64> %a, <2 x i64> %b, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rrkz_128:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vpmullq %xmm1, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0x89,0x40,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rrkz_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vpmullq %xmm1, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0x89,0x40,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <2 x i64> @llvm.x86.avx512.mask.pmull.q.128(<2 x i64> %a, <2 x i64> %b, <2 x i64> zeroinitializer, i8 %mask)
  ret <2 x i64> %res
}

define <2 x i64> @test_mask_mullo_epi64_rm_128(<2 x i64> %a, <2 x i64>* %ptr_b) {
; X86-LABEL: test_mask_mullo_epi64_rm_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vpmullq (%eax), %xmm0, %xmm0 # encoding: [0x62,0xf2,0xfd,0x08,0x40,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rm_128:
; X64:       # %bb.0:
; X64-NEXT:    vpmullq (%rdi), %xmm0, %xmm0 # encoding: [0x62,0xf2,0xfd,0x08,0x40,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <2 x i64>, <2 x i64>* %ptr_b
  %res = call <2 x i64> @llvm.x86.avx512.mask.pmull.q.128(<2 x i64> %a, <2 x i64> %b, <2 x i64> zeroinitializer, i8 -1)
  ret <2 x i64> %res
}

define <2 x i64> @test_mask_mullo_epi64_rmk_128(<2 x i64> %a, <2 x i64>* %ptr_b, <2 x i64> %passThru, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rmk_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vpmullq (%eax), %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0x40,0x08]
; X86-NEXT:    vmovdqa %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rmk_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpmullq (%rdi), %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0x40,0x0f]
; X64-NEXT:    vmovdqa %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <2 x i64>, <2 x i64>* %ptr_b
  %res = call <2 x i64> @llvm.x86.avx512.mask.pmull.q.128(<2 x i64> %a, <2 x i64> %b, <2 x i64> %passThru, i8 %mask)
  ret <2 x i64> %res
}

define <2 x i64> @test_mask_mullo_epi64_rmkz_128(<2 x i64> %a, <2 x i64>* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rmkz_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vpmullq (%eax), %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0x89,0x40,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rmkz_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpmullq (%rdi), %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0x89,0x40,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %b = load <2 x i64>, <2 x i64>* %ptr_b
  %res = call <2 x i64> @llvm.x86.avx512.mask.pmull.q.128(<2 x i64> %a, <2 x i64> %b, <2 x i64> zeroinitializer, i8 %mask)
  ret <2 x i64> %res
}

define <2 x i64> @test_mask_mullo_epi64_rmb_128(<2 x i64> %a, i64* %ptr_b) {
; X86-LABEL: test_mask_mullo_epi64_rmb_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vpbroadcastq (%eax), %xmm1 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x59,0x08]
; X86-NEXT:    vpmullq %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf2,0xfd,0x08,0x40,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rmb_128:
; X64:       # %bb.0:
; X64-NEXT:    vpmullq (%rdi){1to2}, %xmm0, %xmm0 # encoding: [0x62,0xf2,0xfd,0x18,0x40,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load i64, i64* %ptr_b
  %vecinit.i = insertelement <2 x i64> undef, i64 %q, i32 0
  %b = shufflevector <2 x i64> %vecinit.i, <2 x i64> undef, <2 x i32> zeroinitializer
  %res = call <2 x i64> @llvm.x86.avx512.mask.pmull.q.128(<2 x i64> %a, <2 x i64> %b, <2 x i64> zeroinitializer, i8 -1)
  ret <2 x i64> %res
}

define <2 x i64> @test_mask_mullo_epi64_rmbk_128(<2 x i64> %a, i64* %ptr_b, <2 x i64> %passThru, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rmbk_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vpbroadcastq (%eax), %xmm2 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x59,0x10]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vpmullq %xmm2, %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf2,0xfd,0x09,0x40,0xca]
; X86-NEXT:    vmovdqa %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rmbk_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpmullq (%rdi){1to2}, %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf2,0xfd,0x19,0x40,0x0f]
; X64-NEXT:    vmovdqa %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x6f,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load i64, i64* %ptr_b
  %vecinit.i = insertelement <2 x i64> undef, i64 %q, i32 0
  %b = shufflevector <2 x i64> %vecinit.i, <2 x i64> undef, <2 x i32> zeroinitializer
  %res = call <2 x i64> @llvm.x86.avx512.mask.pmull.q.128(<2 x i64> %a, <2 x i64> %b, <2 x i64> %passThru, i8 %mask)
  ret <2 x i64> %res
}

define <2 x i64> @test_mask_mullo_epi64_rmbkz_128(<2 x i64> %a, i64* %ptr_b, i8 %mask) {
; X86-LABEL: test_mask_mullo_epi64_rmbkz_128:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vpbroadcastq (%eax), %xmm1 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x59,0x08]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vpmullq %xmm1, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0x89,0x40,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mask_mullo_epi64_rmbkz_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vpmullq (%rdi){1to2}, %xmm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf2,0xfd,0x99,0x40,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  %q = load i64, i64* %ptr_b
  %vecinit.i = insertelement <2 x i64> undef, i64 %q, i32 0
  %b = shufflevector <2 x i64> %vecinit.i, <2 x i64> undef, <2 x i32> zeroinitializer
  %res = call <2 x i64> @llvm.x86.avx512.mask.pmull.q.128(<2 x i64> %a, <2 x i64> %b, <2 x i64> zeroinitializer, i8 %mask)
  ret <2 x i64> %res
}

