; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=x86-64 -mattr=+avx512f,+avx512dq,+avx512bw,+avx512vl | FileCheck %s --check-prefix=GENERIC
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -print-schedule -mcpu=skx | FileCheck %s --check-prefix=SKX

; This test is an assembly of avx512 instructions to check their scheduling

define <8 x double> @addpd512(<8 x double> %y, <8 x double> %x) {
; GENERIC-LABEL: addpd512:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vaddpd %zmm0, %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: addpd512:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vaddpd %zmm0, %zmm1, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %add.i = fadd <8 x double> %x, %y
  ret <8 x double> %add.i
}

define <8 x double> @addpd512fold(<8 x double> %y) {
; GENERIC-LABEL: addpd512fold:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vaddpd {{.*}}(%rip), %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: addpd512fold:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vaddpd {{.*}}(%rip), %zmm0, %zmm0 # sched: [11:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %add.i = fadd <8 x double> %y, <double 4.500000e+00, double 3.400000e+00, double 2.300000e+00, double 1.200000e+00, double 4.500000e+00, double 3.800000e+00, double 2.300000e+00, double 1.200000e+00>
  ret <8 x double> %add.i
}

define <16 x float> @addps512(<16 x float> %y, <16 x float> %x) {
; GENERIC-LABEL: addps512:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vaddps %zmm0, %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: addps512:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vaddps %zmm0, %zmm1, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %add.i = fadd <16 x float> %x, %y
  ret <16 x float> %add.i
}

define <16 x float> @addps512fold(<16 x float> %y) {
; GENERIC-LABEL: addps512fold:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vaddps {{.*}}(%rip), %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: addps512fold:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vaddps {{.*}}(%rip), %zmm0, %zmm0 # sched: [11:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %add.i = fadd <16 x float> %y, <float 4.500000e+00, float 0x400B333340000000, float 0x4002666660000000, float 0x3FF3333340000000, float 4.500000e+00, float 0x400B333340000000, float 0x4002666660000000, float 0x3FF3333340000000, float 4.500000e+00, float 0x400B333340000000, float 0x4002666660000000, float 4.500000e+00, float 4.500000e+00, float 0x400B333340000000,  float 0x4002666660000000, float 0x3FF3333340000000>
  ret <16 x float> %add.i
}

define <8 x double> @subpd512(<8 x double> %y, <8 x double> %x) {
; GENERIC-LABEL: subpd512:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vsubpd %zmm0, %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: subpd512:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vsubpd %zmm0, %zmm1, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %sub.i = fsub <8 x double> %x, %y
  ret <8 x double> %sub.i
}

define <8 x double> @subpd512fold(<8 x double> %y, <8 x double>* %x) {
; GENERIC-LABEL: subpd512fold:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vsubpd (%rdi), %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: subpd512fold:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vsubpd (%rdi), %zmm0, %zmm0 # sched: [11:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %tmp2 = load <8 x double>, <8 x double>* %x, align 8
  %sub.i = fsub <8 x double> %y, %tmp2
  ret <8 x double> %sub.i
}

define <16 x float> @subps512(<16 x float> %y, <16 x float> %x) {
; GENERIC-LABEL: subps512:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vsubps %zmm0, %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: subps512:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vsubps %zmm0, %zmm1, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %sub.i = fsub <16 x float> %x, %y
  ret <16 x float> %sub.i
}

define <16 x float> @subps512fold(<16 x float> %y, <16 x float>* %x) {
; GENERIC-LABEL: subps512fold:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vsubps (%rdi), %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: subps512fold:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vsubps (%rdi), %zmm0, %zmm0 # sched: [11:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %tmp2 = load <16 x float>, <16 x float>* %x, align 4
  %sub.i = fsub <16 x float> %y, %tmp2
  ret <16 x float> %sub.i
}

define <8 x i64> @imulq512(<8 x i64> %y, <8 x i64> %x) {
; GENERIC-LABEL: imulq512:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmullq %zmm0, %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: imulq512:
; SKX:       # BB#0:
; SKX-NEXT:    vpmullq %zmm0, %zmm1, %zmm0 # sched: [12:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %z = mul <8 x i64>%x, %y
  ret <8 x i64>%z
}

define <4 x i64> @imulq256(<4 x i64> %y, <4 x i64> %x) {
; GENERIC-LABEL: imulq256:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmullq %ymm0, %ymm1, %ymm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: imulq256:
; SKX:       # BB#0:
; SKX-NEXT:    vpmullq %ymm0, %ymm1, %ymm0 # sched: [12:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %z = mul <4 x i64>%x, %y
  ret <4 x i64>%z
}

define <2 x i64> @imulq128(<2 x i64> %y, <2 x i64> %x) {
; GENERIC-LABEL: imulq128:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmullq %xmm0, %xmm1, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: imulq128:
; SKX:       # BB#0:
; SKX-NEXT:    vpmullq %xmm0, %xmm1, %xmm0 # sched: [12:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %z = mul <2 x i64>%x, %y
  ret <2 x i64>%z
}

define <8 x double> @mulpd512(<8 x double> %y, <8 x double> %x) {
; GENERIC-LABEL: mulpd512:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vmulpd %zmm0, %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mulpd512:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vmulpd %zmm0, %zmm1, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %mul.i = fmul <8 x double> %x, %y
  ret <8 x double> %mul.i
}

define <8 x double> @mulpd512fold(<8 x double> %y) {
; GENERIC-LABEL: mulpd512fold:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vmulpd {{.*}}(%rip), %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mulpd512fold:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vmulpd {{.*}}(%rip), %zmm0, %zmm0 # sched: [11:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %mul.i = fmul <8 x double> %y, <double 4.500000e+00, double 3.400000e+00, double 2.300000e+00, double 1.200000e+00, double 4.500000e+00, double 3.400000e+00, double 2.300000e+00, double 1.200000e+00>
  ret <8 x double> %mul.i
}

define <16 x float> @mulps512(<16 x float> %y, <16 x float> %x) {
; GENERIC-LABEL: mulps512:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vmulps %zmm0, %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mulps512:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vmulps %zmm0, %zmm1, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %mul.i = fmul <16 x float> %x, %y
  ret <16 x float> %mul.i
}

define <16 x float> @mulps512fold(<16 x float> %y) {
; GENERIC-LABEL: mulps512fold:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vmulps {{.*}}(%rip), %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mulps512fold:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vmulps {{.*}}(%rip), %zmm0, %zmm0 # sched: [11:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %mul.i = fmul <16 x float> %y, <float 4.500000e+00, float 0x400B333340000000, float 0x4002666660000000, float 0x3FF3333340000000, float 4.500000e+00, float 0x400B333340000000, float 0x4002666660000000, float 0x3FF3333340000000, float 4.500000e+00, float 0x400B333340000000, float 0x4002666660000000, float 0x3FF3333340000000, float 4.500000e+00, float 0x400B333340000000, float 0x4002666660000000, float 0x3FF3333340000000>
  ret <16 x float> %mul.i
}

define <8 x double> @divpd512(<8 x double> %y, <8 x double> %x) {
; GENERIC-LABEL: divpd512:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vdivpd %zmm0, %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: divpd512:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vdivpd %zmm0, %zmm1, %zmm0 # sched: [23:2.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %div.i = fdiv <8 x double> %x, %y
  ret <8 x double> %div.i
}

define <8 x double> @divpd512fold(<8 x double> %y) {
; GENERIC-LABEL: divpd512fold:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vdivpd {{.*}}(%rip), %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: divpd512fold:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vdivpd {{.*}}(%rip), %zmm0, %zmm0 # sched: [30:2.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %div.i = fdiv <8 x double> %y, <double 4.500000e+00, double 3.400000e+00, double 2.300000e+00, double 1.200000e+00, double 4.500000e+00, double 3.400000e+00, double 2.300000e+00, double 1.200000e+00>
  ret <8 x double> %div.i
}

define <16 x float> @divps512(<16 x float> %y, <16 x float> %x) {
; GENERIC-LABEL: divps512:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vdivps %zmm0, %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: divps512:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vdivps %zmm0, %zmm1, %zmm0 # sched: [23:2.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %div.i = fdiv <16 x float> %x, %y
  ret <16 x float> %div.i
}