declare <2 x i64> @llvm.x86.avx512.mask.pmull.q.128(<2 x i64>, <2 x i64>, <2 x i64>, i8)

declare <2 x double> @llvm.x86.avx512.mask.vextractf64x2.256(<4 x double>, i32, <2 x double>, i8)

define <2 x double>@test_int_x86_avx512_mask_vextractf64x2_256(<4 x double> %x0, <2 x double> %x2, i8 %x3) {
; X86-LABEL: test_int_x86_avx512_mask_vextractf64x2_256:
; X86:       # %bb.0:
; X86-NEXT:    vextractf128 $1, %ymm0, %xmm2 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x19,0xc2,0x01]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vextractf64x2 $1, %ymm0, %xmm1 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0x19,0xc1,0x01]
; X86-NEXT:    vaddpd %xmm2, %xmm1, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf1,0x58,0xca]
; X86-NEXT:    vextractf64x2 $1, %ymm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0x19,0xc0,0x01]
; X86-NEXT:    vaddpd %xmm0, %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf1,0x58,0xc0]
; X86-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_vextractf64x2_256:
; X64:       # %bb.0:
; X64-NEXT:    vextractf128 $1, %ymm0, %xmm2 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x19,0xc2,0x01]
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vextractf64x2 $1, %ymm0, %xmm1 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0x19,0xc1,0x01]
; X64-NEXT:    vaddpd %xmm2, %xmm1, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf1,0x58,0xca]
; X64-NEXT:    vextractf64x2 $1, %ymm0, %xmm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0x19,0xc0,0x01]
; X64-NEXT:    vaddpd %xmm0, %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf1,0x58,0xc0]
; X64-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <2 x double> @llvm.x86.avx512.mask.vextractf64x2.256(<4 x double> %x0,i32 1, <2 x double> %x2, i8 %x3)
  %res2 = call <2 x double> @llvm.x86.avx512.mask.vextractf64x2.256(<4 x double> %x0,i32 1, <2 x double> zeroinitializer, i8 %x3)
  %res1 = call <2 x double> @llvm.x86.avx512.mask.vextractf64x2.256(<4 x double> %x0,i32 1, <2 x double> zeroinitializer, i8 -1)
  %res3 = fadd <2 x double> %res, %res1
  %res4 = fadd <2 x double> %res3, %res2
  ret <2 x double> %res4
}

declare <4 x double> @llvm.x86.avx512.mask.insertf64x2.256(<4 x double>, <2 x double>, i32, <4 x double>, i8)

define <4 x double>@test_int_x86_avx512_mask_insertf64x2_256(<4 x double> %x0, <2 x double> %x1, <4 x double> %x3, i8 %x4) {
; X86-LABEL: test_int_x86_avx512_mask_insertf64x2_256:
; X86:       # %bb.0:
; X86-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x18,0xd9,0x01]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vinsertf64x2 $1, %xmm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0x18,0xd1,0x01]
; X86-NEXT:    vaddpd %ymm3, %ymm2, %ymm2 # EVEX TO VEX Compression encoding: [0xc5,0xed,0x58,0xd3]
; X86-NEXT:    vinsertf64x2 $1, %xmm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0x18,0xc1,0x01]
; X86-NEXT:    vaddpd %ymm2, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x58,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_insertf64x2_256:
; X64:       # %bb.0:
; X64-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x18,0xd9,0x01]
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vinsertf64x2 $1, %xmm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0x18,0xd1,0x01]
; X64-NEXT:    vaddpd %ymm3, %ymm2, %ymm2 # EVEX TO VEX Compression encoding: [0xc5,0xed,0x58,0xd3]
; X64-NEXT:    vinsertf64x2 $1, %xmm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0x18,0xc1,0x01]
; X64-NEXT:    vaddpd %ymm2, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x58,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x double> @llvm.x86.avx512.mask.insertf64x2.256(<4 x double> %x0, <2 x double> %x1, i32 1, <4 x double> %x3, i8 %x4)
  %res1 = call <4 x double> @llvm.x86.avx512.mask.insertf64x2.256(<4 x double> %x0, <2 x double> %x1, i32 1, <4 x double> %x3, i8 -1)
  %res2 = call <4 x double> @llvm.x86.avx512.mask.insertf64x2.256(<4 x double> %x0, <2 x double> %x1, i32 1, <4 x double> zeroinitializer, i8 %x4)
  %res3 = fadd <4 x double> %res, %res1
  %res4 = fadd <4 x double> %res2, %res3
  ret <4 x double> %res4
}

declare <4 x i64> @llvm.x86.avx512.mask.inserti64x2.256(<4 x i64>, <2 x i64>, i32, <4 x i64>, i8)