define <16 x float> @divps512fold(<16 x float> %y) {
; GENERIC-LABEL: divps512fold:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vdivps {{.*}}(%rip), %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: divps512fold:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vdivps {{.*}}(%rip), %zmm0, %zmm0 # sched: [24:2.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %div.i = fdiv <16 x float> %y, <float 4.500000e+00, float 0x400B333340000000, float 0x4002666660000000, float 0x3FF3333340000000, float 4.500000e+00, float 4.500000e+00, float 0x4002666660000000, float 0x3FF3333340000000, float 4.500000e+00, float 0x400B333340000000, float 0x4002666660000000, float 0x3FF3333340000000, float 4.500000e+00, float 4.500000e+00, float 0x4002666660000000, float 0x3FF3333340000000>
  ret <16 x float> %div.i
}

define <8 x i64> @vpaddq_test(<8 x i64> %i, <8 x i64> %j) nounwind readnone {
; GENERIC-LABEL: vpaddq_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpaddq %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpaddq_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpaddq %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = add <8 x i64> %i, %j
  ret <8 x i64> %x
}

define <8 x i64> @vpaddq_fold_test(<8 x i64> %i, <8 x i64>* %j) nounwind {
; GENERIC-LABEL: vpaddq_fold_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpaddq (%rdi), %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpaddq_fold_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpaddq (%rdi), %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %tmp = load <8 x i64>, <8 x i64>* %j, align 4
  %x = add <8 x i64> %i, %tmp
  ret <8 x i64> %x
}

define <8 x i64> @vpaddq_broadcast_test(<8 x i64> %i) nounwind {
; GENERIC-LABEL: vpaddq_broadcast_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpaddq_broadcast_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = add <8 x i64> %i, <i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2>
  ret <8 x i64> %x
}

define <8 x i64> @vpaddq_broadcast2_test(<8 x i64> %i, i64* %j) nounwind {
; GENERIC-LABEL: vpaddq_broadcast2_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpaddq (%rdi){1to8}, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpaddq_broadcast2_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpaddq (%rdi){1to8}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %tmp = load i64, i64* %j
  %j.0 = insertelement <8 x i64> undef, i64 %tmp, i32 0
  %j.1 = insertelement <8 x i64> %j.0, i64 %tmp, i32 1
  %j.2 = insertelement <8 x i64> %j.1, i64 %tmp, i32 2
  %j.3 = insertelement <8 x i64> %j.2, i64 %tmp, i32 3
  %j.4 = insertelement <8 x i64> %j.3, i64 %tmp, i32 4
  %j.5 = insertelement <8 x i64> %j.4, i64 %tmp, i32 5
  %j.6 = insertelement <8 x i64> %j.5, i64 %tmp, i32 6
  %j.7 = insertelement <8 x i64> %j.6, i64 %tmp, i32 7
  %x = add <8 x i64> %i, %j.7
  ret <8 x i64> %x
}

define <16 x i32> @vpaddd_test(<16 x i32> %i, <16 x i32> %j) nounwind readnone {
; GENERIC-LABEL: vpaddd_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpaddd %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpaddd_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpaddd %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = add <16 x i32> %i, %j
  ret <16 x i32> %x
}

define <16 x i32> @vpaddd_fold_test(<16 x i32> %i, <16 x i32>* %j) nounwind {
; GENERIC-LABEL: vpaddd_fold_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpaddd (%rdi), %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpaddd_fold_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpaddd (%rdi), %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %tmp = load <16 x i32>, <16 x i32>* %j, align 4
  %x = add <16 x i32> %i, %tmp
  ret <16 x i32> %x
}

define <16 x i32> @vpaddd_broadcast_test(<16 x i32> %i) nounwind {
; GENERIC-LABEL: vpaddd_broadcast_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpaddd {{.*}}(%rip){1to16}, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpaddd_broadcast_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpaddd {{.*}}(%rip){1to16}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = add <16 x i32> %i, <i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3>
  ret <16 x i32> %x
}

define <16 x i32> @vpaddd_mask_test(<16 x i32> %i, <16 x i32> %j, <16 x i32> %mask1) nounwind readnone {
; GENERIC-LABEL: vpaddd_mask_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm3, %xmm3, %xmm3 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm3, %zmm2, %k1
; GENERIC-NEXT:    vpaddd %zmm1, %zmm0, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpaddd_mask_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm3, %xmm3, %xmm3 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm3, %zmm2, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpaddd %zmm1, %zmm0, %zmm0 {%k1} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %x = add <16 x i32> %i, %j
  %r = select <16 x i1> %mask, <16 x i32> %x, <16 x i32> %i
  ret <16 x i32> %r
}

define <16 x i32> @vpaddd_maskz_test(<16 x i32> %i, <16 x i32> %j, <16 x i32> %mask1) nounwind readnone {
; GENERIC-LABEL: vpaddd_maskz_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm3, %xmm3, %xmm3 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm3, %zmm2, %k1
; GENERIC-NEXT:    vpaddd %zmm1, %zmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpaddd_maskz_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm3, %xmm3, %xmm3 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm3, %zmm2, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpaddd %zmm1, %zmm0, %zmm0 {%k1} {z} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %x = add <16 x i32> %i, %j
  %r = select <16 x i1> %mask, <16 x i32> %x, <16 x i32> zeroinitializer
  ret <16 x i32> %r
}

define <16 x i32> @vpaddd_mask_fold_test(<16 x i32> %i, <16 x i32>* %j.ptr, <16 x i32> %mask1) nounwind readnone {
; GENERIC-LABEL: vpaddd_mask_fold_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vpaddd (%rdi), %zmm0, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpaddd_mask_fold_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpaddd (%rdi), %zmm0, %zmm0 {%k1} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %j = load <16 x i32>, <16 x i32>* %j.ptr
  %x = add <16 x i32> %i, %j
  %r = select <16 x i1> %mask, <16 x i32> %x, <16 x i32> %i
  ret <16 x i32> %r
}

define <16 x i32> @vpaddd_mask_broadcast_test(<16 x i32> %i, <16 x i32> %mask1) nounwind readnone {
; GENERIC-LABEL: vpaddd_mask_broadcast_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vpaddd {{.*}}(%rip){1to16}, %zmm0, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpaddd_mask_broadcast_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpaddd {{.*}}(%rip){1to16}, %zmm0, %zmm0 {%k1} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %x = add <16 x i32> %i, <i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4>
  %r = select <16 x i1> %mask, <16 x i32> %x, <16 x i32> %i
  ret <16 x i32> %r
}

define <16 x i32> @vpaddd_maskz_fold_test(<16 x i32> %i, <16 x i32>* %j.ptr, <16 x i32> %mask1) nounwind readnone {
; GENERIC-LABEL: vpaddd_maskz_fold_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vpaddd (%rdi), %zmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpaddd_maskz_fold_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpaddd (%rdi), %zmm0, %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %j = load <16 x i32>, <16 x i32>* %j.ptr
  %x = add <16 x i32> %i, %j
  %r = select <16 x i1> %mask, <16 x i32> %x, <16 x i32> zeroinitializer
  ret <16 x i32> %r
}

define <16 x i32> @vpaddd_maskz_broadcast_test(<16 x i32> %i, <16 x i32> %mask1) nounwind readnone {
; GENERIC-LABEL: vpaddd_maskz_broadcast_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vpaddd {{.*}}(%rip){1to16}, %zmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpaddd_maskz_broadcast_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpaddd {{.*}}(%rip){1to16}, %zmm0, %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %x = add <16 x i32> %i, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
  %r = select <16 x i1> %mask, <16 x i32> %x, <16 x i32> zeroinitializer
  ret <16 x i32> %r
}

define <8 x i64> @vpsubq_test(<8 x i64> %i, <8 x i64> %j) nounwind readnone {
; GENERIC-LABEL: vpsubq_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsubq %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpsubq_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpsubq %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = sub <8 x i64> %i, %j
  ret <8 x i64> %x
}

define <16 x i32> @vpsubd_test(<16 x i32> %i, <16 x i32> %j) nounwind readnone {
; GENERIC-LABEL: vpsubd_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsubd %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpsubd_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpsubd %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = sub <16 x i32> %i, %j
  ret <16 x i32> %x
}

define <16 x i32> @vpmulld_test(<16 x i32> %i, <16 x i32> %j) {
; GENERIC-LABEL: vpmulld_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmulld %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpmulld_test:
; SKX:       # BB#0:
; SKX-NEXT:    vpmulld %zmm1, %zmm0, %zmm0 # sched: [8:0.67]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = mul <16 x i32> %i, %j
  ret <16 x i32> %x
}

declare float @sqrtf(float) readnone
define float @sqrtA(float %a) nounwind uwtable readnone ssp {
; GENERIC-LABEL: sqrtA:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vsqrtss %xmm0, %xmm0, %xmm0 # sched: [114:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sqrtA:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vsqrtss %xmm0, %xmm0, %xmm0 # sched: [12:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %conv1 = tail call float @sqrtf(float %a) nounwind readnone
  ret float %conv1
}

declare double @sqrt(double) readnone
define double @sqrtB(double %a) nounwind uwtable readnone ssp {
; GENERIC-LABEL: sqrtB:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vsqrtsd %xmm0, %xmm0, %xmm0 # sched: [21:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sqrtB:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vsqrtsd %xmm0, %xmm0, %xmm0 # sched: [18:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %call = tail call double @sqrt(double %a) nounwind readnone
  ret double %call
}

declare float @llvm.sqrt.f32(float)
define float @sqrtC(float %a) nounwind {
; GENERIC-LABEL: sqrtC:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vsqrtss %xmm0, %xmm0, %xmm0 # sched: [114:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sqrtC:
; SKX:       # BB#0:
; SKX-NEXT:    vsqrtss %xmm0, %xmm0, %xmm0 # sched: [12:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = call float @llvm.sqrt.f32(float %a)
  ret float %b
}

declare <16 x float> @llvm.sqrt.v16f32(<16 x float>)
define <16 x float> @sqrtD(<16 x float> %a) nounwind {
; GENERIC-LABEL: sqrtD:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vsqrtps %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sqrtD:
; SKX:       # BB#0:
; SKX-NEXT:    vsqrtps %zmm0, %zmm0 # sched: [19:2.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = call <16 x float> @llvm.sqrt.v16f32(<16 x float> %a)
  ret <16 x float> %b
}

declare <8 x double> @llvm.sqrt.v8f64(<8 x double>)
define <8 x double> @sqrtE(<8 x double> %a) nounwind {
; GENERIC-LABEL: sqrtE:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vsqrtpd %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sqrtE:
; SKX:       # BB#0:
; SKX-NEXT:    vsqrtpd %zmm0, %zmm0 # sched: [31:2.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = call <8 x double> @llvm.sqrt.v8f64(<8 x double> %a)
  ret <8 x double> %b
}

define <16 x float> @fadd_broadcast(<16 x float> %a) nounwind {
; GENERIC-LABEL: fadd_broadcast:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vaddps {{.*}}(%rip){1to16}, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: fadd_broadcast:
; SKX:       # BB#0:
; SKX-NEXT:    vaddps {{.*}}(%rip){1to16}, %zmm0, %zmm0 # sched: [11:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = fadd <16 x float> %a, <float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000>
  ret <16 x float> %b
}

define <8 x i64> @addq_broadcast(<8 x i64> %a) nounwind {
; GENERIC-LABEL: addq_broadcast:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: addq_broadcast:
; SKX:       # BB#0:
; SKX-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = add <8 x i64> %a, <i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2>
  ret <8 x i64> %b
}

define <8 x i64> @orq_broadcast(<8 x i64> %a) nounwind {
; GENERIC-LABEL: orq_broadcast:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vorpd {{.*}}(%rip){1to8}, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: orq_broadcast:
; SKX:       # BB#0:
; SKX-NEXT:    vorpd {{.*}}(%rip){1to8}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = or <8 x i64> %a, <i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2, i64 2>
  ret <8 x i64> %b
}

define <16 x i32> @andd512fold(<16 x i32> %y, <16 x i32>* %x) {
; GENERIC-LABEL: andd512fold:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vandps (%rdi), %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: andd512fold:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vandps (%rdi), %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %a = load <16 x i32>, <16 x i32>* %x, align 4
  %b = and <16 x i32> %y, %a
  ret <16 x i32> %b
}

define <8 x i64> @andqbrst(<8 x i64> %p1, i64* %ap) {
; GENERIC-LABEL: andqbrst:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vandpd (%rdi){1to8}, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: andqbrst:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vandpd (%rdi){1to8}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %a = load i64, i64* %ap, align 8
  %b = insertelement <8 x i64> undef, i64 %a, i32 0
  %c = shufflevector <8 x i64> %b, <8 x i64> undef, <8 x i32> zeroinitializer
  %d = and <8 x i64> %p1, %c
  ret <8 x i64>%d
}

define <16 x float> @test_mask_vaddps(<16 x float> %dst, <16 x float> %i,
; GENERIC-LABEL: test_mask_vaddps:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm4, %zmm3, %k1
; GENERIC-NEXT:    vaddps %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mask_vaddps:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm4, %zmm3, %k1 # sched: [3:1.00]
; SKX-NEXT:    vaddps %zmm2, %zmm1, %zmm0 {%k1} # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
                                     <16 x float> %j, <16 x i32> %mask1)
                                     nounwind readnone {
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %x = fadd <16 x float> %i, %j
  %r = select <16 x i1> %mask, <16 x float> %x, <16 x float> %dst
  ret <16 x float> %r
}

define <16 x float> @test_mask_vmulps(<16 x float> %dst, <16 x float> %i, <16 x float> %j, <16 x i32> %mask1) nounwind readnone {
; GENERIC-LABEL: test_mask_vmulps:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm4, %zmm3, %k1
; GENERIC-NEXT:    vmulps %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mask_vmulps:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm4, %zmm3, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmulps %zmm2, %zmm1, %zmm0 {%k1} # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %x = fmul <16 x float> %i, %j
  %r = select <16 x i1> %mask, <16 x float> %x, <16 x float> %dst
  ret <16 x float> %r
}

define <16 x float> @test_mask_vminps(<16 x float> %dst, <16 x float> %i, <16 x float> %j, <16 x i32> %mask1) nounwind readnone {
; GENERIC-LABEL: test_mask_vminps:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm4, %zmm3, %k1
; GENERIC-NEXT:    vminps %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mask_vminps:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm4, %zmm3, %k1 # sched: [3:1.00]
; SKX-NEXT:    vminps %zmm2, %zmm1, %zmm0 {%k1} # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %cmp_res = fcmp olt <16 x float> %i, %j
  %min = select <16 x i1> %cmp_res, <16 x float> %i, <16 x float> %j
  %r = select <16 x i1> %mask, <16 x float> %min, <16 x float> %dst
  ret <16 x float> %r
}

define <8 x double> @test_mask_vminpd(<8 x double> %dst, <8 x double> %i, <8 x double> %j, <8 x i32> %mask1) nounwind readnone {
; GENERIC-LABEL: test_mask_vminpd:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %ymm4, %ymm3, %k1
; GENERIC-NEXT:    vminpd %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mask_vminpd:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %ymm4, %ymm3, %k1 # sched: [3:1.00]
; SKX-NEXT:    vminpd %zmm2, %zmm1, %zmm0 {%k1} # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %cmp_res = fcmp olt <8 x double> %i, %j
  %min = select <8 x i1> %cmp_res, <8 x double> %i, <8 x double> %j
  %r = select <8 x i1> %mask, <8 x double> %min, <8 x double> %dst
  ret <8 x double> %r
}

define <16 x float> @test_mask_vmaxps(<16 x float> %dst, <16 x float> %i, <16 x float> %j, <16 x i32> %mask1) nounwind readnone {
; GENERIC-LABEL: test_mask_vmaxps:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm4, %zmm3, %k1
; GENERIC-NEXT:    vmaxps %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mask_vmaxps:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm4, %zmm3, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmaxps %zmm2, %zmm1, %zmm0 {%k1} # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %cmp_res = fcmp ogt <16 x float> %i, %j
  %max = select <16 x i1> %cmp_res, <16 x float> %i, <16 x float> %j
  %r = select <16 x i1> %mask, <16 x float> %max, <16 x float> %dst
  ret <16 x float> %r
}