define <4 x i64>@test_int_x86_avx512_mask_inserti64x2_256(<4 x i64> %x0, <2 x i64> %x1, <4 x i64> %x3, i8 %x4) {
; X86-LABEL: test_int_x86_avx512_mask_inserti64x2_256:
; X86:       # %bb.0:
; X86-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x38,0xd9,0x01]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vinserti64x2 $1, %xmm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0x38,0xd1,0x01]
; X86-NEXT:    vinserti64x2 $1, %xmm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0x38,0xc1,0x01]
; X86-NEXT:    vpaddq %ymm0, %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xe5,0xd4,0xc0]
; X86-NEXT:    vpaddq %ymm0, %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xed,0xd4,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_inserti64x2_256:
; X64:       # %bb.0:
; X64-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm3 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x38,0xd9,0x01]
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vinserti64x2 $1, %xmm1, %ymm0, %ymm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0x38,0xd1,0x01]
; X64-NEXT:    vinserti64x2 $1, %xmm1, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0x38,0xc1,0x01]
; X64-NEXT:    vpaddq %ymm0, %ymm3, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xe5,0xd4,0xc0]
; X64-NEXT:    vpaddq %ymm0, %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xed,0xd4,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x i64> @llvm.x86.avx512.mask.inserti64x2.256(<4 x i64> %x0, <2 x i64> %x1, i32 1, <4 x i64> %x3, i8 %x4)
  %res1 = call <4 x i64> @llvm.x86.avx512.mask.inserti64x2.256(<4 x i64> %x0, <2 x i64> %x1, i32 1, <4 x i64> %x3, i8 -1)
  %res2 = call <4 x i64> @llvm.x86.avx512.mask.inserti64x2.256(<4 x i64> %x0, <2 x i64> %x1, i32 1, <4 x i64> zeroinitializer, i8 %x4)
  %res3 = add <4 x i64> %res, %res1
  %res4 = add <4 x i64> %res3, %res2
  ret <4 x i64> %res4
}

declare <4 x i32> @llvm.x86.avx512.cvtmask2d.128(i8)

define <4 x i32>@test_int_x86_avx512_cvtmask2d_128(i8 %x0) {
; X86-LABEL: test_int_x86_avx512_cvtmask2d_128:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k0 # encoding: [0xc5,0xf9,0x90,0x44,0x24,0x04]
; X86-NEXT:    vpmovm2d %k0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_cvtmask2d_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k0 # encoding: [0xc5,0xf8,0x92,0xc7]
; X64-NEXT:    vpmovm2d %k0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x i32> @llvm.x86.avx512.cvtmask2d.128(i8 %x0)
  ret <4 x i32> %res
}

declare <8 x i32> @llvm.x86.avx512.cvtmask2d.256(i8)

define <8 x i32>@test_int_x86_avx512_cvtmask2d_256(i8 %x0) {
; X86-LABEL: test_int_x86_avx512_cvtmask2d_256:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k0 # encoding: [0xc5,0xf9,0x90,0x44,0x24,0x04]
; X86-NEXT:    vpmovm2d %k0, %ymm0 # encoding: [0x62,0xf2,0x7e,0x28,0x38,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_cvtmask2d_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k0 # encoding: [0xc5,0xf8,0x92,0xc7]
; X64-NEXT:    vpmovm2d %k0, %ymm0 # encoding: [0x62,0xf2,0x7e,0x28,0x38,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x i32> @llvm.x86.avx512.cvtmask2d.256(i8 %x0)
  ret <8 x i32> %res
}

declare <2 x i64> @llvm.x86.avx512.cvtmask2q.128(i8)

define <2 x i64>@test_int_x86_avx512_cvtmask2q_128(i8 %x0) {
; X86-LABEL: test_int_x86_avx512_cvtmask2q_128:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k0 # encoding: [0xc5,0xf9,0x90,0x44,0x24,0x04]
; X86-NEXT:    vpmovm2q %k0, %xmm0 # encoding: [0x62,0xf2,0xfe,0x08,0x38,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_cvtmask2q_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k0 # encoding: [0xc5,0xf8,0x92,0xc7]
; X64-NEXT:    vpmovm2q %k0, %xmm0 # encoding: [0x62,0xf2,0xfe,0x08,0x38,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <2 x i64> @llvm.x86.avx512.cvtmask2q.128(i8 %x0)
  ret <2 x i64> %res
}

declare <4 x i64> @llvm.x86.avx512.cvtmask2q.256(i8)

define <4 x i64>@test_int_x86_avx512_cvtmask2q_256(i8 %x0) {
; X86-LABEL: test_int_x86_avx512_cvtmask2q_256:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k0 # encoding: [0xc5,0xf9,0x90,0x44,0x24,0x04]
; X86-NEXT:    vpmovm2q %k0, %ymm0 # encoding: [0x62,0xf2,0xfe,0x28,0x38,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_cvtmask2q_256:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k0 # encoding: [0xc5,0xf8,0x92,0xc7]
; X64-NEXT:    vpmovm2q %k0, %ymm0 # encoding: [0x62,0xf2,0xfe,0x28,0x38,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x i64> @llvm.x86.avx512.cvtmask2q.256(i8 %x0)
  ret <4 x i64> %res
}