define <8 x double> @test_mask_vmaxpd(<8 x double> %dst, <8 x double> %i, <8 x double> %j, <8 x i32> %mask1) nounwind readnone {
; GENERIC-LABEL: test_mask_vmaxpd:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %ymm4, %ymm3, %k1
; GENERIC-NEXT:    vmaxpd %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mask_vmaxpd:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %ymm4, %ymm3, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmaxpd %zmm2, %zmm1, %zmm0 {%k1} # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %cmp_res = fcmp ogt <8 x double> %i, %j
  %max = select <8 x i1> %cmp_res, <8 x double> %i, <8 x double> %j
  %r = select <8 x i1> %mask, <8 x double> %max, <8 x double> %dst
  ret <8 x double> %r
}

define <16 x float> @test_mask_vsubps(<16 x float> %dst, <16 x float> %i, <16 x float> %j, <16 x i32> %mask1) nounwind readnone {
; GENERIC-LABEL: test_mask_vsubps:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm4, %zmm3, %k1
; GENERIC-NEXT:    vsubps %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mask_vsubps:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm4, %zmm3, %k1 # sched: [3:1.00]
; SKX-NEXT:    vsubps %zmm2, %zmm1, %zmm0 {%k1} # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %x = fsub <16 x float> %i, %j
  %r = select <16 x i1> %mask, <16 x float> %x, <16 x float> %dst
  ret <16 x float> %r
}

define <16 x float> @test_mask_vdivps(<16 x float> %dst, <16 x float> %i, <16 x float> %j, <16 x i32> %mask1) nounwind readnone {
; GENERIC-LABEL: test_mask_vdivps:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm4, %zmm3, %k1
; GENERIC-NEXT:    vdivps %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mask_vdivps:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm4, %zmm3, %k1 # sched: [3:1.00]
; SKX-NEXT:    vdivps %zmm2, %zmm1, %zmm0 {%k1} # sched: [23:2.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %x = fdiv <16 x float> %i, %j
  %r = select <16 x i1> %mask, <16 x float> %x, <16 x float> %dst
  ret <16 x float> %r
}

define <8 x double> @test_mask_vaddpd(<8 x double> %dst, <8 x double> %i, <8 x double> %j, <8 x i64> %mask1) nounwind readnone {
; GENERIC-LABEL: test_mask_vaddpd:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqq %zmm4, %zmm3, %k1
; GENERIC-NEXT:    vaddpd %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mask_vaddpd:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm4, %xmm4, %xmm4 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqq %zmm4, %zmm3, %k1 # sched: [3:1.00]
; SKX-NEXT:    vaddpd %zmm2, %zmm1, %zmm0 {%k1} # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <8 x i64> %mask1, zeroinitializer
  %x = fadd <8 x double> %i, %j
  %r = select <8 x i1> %mask, <8 x double> %x, <8 x double> %dst
  ret <8 x double> %r
}

define <8 x double> @test_maskz_vaddpd(<8 x double> %i, <8 x double> %j, <8 x i64> %mask1) nounwind readnone {
; GENERIC-LABEL: test_maskz_vaddpd:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm3, %xmm3, %xmm3 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqq %zmm3, %zmm2, %k1
; GENERIC-NEXT:    vaddpd %zmm1, %zmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_maskz_vaddpd:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm3, %xmm3, %xmm3 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqq %zmm3, %zmm2, %k1 # sched: [3:1.00]
; SKX-NEXT:    vaddpd %zmm1, %zmm0, %zmm0 {%k1} {z} # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <8 x i64> %mask1, zeroinitializer
  %x = fadd <8 x double> %i, %j
  %r = select <8 x i1> %mask, <8 x double> %x, <8 x double> zeroinitializer
  ret <8 x double> %r
}

define <8 x double> @test_mask_fold_vaddpd(<8 x double> %dst, <8 x double> %i, <8 x double>* %j,  <8 x i64> %mask1) nounwind {
; GENERIC-LABEL: test_mask_fold_vaddpd:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm3, %xmm3, %xmm3 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqq %zmm3, %zmm2, %k1
; GENERIC-NEXT:    vaddpd (%rdi), %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mask_fold_vaddpd:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm3, %xmm3, %xmm3 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqq %zmm3, %zmm2, %k1 # sched: [3:1.00]
; SKX-NEXT:    vaddpd (%rdi), %zmm1, %zmm0 {%k1} # sched: [11:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <8 x i64> %mask1, zeroinitializer
  %tmp = load <8 x double>, <8 x double>* %j, align 8
  %x = fadd <8 x double> %i, %tmp
  %r = select <8 x i1> %mask, <8 x double> %x, <8 x double> %dst
  ret <8 x double> %r
}

define <8 x double> @test_maskz_fold_vaddpd(<8 x double> %i, <8 x double>* %j, <8 x i64> %mask1) nounwind {
; GENERIC-LABEL: test_maskz_fold_vaddpd:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqq %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vaddpd (%rdi), %zmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_maskz_fold_vaddpd:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqq %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vaddpd (%rdi), %zmm0, %zmm0 {%k1} {z} # sched: [11:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <8 x i64> %mask1, zeroinitializer
  %tmp = load <8 x double>, <8 x double>* %j, align 8
  %x = fadd <8 x double> %i, %tmp
  %r = select <8 x i1> %mask, <8 x double> %x, <8 x double> zeroinitializer
  ret <8 x double> %r
}

define <8 x double> @test_broadcast_vaddpd(<8 x double> %i, double* %j) nounwind {
; GENERIC-LABEL: test_broadcast_vaddpd:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vaddpd (%rdi){1to8}, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_broadcast_vaddpd:
; SKX:       # BB#0:
; SKX-NEXT:    vaddpd (%rdi){1to8}, %zmm0, %zmm0 # sched: [11:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %tmp = load double, double* %j
  %b = insertelement <8 x double> undef, double %tmp, i32 0
  %c = shufflevector <8 x double> %b, <8 x double> undef,
                     <8 x i32> zeroinitializer
  %x = fadd <8 x double> %c, %i
  ret <8 x double> %x
}

define <8 x double> @test_mask_broadcast_vaddpd(<8 x double> %dst, <8 x double> %i, double* %j, <8 x i64> %mask1) nounwind {
; GENERIC-LABEL: test_mask_broadcast_vaddpd:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm0, %xmm0, %xmm0 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqq %zmm0, %zmm2, %k1
; GENERIC-NEXT:    vaddpd (%rdi){1to8}, %zmm1, %zmm1 {%k1}
; GENERIC-NEXT:    vmovapd %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mask_broadcast_vaddpd:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm0, %xmm0, %xmm0 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqq %zmm0, %zmm2, %k1 # sched: [3:1.00]
; SKX-NEXT:    vaddpd (%rdi){1to8}, %zmm1, %zmm1 {%k1} # sched: [11:0.50]
; SKX-NEXT:    vmovapd %zmm1, %zmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <8 x i64> %mask1, zeroinitializer
  %tmp = load double, double* %j
  %b = insertelement <8 x double> undef, double %tmp, i32 0
  %c = shufflevector <8 x double> %b, <8 x double> undef,
                     <8 x i32> zeroinitializer
  %x = fadd <8 x double> %c, %i
  %r = select <8 x i1> %mask, <8 x double> %x, <8 x double> %i
  ret <8 x double> %r
}

define <8 x double> @test_maskz_broadcast_vaddpd(<8 x double> %i, double* %j,
; GENERIC-LABEL: test_maskz_broadcast_vaddpd:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqq %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vaddpd (%rdi){1to8}, %zmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_maskz_broadcast_vaddpd:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqq %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vaddpd (%rdi){1to8}, %zmm0, %zmm0 {%k1} {z} # sched: [11:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
                                       <8 x i64> %mask1) nounwind {
  %mask = icmp ne <8 x i64> %mask1, zeroinitializer
  %tmp = load double, double* %j
  %b = insertelement <8 x double> undef, double %tmp, i32 0
  %c = shufflevector <8 x double> %b, <8 x double> undef,
                     <8 x i32> zeroinitializer
  %x = fadd <8 x double> %c, %i
  %r = select <8 x i1> %mask, <8 x double> %x, <8 x double> zeroinitializer
  ret <8 x double> %r
}

define <16 x float>  @test_fxor(<16 x float> %a) {
; GENERIC-LABEL: test_fxor:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorps {{.*}}(%rip){1to16}, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_fxor:
; SKX:       # BB#0:
; SKX-NEXT:    vxorps {{.*}}(%rip){1to16}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]