declare <4 x double> @llvm.x86.avx512.mask.broadcastf64x2.256(<2 x double>, <4 x double>, i8)

define <4 x double>@test_int_x86_avx512_mask_broadcastf64x2_256(<2 x double> %x0, <4 x double> %x2, i8 %mask) {
; X86-LABEL: test_int_x86_avx512_mask_broadcastf64x2_256:
; X86:       # %bb.0:
; X86-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; X86-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm2 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x18,0xd0,0x01]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vinsertf64x2 $1, %xmm0, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0x18,0xc8,0x01]
; X86-NEXT:    vaddpd %ymm1, %ymm2, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xed,0x58,0xc9]
; X86-NEXT:    vinsertf64x2 $1, %xmm0, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0x18,0xc0,0x01]
; X86-NEXT:    vaddpd %ymm1, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x58,0xc1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_broadcastf64x2_256:
; X64:       # %bb.0:
; X64-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; X64-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm2 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x18,0xd0,0x01]
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vinsertf64x2 $1, %xmm0, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0x18,0xc8,0x01]
; X64-NEXT:    vaddpd %ymm1, %ymm2, %ymm1 # EVEX TO VEX Compression encoding: [0xc5,0xed,0x58,0xc9]
; X64-NEXT:    vinsertf64x2 $1, %xmm0, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0x18,0xc0,0x01]
; X64-NEXT:    vaddpd %ymm1, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0x58,0xc1]
; X64-NEXT:    retq # encoding: [0xc3]

  %res1 = call <4 x double> @llvm.x86.avx512.mask.broadcastf64x2.256(<2 x double> %x0, <4 x double> %x2, i8 -1)
  %res2 = call <4 x double> @llvm.x86.avx512.mask.broadcastf64x2.256(<2 x double> %x0, <4 x double> %x2, i8 %mask)
  %res3 = call <4 x double> @llvm.x86.avx512.mask.broadcastf64x2.256(<2 x double> %x0, <4 x double> zeroinitializer, i8 %mask)
  %res4 = fadd <4 x double> %res1, %res2
  %res5 = fadd <4 x double> %res3, %res4
  ret <4 x double> %res5
}

define <4 x double>@test_int_x86_avx512_mask_broadcastf64x2_256_load(<2 x double>* %x0ptr, <4 x double> %x2, i8 %mask) {
; X86-LABEL: test_int_x86_avx512_mask_broadcastf64x2_256_load:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vbroadcastf64x2 (%eax), %ymm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x1a,0x00]
; X86-NEXT:    # ymm0 {%k1} = mem[0,1,0,1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_broadcastf64x2_256_load:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vbroadcastf64x2 (%rdi), %ymm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x1a,0x07]
; X64-NEXT:    # ymm0 {%k1} = mem[0,1,0,1]
; X64-NEXT:    retq # encoding: [0xc3]

  %x0 = load <2 x double>, <2 x double>* %x0ptr
  %res = call <4 x double> @llvm.x86.avx512.mask.broadcastf64x2.256(<2 x double> %x0, <4 x double> %x2, i8 %mask)
  ret <4 x double> %res
}

declare <4 x i64> @llvm.x86.avx512.mask.broadcasti64x2.256(<2 x i64>, <4 x i64>, i8)

define <4 x i64>@test_int_x86_avx512_mask_broadcasti64x2_256(<2 x i64> %x0, <4 x i64> %x2, i8 %mask) {
; X86-LABEL: test_int_x86_avx512_mask_broadcasti64x2_256:
; X86:       # %bb.0:
; X86-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; X86-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm2 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x38,0xd0,0x01]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vinserti64x2 $1, %xmm0, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0x38,0xc8,0x01]
; X86-NEXT:    vinserti64x2 $1, %xmm0, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0x38,0xc0,0x01]
; X86-NEXT:    vpaddq %ymm0, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf5,0xd4,0xc0]
; X86-NEXT:    vpaddq %ymm0, %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xed,0xd4,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_broadcasti64x2_256:
; X64:       # %bb.0:
; X64-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; X64-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm2 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x38,0xd0,0x01]
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vinserti64x2 $1, %xmm0, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0x38,0xc8,0x01]
; X64-NEXT:    vinserti64x2 $1, %xmm0, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf3,0xfd,0xa9,0x38,0xc0,0x01]
; X64-NEXT:    vpaddq %ymm0, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf5,0xd4,0xc0]
; X64-NEXT:    vpaddq %ymm0, %ymm2, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xed,0xd4,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]

  %res1 = call <4 x i64> @llvm.x86.avx512.mask.broadcasti64x2.256(<2 x i64> %x0, <4 x i64> %x2, i8 -1)
  %res2 = call <4 x i64> @llvm.x86.avx512.mask.broadcasti64x2.256(<2 x i64> %x0, <4 x i64> %x2, i8 %mask)
  %res3 = call <4 x i64> @llvm.x86.avx512.mask.broadcasti64x2.256(<2 x i64> %x0, <4 x i64> zeroinitializer, i8 %mask)
  %res4 = add <4 x i64> %res1, %res2
  %res5 = add <4 x i64> %res3, %res4
  ret <4 x i64> %res5
}

define <4 x i64>@test_int_x86_avx512_mask_broadcasti64x2_256_load(<2 x i64>* %x0ptr, <4 x i64> %x2, i8 %mask) {
; X86-LABEL: test_int_x86_avx512_mask_broadcasti64x2_256_load:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x08]
; X86-NEXT:    vbroadcasti64x2 (%eax), %ymm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x5a,0x00]
; X86-NEXT:    # ymm0 {%k1} = mem[0,1,0,1]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_broadcasti64x2_256_load:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %esi, %k1 # encoding: [0xc5,0xf8,0x92,0xce]
; X64-NEXT:    vbroadcasti64x2 (%rdi), %ymm0 {%k1} # encoding: [0x62,0xf2,0xfd,0x29,0x5a,0x07]
; X64-NEXT:    # ymm0 {%k1} = mem[0,1,0,1]
; X64-NEXT:    retq # encoding: [0xc3]

  %x0 = load <2 x i64>, <2 x i64>* %x0ptr
  %res = call <4 x i64> @llvm.x86.avx512.mask.broadcasti64x2.256(<2 x i64> %x0, <4 x i64> %x2, i8 %mask)
  ret <4 x i64> %res
}

declare <8 x float> @llvm.x86.avx512.mask.broadcastf32x2.256(<4 x float>, <8 x float>, i8)

define <8 x float>@test_int_x86_avx512_mask_broadcastf32x2_256(<4 x float> %x0, <8 x float> %x2, i8 %x3) {
; X86-LABEL: test_int_x86_avx512_mask_broadcastf32x2_256:
; X86:       # %bb.0:
; X86-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; X86-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm2 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x18,0xd0,0x01]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vinsertf32x4 $1, %xmm0, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf3,0x7d,0x29,0x18,0xc8,0x01]
; X86-NEXT:    vinsertf32x4 $1, %xmm0, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xa9,0x18,0xc0,0x01]
; X86-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf4,0x58,0xc0]
; X86-NEXT:    vaddps %ymm2, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x58,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_broadcastf32x2_256:
; X64:       # %bb.0:
; X64-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; X64-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm2 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x18,0xd0,0x01]
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vinsertf32x4 $1, %xmm0, %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf3,0x7d,0x29,0x18,0xc8,0x01]
; X64-NEXT:    vinsertf32x4 $1, %xmm0, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xa9,0x18,0xc0,0x01]
; X64-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf4,0x58,0xc0]
; X64-NEXT:    vaddps %ymm2, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfc,0x58,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x float> @llvm.x86.avx512.mask.broadcastf32x2.256(<4 x float>  %x0, <8 x float> %x2, i8 %x3)
  %res1 = call <8 x float> @llvm.x86.avx512.mask.broadcastf32x2.256(<4 x float> %x0, <8 x float> zeroinitializer, i8 %x3)
  %res2 = call <8 x float> @llvm.x86.avx512.mask.broadcastf32x2.256(<4 x float> %x0, <8 x float> %x2, i8 -1)
  %res3 = fadd <8 x float> %res, %res1
  %res4 = fadd <8 x float> %res3, %res2
  ret <8 x float> %res4
}

declare <8 x i32> @llvm.x86.avx512.mask.broadcasti32x2.256(<4 x i32>, <8 x i32>, i8)

define <8 x i32>@test_int_x86_avx512_mask_broadcasti32x2_256(<4 x i32> %x0, <8 x i32> %x2, i8 %x3, i64 * %y_ptr) {
; X86-LABEL: test_int_x86_avx512_mask_broadcasti32x2_256:
; X86:       # %bb.0:
; X86-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x08]
; X86-NEXT:    vmovq (%eax), %xmm2 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x7e,0x10]
; X86-NEXT:    # xmm2 = mem[0],zero
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vinserti32x4 $1, %xmm2, %ymm2, %ymm1 {%k1} # encoding: [0x62,0xf3,0x6d,0x29,0x38,0xca,0x01]
; X86-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm2 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x38,0xd0,0x01]
; X86-NEXT:    vinserti32x4 $1, %xmm0, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xa9,0x38,0xc0,0x01]
; X86-NEXT:    vpaddd %ymm2, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0xfe,0xc2]
; X86-NEXT:    vpaddd %ymm0, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf5,0xfe,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_broadcasti32x2_256:
; X64:       # %bb.0:
; X64-NEXT:    # kill: def $xmm0 killed $xmm0 def $ymm0
; X64-NEXT:    vmovq (%rsi), %xmm2 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x7e,0x16]
; X64-NEXT:    # xmm2 = mem[0],zero
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vinserti32x4 $1, %xmm2, %ymm2, %ymm1 {%k1} # encoding: [0x62,0xf3,0x6d,0x29,0x38,0xca,0x01]
; X64-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm2 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x7d,0x38,0xd0,0x01]
; X64-NEXT:    vinserti32x4 $1, %xmm0, %ymm0, %ymm0 {%k1} {z} # encoding: [0x62,0xf3,0x7d,0xa9,0x38,0xc0,0x01]
; X64-NEXT:    vpaddd %ymm2, %ymm0, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xfd,0xfe,0xc2]
; X64-NEXT:    vpaddd %ymm0, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf5,0xfe,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %y_64  = load i64, i64 * %y_ptr
  %y_v2i64 = insertelement <2 x i64> undef, i64 %y_64, i32 0
  %y = bitcast <2 x i64> %y_v2i64 to <4 x i32>
  %res = call <8 x i32> @llvm.x86.avx512.mask.broadcasti32x2.256(<4 x i32>  %y, <8 x i32> %x2, i8 %x3)
  %res1 = call <8 x i32> @llvm.x86.avx512.mask.broadcasti32x2.256(<4 x i32> %x0, <8 x i32> zeroinitializer, i8 %x3)
  %res2 = call <8 x i32> @llvm.x86.avx512.mask.broadcasti32x2.256(<4 x i32> %x0, <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 <4 x i32> @llvm.x86.avx512.mask.broadcasti32x2.128(<4 x i32>, <4 x i32>, i8)