  %res = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %a
  ret <16 x float>%res
}

define <8 x float>  @test_fxor_8f32(<8 x float> %a) {
; GENERIC-LABEL: test_fxor_8f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorps {{.*}}(%rip){1to8}, %ymm0, %ymm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_fxor_8f32:
; SKX:       # BB#0:
; SKX-NEXT:    vxorps {{.*}}(%rip){1to8}, %ymm0, %ymm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %a
  ret <8 x float>%res
}

define <8 x double> @fabs_v8f64(<8 x double> %p)
; GENERIC-LABEL: fabs_v8f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vandpd {{.*}}(%rip){1to8}, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: fabs_v8f64:
; SKX:       # BB#0:
; SKX-NEXT:    vandpd {{.*}}(%rip){1to8}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
{
  %t = call <8 x double> @llvm.fabs.v8f64(<8 x double> %p)
  ret <8 x double> %t
}
declare <8 x double> @llvm.fabs.v8f64(<8 x double> %p)

define <16 x float> @fabs_v16f32(<16 x float> %p)
; GENERIC-LABEL: fabs_v16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vandps {{.*}}(%rip){1to16}, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: fabs_v16f32:
; SKX:       # BB#0:
; SKX-NEXT:    vandps {{.*}}(%rip){1to16}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
{
  %t = call <16 x float> @llvm.fabs.v16f32(<16 x float> %p)
  ret <16 x float> %t
}
declare <16 x float> @llvm.fabs.v16f32(<16 x float> %p)

define double @test1(double %a, double %b) nounwind {
; GENERIC-LABEL: test1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vucomisd %xmm1, %xmm0 # sched: [2:1.00]
; GENERIC-NEXT:    jne .LBB64_1 # sched: [1:1.00]
; GENERIC-NEXT:    jnp .LBB64_2 # sched: [1:1.00]
; GENERIC-NEXT:  .LBB64_1: # %l1
; GENERIC-NEXT:    vsubsd %xmm1, %xmm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
; GENERIC-NEXT:  .LBB64_2: # %l2
; GENERIC-NEXT:    vaddsd %xmm1, %xmm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test1:
; SKX:       # BB#0:
; SKX-NEXT:    vucomisd %xmm1, %xmm0 # sched: [2:1.00]
; SKX-NEXT:    jne .LBB64_1 # sched: [1:0.50]
; SKX-NEXT:    jnp .LBB64_2 # sched: [1:0.50]
; SKX-NEXT:  .LBB64_1: # %l1
; SKX-NEXT:    vsubsd %xmm1, %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
; SKX-NEXT:  .LBB64_2: # %l2
; SKX-NEXT:    vaddsd %xmm1, %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %tobool = fcmp une double %a, %b
  br i1 %tobool, label %l1, label %l2

l1:
  %c = fsub double %a, %b
  ret double %c
l2:
  %c1 = fadd double %a, %b
  ret double %c1
}

define float @test2(float %a, float %b) nounwind {
; GENERIC-LABEL: test2:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vucomiss %xmm0, %xmm1 # sched: [2:1.00]
; GENERIC-NEXT:    jbe .LBB65_2 # sched: [1:1.00]
; GENERIC-NEXT:  # BB#1: # %l1
; GENERIC-NEXT:    vsubss %xmm1, %xmm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
; GENERIC-NEXT:  .LBB65_2: # %l2
; GENERIC-NEXT:    vaddss %xmm1, %xmm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test2:
; SKX:       # BB#0:
; SKX-NEXT:    vucomiss %xmm0, %xmm1 # sched: [2:1.00]
; SKX-NEXT:    jbe .LBB65_2 # sched: [1:0.50]
; SKX-NEXT:  # BB#1: # %l1
; SKX-NEXT:    vsubss %xmm1, %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
; SKX-NEXT:  .LBB65_2: # %l2
; SKX-NEXT:    vaddss %xmm1, %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %tobool = fcmp olt float %a, %b
  br i1 %tobool, label %l1, label %l2

l1:
  %c = fsub float %a, %b
  ret float %c
l2:
  %c1 = fadd float %a, %b
  ret float %c1
}

define i32 @test3(float %a, float %b) {
; GENERIC-LABEL: test3:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcmpeqss %xmm1, %xmm0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    movzbl %al, %eax # sched: [1:0.33]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test3:
; SKX:       # BB#0:
; SKX-NEXT:    vcmpeqss %xmm1, %xmm0, %k0
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    movzbl %al, %eax # sched: [1:0.25]
; SKX-NEXT:    retq # sched: [7:1.00]

  %cmp10.i = fcmp oeq float %a, %b
  %conv11.i = zext i1 %cmp10.i to i32
  ret i32 %conv11.i
}

define float @test5(float %p) #0 {
; GENERIC-LABEL: test5:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vxorps %xmm1, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vucomiss %xmm1, %xmm0 # sched: [2:1.00]
; GENERIC-NEXT:    jne .LBB67_1 # sched: [1:1.00]
; GENERIC-NEXT:    jp .LBB67_1 # sched: [1:1.00]
; GENERIC-NEXT:  # BB#2: # %return
; GENERIC-NEXT:    retq # sched: [1:1.00]
; GENERIC-NEXT:  .LBB67_1: # %if.end
; GENERIC-NEXT:    seta %al # sched: [2:1.00]
; GENERIC-NEXT:    movzbl %al, %eax # sched: [1:0.33]
; GENERIC-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero sched: [6:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test5:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vxorps %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vucomiss %xmm1, %xmm0 # sched: [2:1.00]
; SKX-NEXT:    jne .LBB67_1 # sched: [1:0.50]
; SKX-NEXT:    jp .LBB67_1 # sched: [1:0.50]
; SKX-NEXT:  # BB#2: # %return
; SKX-NEXT:    retq # sched: [7:1.00]
; SKX-NEXT:  .LBB67_1: # %if.end
; SKX-NEXT:    seta %al # sched: [2:1.00]
; SKX-NEXT:    movzbl %al, %eax # sched: [1:0.25]
; SKX-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero sched: [5:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %cmp = fcmp oeq float %p, 0.000000e+00
  br i1 %cmp, label %return, label %if.end

if.end:                                           ; preds = %entry
  %cmp1 = fcmp ogt float %p, 0.000000e+00
  %cond = select i1 %cmp1, float 1.000000e+00, float -1.000000e+00
  br label %return

return:                                           ; preds = %if.end, %entry
  %retval.0 = phi float [ %cond, %if.end ], [ %p, %entry ]
  ret float %retval.0
}

define i32 @test6(i32 %a, i32 %b) {
; GENERIC-LABEL: test6:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    xorl %eax, %eax # sched: [1:0.33]
; GENERIC-NEXT:    cmpl %esi, %edi # sched: [1:0.33]
; GENERIC-NEXT:    sete %al # sched: [1:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test6:
; SKX:       # BB#0:
; SKX-NEXT:    xorl %eax, %eax # sched: [1:0.25]
; SKX-NEXT:    cmpl %esi, %edi # sched: [1:0.25]
; SKX-NEXT:    sete %al # sched: [1:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %cmp = icmp eq i32 %a, %b
  %res = zext i1 %cmp to i32
  ret i32 %res
}

define i32 @test7(double %x, double %y) #2 {
; GENERIC-LABEL: test7:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    xorl %eax, %eax # sched: [1:0.33]
; GENERIC-NEXT:    vucomisd %xmm1, %xmm0 # sched: [2:1.00]
; GENERIC-NEXT:    setne %al # sched: [1:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test7:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    xorl %eax, %eax # sched: [1:0.25]
; SKX-NEXT:    vucomisd %xmm1, %xmm0 # sched: [2:1.00]
; SKX-NEXT:    setne %al # sched: [1:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = fcmp one double %x, %y
  %or = zext i1 %0 to i32
  ret i32 %or
}