define <4 x i32>@test_int_x86_avx512_mask_broadcasti32x2_128(<4 x i32> %x0, <4 x i32> %x2, i8 %x3) {
; X86-LABEL: test_int_x86_avx512_mask_broadcasti32x2_128:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vmovdqa32 %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7d,0x09,0x6f,0xc8]
; X86-NEXT:    vmovdqa32 %xmm0, %xmm2 {%k1} {z} # encoding: [0x62,0xf1,0x7d,0x89,0x6f,0xd0]
; X86-NEXT:    vpaddd %xmm0, %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe9,0xfe,0xc0]
; X86-NEXT:    vpaddd %xmm0, %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf1,0xfe,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_broadcasti32x2_128:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vmovdqa32 %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0x7d,0x09,0x6f,0xc8]
; X64-NEXT:    vmovdqa32 %xmm0, %xmm2 {%k1} {z} # encoding: [0x62,0xf1,0x7d,0x89,0x6f,0xd0]
; X64-NEXT:    vpaddd %xmm0, %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe9,0xfe,0xc0]
; X64-NEXT:    vpaddd %xmm0, %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf1,0xfe,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x i32> @llvm.x86.avx512.mask.broadcasti32x2.128(<4 x i32>  %x0, <4 x i32> %x2, i8 %x3)
  %res1 = call <4 x i32> @llvm.x86.avx512.mask.broadcasti32x2.128(<4 x i32> %x0, <4 x i32> zeroinitializer, i8 %x3)
  %res2 = call <4 x i32> @llvm.x86.avx512.mask.broadcasti32x2.128(<4 x i32> %x0, <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 i8 @llvm.x86.avx512.cvtd2mask.128(<4 x i32>)

define i8@test_int_x86_avx512_cvtd2mask_128(<4 x i32> %x0) {
; CHECK-LABEL: test_int_x86_avx512_cvtd2mask_128:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vpmovd2m %xmm0, %k0 # encoding: [0x62,0xf2,0x7e,0x08,0x39,0xc0]
; CHECK-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; CHECK-NEXT:    # kill: def $al killed $al killed $eax
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
    %res = call i8 @llvm.x86.avx512.cvtd2mask.128(<4 x i32> %x0)
    ret i8 %res
}

declare i8 @llvm.x86.avx512.cvtd2mask.256(<8 x i32>)

define i8@test_int_x86_avx512_cvtd2mask_256(<8 x i32> %x0) {
; CHECK-LABEL: test_int_x86_avx512_cvtd2mask_256:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vpmovd2m %ymm0, %k0 # encoding: [0x62,0xf2,0x7e,0x28,0x39,0xc0]
; CHECK-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; CHECK-NEXT:    # kill: def $al killed $al killed $eax
; CHECK-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
    %res = call i8 @llvm.x86.avx512.cvtd2mask.256(<8 x i32> %x0)
    ret i8 %res
}

declare i8 @llvm.x86.avx512.cvtq2mask.128(<2 x i64>)

define i8@test_int_x86_avx512_cvtq2mask_128(<2 x i64> %x0) {
; CHECK-LABEL: test_int_x86_avx512_cvtq2mask_128:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vpmovq2m %xmm0, %k0 # encoding: [0x62,0xf2,0xfe,0x08,0x39,0xc0]
; CHECK-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; CHECK-NEXT:    # kill: def $al killed $al killed $eax
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
    %res = call i8 @llvm.x86.avx512.cvtq2mask.128(<2 x i64> %x0)
    ret i8 %res
}

declare i8 @llvm.x86.avx512.cvtq2mask.256(<4 x i64>)

define i8@test_int_x86_avx512_cvtq2mask_256(<4 x i64> %x0) {
; CHECK-LABEL: test_int_x86_avx512_cvtq2mask_256:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vpmovq2m %ymm0, %k0 # encoding: [0x62,0xf2,0xfe,0x28,0x39,0xc0]
; CHECK-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; CHECK-NEXT:    # kill: def $al killed $al killed $eax
; CHECK-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
    %res = call i8 @llvm.x86.avx512.cvtq2mask.256(<4 x i64> %x0)
    ret i8 %res
}

declare <2 x double> @llvm.x86.avx512.mask.cvtqq2pd.128(<2 x i64>, <2 x double>, i8)

define <2 x double>@test_int_x86_avx512_mask_cvt_qq2pd_128(<2 x i64> %x0, <2 x double> %x1, i8 %x2) {
; X86-LABEL: test_int_x86_avx512_mask_cvt_qq2pd_128:
; X86:       # %bb.0:
; X86-NEXT:    vcvtqq2pd %xmm0, %xmm2 # encoding: [0x62,0xf1,0xfe,0x08,0xe6,0xd0]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vcvtqq2pd %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0xfe,0x09,0xe6,0xc8]
; X86-NEXT:    vaddpd %xmm2, %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf1,0x58,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_cvt_qq2pd_128:
; X64:       # %bb.0:
; X64-NEXT:    vcvtqq2pd %xmm0, %xmm2 # encoding: [0x62,0xf1,0xfe,0x08,0xe6,0xd0]
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vcvtqq2pd %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0xfe,0x09,0xe6,0xc8]
; X64-NEXT:    vaddpd %xmm2, %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf1,0x58,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <2 x double> @llvm.x86.avx512.mask.cvtqq2pd.128(<2 x i64> %x0, <2 x double> %x1, i8 %x2)
  %res1 = call <2 x double> @llvm.x86.avx512.mask.cvtqq2pd.128(<2 x i64> %x0, <2 x double> %x1, i8 -1)
  %res2 = fadd <2 x double> %res, %res1
  ret <2 x double> %res2
}

declare <4 x double> @llvm.x86.avx512.mask.cvtqq2pd.256(<4 x i64>, <4 x double>, i8)

define <4 x double>@test_int_x86_avx512_mask_cvt_qq2pd_256(<4 x i64> %x0, <4 x double> %x1, i8 %x2) {
; X86-LABEL: test_int_x86_avx512_mask_cvt_qq2pd_256:
; X86:       # %bb.0:
; X86-NEXT:    vcvtqq2pd %ymm0, %ymm2 # encoding: [0x62,0xf1,0xfe,0x28,0xe6,0xd0]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vcvtqq2pd %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0xfe,0x29,0xe6,0xc8]
; X86-NEXT:    vaddpd %ymm2, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf5,0x58,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_cvt_qq2pd_256:
; X64:       # %bb.0:
; X64-NEXT:    vcvtqq2pd %ymm0, %ymm2 # encoding: [0x62,0xf1,0xfe,0x28,0xe6,0xd0]
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vcvtqq2pd %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0xfe,0x29,0xe6,0xc8]
; X64-NEXT:    vaddpd %ymm2, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf5,0x58,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x double> @llvm.x86.avx512.mask.cvtqq2pd.256(<4 x i64> %x0, <4 x double> %x1, i8 %x2)
  %res1 = call <4 x double> @llvm.x86.avx512.mask.cvtqq2pd.256(<4 x i64> %x0, <4 x double> %x1, i8 -1)
  %res2 = fadd <4 x double> %res, %res1
  ret <4 x double> %res2
}

declare <2 x double> @llvm.x86.avx512.mask.cvtuqq2pd.128(<2 x i64>, <2 x double>, i8)

define <2 x double>@test_int_x86_avx512_mask_cvt_uqq2pd_128(<2 x i64> %x0, <2 x double> %x1, i8 %x2) {
; X86-LABEL: test_int_x86_avx512_mask_cvt_uqq2pd_128:
; X86:       # %bb.0:
; X86-NEXT:    vcvtuqq2pd %xmm0, %xmm2 # encoding: [0x62,0xf1,0xfe,0x08,0x7a,0xd0]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vcvtuqq2pd %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0xfe,0x09,0x7a,0xc8]
; X86-NEXT:    vaddpd %xmm2, %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf1,0x58,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_cvt_uqq2pd_128:
; X64:       # %bb.0:
; X64-NEXT:    vcvtuqq2pd %xmm0, %xmm2 # encoding: [0x62,0xf1,0xfe,0x08,0x7a,0xd0]
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vcvtuqq2pd %xmm0, %xmm1 {%k1} # encoding: [0x62,0xf1,0xfe,0x09,0x7a,0xc8]
; X64-NEXT:    vaddpd %xmm2, %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf1,0x58,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <2 x double> @llvm.x86.avx512.mask.cvtuqq2pd.128(<2 x i64> %x0, <2 x double> %x1, i8 %x2)
  %res1 = call <2 x double> @llvm.x86.avx512.mask.cvtuqq2pd.128(<2 x i64> %x0, <2 x double> %x1, i8 -1)
  %res2 = fadd <2 x double> %res, %res1
  ret <2 x double> %res2
}

declare <4 x double> @llvm.x86.avx512.mask.cvtuqq2pd.256(<4 x i64>, <4 x double>, i8)