define i32 @test8(i32 %a1, i32 %a2, i32 %a3) {
; GENERIC-LABEL: test8:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    xorl $-2147483648, %esi # imm = 0x80000000
; GENERIC-NEXT:    # sched: [1:0.33]
; GENERIC-NEXT:    testl %edx, %edx # sched: [1:0.33]
; GENERIC-NEXT:    movl $1, %eax # sched: [1:0.33]
; GENERIC-NEXT:    cmovel %eax, %edx # sched: [2:0.67]
; GENERIC-NEXT:    notl %edi # sched: [1:0.33]
; GENERIC-NEXT:    orl %edi, %esi # sched: [1:0.33]
; GENERIC-NEXT:    cmovnel %edx, %eax # sched: [2:0.67]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test8:
; SKX:       # BB#0:
; SKX-NEXT:    notl %edi # sched: [1:0.25]
; SKX-NEXT:    xorl $-2147483648, %esi # imm = 0x80000000
; SKX-NEXT:    # sched: [1:0.25]
; SKX-NEXT:    testl %edx, %edx # sched: [1:0.25]
; SKX-NEXT:    movl $1, %eax # sched: [1:0.25]
; SKX-NEXT:    cmovel %eax, %edx # sched: [1:0.50]
; SKX-NEXT:    orl %edi, %esi # sched: [1:0.25]
; SKX-NEXT:    cmovnel %edx, %eax # sched: [1:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %tmp1 = icmp eq i32 %a1, -1
  %tmp2 = icmp eq i32 %a2, -2147483648
  %tmp3 = and i1 %tmp1, %tmp2
  %tmp4 = icmp eq i32 %a3, 0
  %tmp5 = or i1 %tmp3, %tmp4
  %res = select i1 %tmp5, i32 1, i32 %a3
  ret i32 %res
}

define i32 @test9(i64 %a) {
; GENERIC-LABEL: test9:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    testb $1, %dil # sched: [1:0.33]
; GENERIC-NEXT:    jne .LBB71_2 # sched: [1:1.00]
; GENERIC-NEXT:  # BB#1: # %A
; GENERIC-NEXT:    movl $6, %eax # sched: [1:0.33]
; GENERIC-NEXT:    retq # sched: [1:1.00]
; GENERIC-NEXT:  .LBB71_2: # %B
; GENERIC-NEXT:    movl $7, %eax # sched: [1:0.33]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test9:
; SKX:       # BB#0:
; SKX-NEXT:    testb $1, %dil # sched: [1:0.25]
; SKX-NEXT:    jne .LBB71_2 # sched: [1:0.50]
; SKX-NEXT:  # BB#1: # %A
; SKX-NEXT:    movl $6, %eax # sched: [1:0.25]
; SKX-NEXT:    retq # sched: [7:1.00]
; SKX-NEXT:  .LBB71_2: # %B
; SKX-NEXT:    movl $7, %eax # sched: [1:0.25]
; SKX-NEXT:    retq # sched: [7:1.00]
 %b = and i64 %a, 1
 %cmp10.i = icmp eq i64 %b, 0
 br i1 %cmp10.i, label %A, label %B
A:
 ret i32 6
B:
 ret i32 7
}

define i32 @test10(i64 %b, i64 %c, i1 %d) {
; GENERIC-LABEL: test10:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    movl %edx, %eax # sched: [1:0.33]
; GENERIC-NEXT:    andb $1, %al # sched: [1:0.33]
; GENERIC-NEXT:    cmpq %rsi, %rdi # sched: [1:0.33]
; GENERIC-NEXT:    sete %cl # sched: [1:0.50]
; GENERIC-NEXT:    orb %dl, %cl # sched: [1:0.33]
; GENERIC-NEXT:    andb $1, %cl # sched: [1:0.33]
; GENERIC-NEXT:    cmpb %cl, %al # sched: [1:0.33]
; GENERIC-NEXT:    je .LBB72_1 # sched: [1:1.00]
; GENERIC-NEXT:  # BB#2: # %if.end.i
; GENERIC-NEXT:    movl $6, %eax # sched: [1:0.33]
; GENERIC-NEXT:    retq # sched: [1:1.00]
; GENERIC-NEXT:  .LBB72_1: # %if.then.i
; GENERIC-NEXT:    movl $5, %eax # sched: [1:0.33]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test10:
; SKX:       # BB#0:
; SKX-NEXT:    movl %edx, %eax # sched: [1:0.25]
; SKX-NEXT:    andb $1, %al # sched: [1:0.25]
; SKX-NEXT:    cmpq %rsi, %rdi # sched: [1:0.25]
; SKX-NEXT:    sete %cl # sched: [1:0.50]
; SKX-NEXT:    orb %dl, %cl # sched: [1:0.25]
; SKX-NEXT:    andb $1, %cl # sched: [1:0.25]
; SKX-NEXT:    cmpb %cl, %al # sched: [1:0.25]
; SKX-NEXT:    je .LBB72_1 # sched: [1:0.50]
; SKX-NEXT:  # BB#2: # %if.end.i
; SKX-NEXT:    movl $6, %eax # sched: [1:0.25]
; SKX-NEXT:    retq # sched: [7:1.00]
; SKX-NEXT:  .LBB72_1: # %if.then.i
; SKX-NEXT:    movl $5, %eax # sched: [1:0.25]
; SKX-NEXT:    retq # sched: [7:1.00]

  %cmp8.i = icmp eq i64 %b, %c
  %or1 = or i1 %d, %cmp8.i
  %xor1 = xor i1 %d, %or1
  br i1 %xor1, label %if.end.i, label %if.then.i

if.then.i:
 ret i32 5

if.end.i:
  ret i32 6
}

define <16 x float> @sitof32(<16 x i32> %a) nounwind {
; GENERIC-LABEL: sitof32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtdq2ps %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sitof32:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtdq2ps %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <16 x i32> %a to <16 x float>
  ret <16 x float> %b
}

define <8 x double> @sltof864(<8 x i64> %a) {
; GENERIC-LABEL: sltof864:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtqq2pd %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sltof864:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtqq2pd %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <8 x i64> %a to <8 x double>
  ret <8 x double> %b
}

define <4 x double> @slto4f64(<4 x i64> %a) {
; GENERIC-LABEL: slto4f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtqq2pd %ymm0, %ymm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: slto4f64:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtqq2pd %ymm0, %ymm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <4 x i64> %a to <4 x double>
  ret <4 x double> %b
}

define <2 x double> @slto2f64(<2 x i64> %a) {
; GENERIC-LABEL: slto2f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtqq2pd %xmm0, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: slto2f64:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtqq2pd %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <2 x i64> %a to <2 x double>
  ret <2 x double> %b
}

define <2 x float> @sltof2f32(<2 x i64> %a) {
; GENERIC-LABEL: sltof2f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtqq2ps %xmm0, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sltof2f32:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtqq2ps %xmm0, %xmm0 # sched: [5:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <2 x i64> %a to <2 x float>
  ret <2 x float>%b
}

define <4 x float> @slto4f32_mem(<4 x i64>* %a) {
; GENERIC-LABEL: slto4f32_mem:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtqq2psy (%rdi), %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: slto4f32_mem:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtqq2psy (%rdi), %xmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %a1 = load <4 x i64>, <4 x i64>* %a, align 8
  %b = sitofp <4 x i64> %a1 to <4 x float>
  ret <4 x float>%b
}

define <4 x i64> @f64to4sl(<4 x double> %a) {
; GENERIC-LABEL: f64to4sl:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttpd2qq %ymm0, %ymm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f64to4sl:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttpd2qq %ymm0, %ymm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = fptosi <4 x double> %a to <4 x i64>
  ret <4 x i64> %b
}

define <4 x i64> @f32to4sl(<4 x float> %a) {
; GENERIC-LABEL: f32to4sl:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttps2qq %xmm0, %ymm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f32to4sl:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttps2qq %xmm0, %ymm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = fptosi <4 x float> %a to <4 x i64>
  ret <4 x i64> %b
}

define <4 x float> @slto4f32(<4 x i64> %a) {
; GENERIC-LABEL: slto4f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtqq2ps %ymm0, %xmm0
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: slto4f32:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtqq2ps %ymm0, %xmm0 # sched: [7:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <4 x i64> %a to <4 x float>
  ret <4 x float> %b
}

define <4 x float> @ulto4f32(<4 x i64> %a) {
; GENERIC-LABEL: ulto4f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtuqq2ps %ymm0, %xmm0
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ulto4f32:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtuqq2ps %ymm0, %xmm0 # sched: [7:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <4 x i64> %a to <4 x float>
  ret <4 x float> %b
}

define <8 x double> @ulto8f64(<8 x i64> %a) {
; GENERIC-LABEL: ulto8f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtuqq2pd %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ulto8f64:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtuqq2pd %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <8 x i64> %a to <8 x double>
  ret <8 x double> %b
}

define <16 x double> @ulto16f64(<16 x i64> %a) {
; GENERIC-LABEL: ulto16f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtuqq2pd %zmm0, %zmm0
; GENERIC-NEXT:    vcvtuqq2pd %zmm1, %zmm1
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ulto16f64:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtuqq2pd %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    vcvtuqq2pd %zmm1, %zmm1 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <16 x i64> %a to <16 x double>
  ret <16 x double> %b
}

define <16 x i32> @f64to16si(<16 x float> %a) nounwind {
; GENERIC-LABEL: f64to16si:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttps2dq %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f64to16si:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttps2dq %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = fptosi <16 x float> %a to <16 x i32>
  ret <16 x i32> %b
}

define <16 x i32> @f32to16ui(<16 x float> %a) nounwind {
; GENERIC-LABEL: f32to16ui:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttps2udq %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f32to16ui:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttps2udq %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = fptoui <16 x float> %a to <16 x i32>
  ret <16 x i32> %b
}

define <16 x i8> @f32to16uc(<16 x float> %f) {
; GENERIC-LABEL: f32to16uc:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttps2udq %zmm0, %zmm0
; GENERIC-NEXT:    vpmovdb %zmm0, %xmm0
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f32to16uc:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttps2udq %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    vpmovdb %zmm0, %xmm0 # sched: [4:2.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = fptoui <16 x float> %f to <16 x i8>
  ret <16 x i8> %res
}

define <16 x i16> @f32to16us(<16 x float> %f) {
; GENERIC-LABEL: f32to16us:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttps2udq %zmm0, %zmm0
; GENERIC-NEXT:    vpmovdw %zmm0, %ymm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f32to16us:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttps2udq %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    vpmovdw %zmm0, %ymm0 # sched: [4:2.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = fptoui <16 x float> %f to <16 x i16>
  ret <16 x i16> %res
}

define <8 x i32> @f32to8ui(<8 x float> %a) nounwind {
; GENERIC-LABEL: f32to8ui:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttps2udq %ymm0, %ymm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f32to8ui:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttps2udq %ymm0, %ymm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = fptoui <8 x float> %a to <8 x i32>
  ret <8 x i32> %b
}

define <4 x i32> @f32to4ui(<4 x float> %a) nounwind {
; GENERIC-LABEL: f32to4ui:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttps2udq %xmm0, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f32to4ui:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttps2udq %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = fptoui <4 x float> %a to <4 x i32>
  ret <4 x i32> %b
}

define <8 x i32> @f64to8ui(<8 x double> %a) nounwind {
; GENERIC-LABEL: f64to8ui:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttpd2udq %zmm0, %ymm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f64to8ui:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttpd2udq %zmm0, %ymm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = fptoui <8 x double> %a to <8 x i32>
  ret <8 x i32> %b
}

define <8 x i16> @f64to8us(<8 x double> %f) {
; GENERIC-LABEL: f64to8us:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttpd2dq %zmm0, %ymm0
; GENERIC-NEXT:    vpmovdw %ymm0, %xmm0
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f64to8us:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttpd2dq %zmm0, %ymm0 # sched: [7:1.00]
; SKX-NEXT:    vpmovdw %ymm0, %xmm0 # sched: [4:2.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = fptoui <8 x double> %f to <8 x i16>
  ret <8 x i16> %res
}

define <8 x i8> @f64to8uc(<8 x double> %f) {
; GENERIC-LABEL: f64to8uc:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttpd2dq %zmm0, %ymm0
; GENERIC-NEXT:    vpmovdw %ymm0, %xmm0
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f64to8uc:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttpd2dq %zmm0, %ymm0 # sched: [7:1.00]
; SKX-NEXT:    vpmovdw %ymm0, %xmm0 # sched: [4:2.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = fptoui <8 x double> %f to <8 x i8>
  ret <8 x i8> %res
}

define <4 x i32> @f64to4ui(<4 x double> %a) nounwind {
; GENERIC-LABEL: f64to4ui:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttpd2udq %ymm0, %xmm0
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f64to4ui:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttpd2udq %ymm0, %xmm0 # sched: [7:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = fptoui <4 x double> %a to <4 x i32>
  ret <4 x i32> %b
}

define <8 x double> @sito8f64(<8 x i32> %a) {
; GENERIC-LABEL: sito8f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtdq2pd %ymm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sito8f64:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtdq2pd %ymm0, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <8 x i32> %a to <8 x double>
  ret <8 x double> %b
}
define <8 x double> @i32to8f64_mask(<8 x double> %a, <8 x i32> %b, i8 %c) nounwind {
; GENERIC-LABEL: i32to8f64_mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vcvtdq2pd %ymm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: i32to8f64_mask:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm1, %zmm0 {%k1} # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
; VLNOBW-LABEL: i32to8f64_mask:
; VLNOBW:       # BB#0:
; VLNOBW-NEXT:    kmovw %edi, %k1
; VLNOBW-NEXT:    vcvtdq2pd %ymm1, %zmm0 {%k1}
; VLNOBW-NEXT:    ret{{[l|q]}}
  %1 = bitcast i8 %c to <8 x i1>
  %2 = sitofp <8 x i32> %b to <8 x double>
  %3 = select <8 x i1> %1, <8 x double> %2, <8 x double> %a
  ret <8 x double> %3
}
define <8 x double> @sito8f64_maskz(<8 x i32> %a, i8 %b) nounwind {
; GENERIC-LABEL: sito8f64_maskz:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vcvtdq2pd %ymm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sito8f64_maskz:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm0, %zmm0 {%k1} {z} # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
; VLNOBW-LABEL: sito8f64_maskz:
; VLNOBW:       # BB#0:
; VLNOBW-NEXT:    kmovw %edi, %k1
; VLNOBW-NEXT:    vcvtdq2pd %ymm0, %zmm0 {%k1} {z}
; VLNOBW-NEXT:    ret{{[l|q]}}
  %1 = bitcast i8 %b to <8 x i1>
  %2 = sitofp <8 x i32> %a to <8 x double>
  %3 = select <8 x i1> %1, <8 x double> %2, <8 x double> zeroinitializer
  ret <8 x double> %3
}

define <8 x i32> @f64to8si(<8 x double> %a) {
; GENERIC-LABEL: f64to8si:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttpd2dq %zmm0, %ymm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f64to8si:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttpd2dq %zmm0, %ymm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = fptosi <8 x double> %a to <8 x i32>
  ret <8 x i32> %b
}

define <4 x i32> @f64to4si(<4 x double> %a) {
; GENERIC-LABEL: f64to4si:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttpd2dq %ymm0, %xmm0 # sched: [4:1.00]
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f64to4si:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttpd2dq %ymm0, %xmm0 # sched: [7:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = fptosi <4 x double> %a to <4 x i32>
  ret <4 x i32> %b
}

define <16 x float> @f64to16f32(<16 x double> %b) nounwind {
; GENERIC-LABEL: f64to16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtpd2ps %zmm0, %ymm0
; GENERIC-NEXT:    vcvtpd2ps %zmm1, %ymm1
; GENERIC-NEXT:    vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f64to16f32:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtpd2ps %zmm0, %ymm0 # sched: [7:1.00]
; SKX-NEXT:    vcvtpd2ps %zmm1, %ymm1 # sched: [7:1.00]
; SKX-NEXT:    vinsertf64x4 $1, %ymm1, %zmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = fptrunc <16 x double> %b to <16 x float>
  ret <16 x float> %a
}

define <4 x float> @f64to4f32(<4 x double> %b) {
; GENERIC-LABEL: f64to4f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtpd2ps %ymm0, %xmm0 # sched: [4:1.00]
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f64to4f32:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtpd2ps %ymm0, %xmm0 # sched: [7:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = fptrunc <4 x double> %b to <4 x float>
  ret <4 x float> %a
}