define <4 x double>@test_int_x86_avx512_mask_cvt_uqq2pd_256(<4 x i64> %x0, <4 x double> %x1, i8 %x2) {
; X86-LABEL: test_int_x86_avx512_mask_cvt_uqq2pd_256:
; X86:       # %bb.0:
; X86-NEXT:    vcvtuqq2pd %ymm0, %ymm2 # encoding: [0x62,0xf1,0xfe,0x28,0x7a,0xd0]
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vcvtuqq2pd %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0xfe,0x29,0x7a,0xc8]
; X86-NEXT:    vaddpd %ymm2, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf5,0x58,0xc2]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_cvt_uqq2pd_256:
; X64:       # %bb.0:
; X64-NEXT:    vcvtuqq2pd %ymm0, %ymm2 # encoding: [0x62,0xf1,0xfe,0x28,0x7a,0xd0]
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vcvtuqq2pd %ymm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0xfe,0x29,0x7a,0xc8]
; X64-NEXT:    vaddpd %ymm2, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf5,0x58,0xc2]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x double> @llvm.x86.avx512.mask.cvtuqq2pd.256(<4 x i64> %x0, <4 x double> %x1, i8 %x2)
  %res1 = call <4 x double> @llvm.x86.avx512.mask.cvtuqq2pd.256(<4 x i64> %x0, <4 x double> %x1, i8 -1)
  %res2 = fadd <4 x double> %res, %res1
  ret <4 x double> %res2
}

declare i8 @llvm.x86.avx512.mask.fpclass.ps.128(<4 x float>, i32, i8)

define i8 @test_int_x86_avx512_mask_fpclass_ps_128(<4 x float> %x0) {
; CHECK-LABEL: test_int_x86_avx512_mask_fpclass_ps_128:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vfpclassps $4, %xmm0, %k1 # encoding: [0x62,0xf3,0x7d,0x08,0x66,0xc8,0x04]
; CHECK-NEXT:    vfpclassps $2, %xmm0, %k0 {%k1} # encoding: [0x62,0xf3,0x7d,0x09,0x66,0xc0,0x02]
; CHECK-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; CHECK-NEXT:    # kill: def $al killed $al killed $eax
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i8 @llvm.x86.avx512.mask.fpclass.ps.128(<4 x float> %x0, i32 2, i8 -1)
  %res1 = call i8 @llvm.x86.avx512.mask.fpclass.ps.128(<4 x float> %x0, i32 4, i8 %res)
  ret i8 %res1
}

declare i8 @llvm.x86.avx512.mask.fpclass.ps.256(<8 x float>, i32, i8)

define i8 @test_int_x86_avx512_mask_fpclass_ps_256(<8 x float> %x0) {
; CHECK-LABEL: test_int_x86_avx512_mask_fpclass_ps_256:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vfpclassps $4, %ymm0, %k1 # encoding: [0x62,0xf3,0x7d,0x28,0x66,0xc8,0x04]
; CHECK-NEXT:    vfpclassps $2, %ymm0, %k0 {%k1} # encoding: [0x62,0xf3,0x7d,0x29,0x66,0xc0,0x02]
; CHECK-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; CHECK-NEXT:    # kill: def $al killed $al killed $eax
; CHECK-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i8 @llvm.x86.avx512.mask.fpclass.ps.256(<8 x float> %x0, i32 2, i8 -1)
  %res1 = call i8 @llvm.x86.avx512.mask.fpclass.ps.256(<8 x float> %x0, i32 4, i8 %res)
  ret i8 %res1
}

declare i8 @llvm.x86.avx512.mask.fpclass.pd.128(<2 x double>, i32, i8)

define i8 @test_int_x86_avx512_mask_fpclass_pd_128(<2 x double> %x0) {
; CHECK-LABEL: test_int_x86_avx512_mask_fpclass_pd_128:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vfpclasspd $2, %xmm0, %k1 # encoding: [0x62,0xf3,0xfd,0x08,0x66,0xc8,0x02]
; CHECK-NEXT:    vfpclasspd $4, %xmm0, %k0 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0x66,0xc0,0x04]
; CHECK-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; CHECK-NEXT:    # kill: def $al killed $al killed $eax
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res =  call i8 @llvm.x86.avx512.mask.fpclass.pd.128(<2 x double> %x0, i32 4, i8 -1)
  %res1 = call i8 @llvm.x86.avx512.mask.fpclass.pd.128(<2 x double> %x0, i32 2, i8 %res)
  ret i8 %res1
}

declare i8 @llvm.x86.avx512.mask.fpclass.pd.256(<4 x double>, i32, i8)

define i8 @test_int_x86_avx512_mask_fpclass_pd_256(<4 x double> %x0) {
; CHECK-LABEL: test_int_x86_avx512_mask_fpclass_pd_256:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vfpclasspd $4, %ymm0, %k1 # encoding: [0x62,0xf3,0xfd,0x28,0x66,0xc8,0x04]
; CHECK-NEXT:    vfpclasspd $2, %ymm0, %k0 {%k1} # encoding: [0x62,0xf3,0xfd,0x29,0x66,0xc0,0x02]
; CHECK-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; CHECK-NEXT:    # kill: def $al killed $al killed $eax
; CHECK-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i8 @llvm.x86.avx512.mask.fpclass.pd.256(<4 x double> %x0, i32 2, i8 -1)
  %res1 = call i8 @llvm.x86.avx512.mask.fpclass.pd.256(<4 x double> %x0, i32 4, i8 %res)
  ret i8 %res1
}