define <4 x float> @f64to4f32_mask(<4 x double> %b, <4 x i1> %mask) {
; GENERIC-LABEL: f64to4f32_mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm1, %xmm1, %k1
; GENERIC-NEXT:    vcvtpd2ps %ymm0, %xmm0 {%k1} {z}
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f64to4f32_mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %xmm1, %xmm1 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm1, %xmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vcvtpd2ps %ymm0, %xmm0 {%k1} {z} # sched: [7:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = fptrunc <4 x double> %b to <4 x float>
  %c = select <4 x i1>%mask, <4 x float>%a, <4 x float> zeroinitializer
  ret <4 x float> %c
}

define <4 x float> @f64tof32_inreg(<2 x double> %a0, <4 x float> %a1) nounwind {
; GENERIC-LABEL: f64tof32_inreg:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtsd2ss %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f64tof32_inreg:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtsd2ss %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %ext = extractelement <2 x double> %a0, i32 0
  %cvt = fptrunc double %ext to float
  %res = insertelement <4 x float> %a1, float %cvt, i32 0
  ret <4 x float> %res
}

define <8 x double> @f32to8f64(<8 x float> %b) nounwind {
; GENERIC-LABEL: f32to8f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtps2pd %ymm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f32to8f64:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtps2pd %ymm0, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = fpext <8 x float> %b to <8 x double>
  ret <8 x double> %a
}

define <4 x double> @f32to4f64_mask(<4 x float> %b, <4 x double> %b1, <4 x double> %a1) {
; GENERIC-LABEL: f32to4f64_mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcmpltpd %ymm2, %ymm1, %k1
; GENERIC-NEXT:    vcvtps2pd %xmm0, %ymm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f32to4f64_mask:
; SKX:       # BB#0:
; SKX-NEXT:    vcmpltpd %ymm2, %ymm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vcvtps2pd %xmm0, %ymm0 {%k1} {z} # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = fpext <4 x float> %b to <4 x double>
  %mask = fcmp ogt <4 x double> %a1, %b1
  %c = select <4 x i1> %mask, <4 x double> %a, <4 x double> zeroinitializer
  ret <4 x double> %c
}

define <2 x double> @f32tof64_inreg(<2 x double> %a0, <4 x float> %a1) nounwind {
; GENERIC-LABEL: f32tof64_inreg:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtss2sd %xmm1, %xmm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f32tof64_inreg:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtss2sd %xmm1, %xmm0, %xmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %ext = extractelement <4 x float> %a1, i32 0
  %cvt = fpext float %ext to double
  %res = insertelement <2 x double> %a0, double %cvt, i32 0
  ret <2 x double> %res
}

define double @sltof64_load(i64* nocapture %e) {
; GENERIC-LABEL: sltof64_load:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vcvtsi2sdq (%rdi), %xmm0, %xmm0 # sched: [9:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sltof64_load:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vcvtsi2sdq (%rdi), %xmm0, %xmm0 # sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %tmp1 = load i64, i64* %e, align 8
  %conv = sitofp i64 %tmp1 to double
  ret double %conv
}

define double @sitof64_load(i32* %e) {
; GENERIC-LABEL: sitof64_load:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vcvtsi2sdl (%rdi), %xmm0, %xmm0 # sched: [9:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sitof64_load:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vcvtsi2sdl (%rdi), %xmm0, %xmm0 # sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %tmp1 = load i32, i32* %e, align 4
  %conv = sitofp i32 %tmp1 to double
  ret double %conv
}

define float @sitof32_load(i32* %e) {
; GENERIC-LABEL: sitof32_load:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vcvtsi2ssl (%rdi), %xmm0, %xmm0 # sched: [10:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sitof32_load:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vcvtsi2ssl (%rdi), %xmm0, %xmm0 # sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %tmp1 = load i32, i32* %e, align 4
  %conv = sitofp i32 %tmp1 to float
  ret float %conv
}

define float @sltof32_load(i64* %e) {
; GENERIC-LABEL: sltof32_load:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vcvtsi2ssq (%rdi), %xmm0, %xmm0 # sched: [10:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sltof32_load:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vcvtsi2ssq (%rdi), %xmm0, %xmm0 # sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %tmp1 = load i64, i64* %e, align 8
  %conv = sitofp i64 %tmp1 to float
  ret float %conv
}

define void @f32tof64_loadstore() {
; GENERIC-LABEL: f32tof64_loadstore:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero sched: [6:0.50]
; GENERIC-NEXT:    vcvtss2sd %xmm0, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vmovsd %xmm0, -{{[0-9]+}}(%rsp) # sched: [5:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f32tof64_loadstore:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero sched: [5:0.50]
; SKX-NEXT:    vcvtss2sd %xmm0, %xmm0, %xmm0 # sched: [5:1.00]
; SKX-NEXT:    vmovsd %xmm0, -{{[0-9]+}}(%rsp) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %f = alloca float, align 4
  %d = alloca double, align 8
  %tmp = load float, float* %f, align 4
  %conv = fpext float %tmp to double
  store double %conv, double* %d, align 8
  ret void
}

define void @f64tof32_loadstore() nounwind uwtable {
; GENERIC-LABEL: f64tof32_loadstore:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vmovsd {{.*#+}} xmm0 = mem[0],zero sched: [6:0.50]
; GENERIC-NEXT:    vcvtsd2ss %xmm0, %xmm0, %xmm0 # sched: [4:1.00]
; GENERIC-NEXT:    vmovss %xmm0, -{{[0-9]+}}(%rsp) # sched: [5:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: f64tof32_loadstore:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vmovsd {{.*#+}} xmm0 = mem[0],zero sched: [5:0.50]
; SKX-NEXT:    vcvtsd2ss %xmm0, %xmm0, %xmm0 # sched: [5:1.00]
; SKX-NEXT:    vmovss %xmm0, -{{[0-9]+}}(%rsp) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %f = alloca float, align 4
  %d = alloca double, align 8
  %tmp = load double, double* %d, align 8
  %conv = fptrunc double %tmp to float
  store float %conv, float* %f, align 4
  ret void
}

define double @long_to_double(i64 %x) {
; GENERIC-LABEL: long_to_double:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovq %rdi, %xmm0 # sched: [1:0.33]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: long_to_double:
; SKX:       # BB#0:
; SKX-NEXT:    vmovq %rdi, %xmm0 # sched: [1:0.25]
; SKX-NEXT:    retq # sched: [7:1.00]
   %res = bitcast i64 %x to double
   ret double %res
}

define i64 @double_to_long(double %x) {
; GENERIC-LABEL: double_to_long:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovq %xmm0, %rax # sched: [1:0.33]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: double_to_long:
; SKX:       # BB#0:
; SKX-NEXT:    vmovq %xmm0, %rax # sched: [1:0.25]
; SKX-NEXT:    retq # sched: [7:1.00]
   %res = bitcast double %x to i64
   ret i64 %res
}

define float @int_to_float(i32 %x) {
; GENERIC-LABEL: int_to_float:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovd %edi, %xmm0 # sched: [1:0.33]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: int_to_float:
; SKX:       # BB#0:
; SKX-NEXT:    vmovd %edi, %xmm0 # sched: [1:0.25]
; SKX-NEXT:    retq # sched: [7:1.00]
   %res = bitcast i32 %x to float
   ret float %res
}

define i32 @float_to_int(float %x) {
; GENERIC-LABEL: float_to_int:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovd %xmm0, %eax # sched: [1:0.33]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: float_to_int:
; SKX:       # BB#0:
; SKX-NEXT:    vmovd %xmm0, %eax # sched: [1:0.25]
; SKX-NEXT:    retq # sched: [7:1.00]
   %res = bitcast float %x to i32
   ret i32 %res
}

define <16 x double> @uito16f64(<16 x i32> %a) nounwind {
; GENERIC-LABEL: uito16f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtudq2pd %ymm0, %zmm2
; GENERIC-NEXT:    vextractf64x4 $1, %zmm0, %ymm0
; GENERIC-NEXT:    vcvtudq2pd %ymm0, %zmm1
; GENERIC-NEXT:    vmovaps %zmm2, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: uito16f64:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtudq2pd %ymm0, %zmm2 # sched: [7:1.00]
; SKX-NEXT:    vextractf64x4 $1, %zmm0, %ymm0 # sched: [3:1.00]
; SKX-NEXT:    vcvtudq2pd %ymm0, %zmm1 # sched: [7:1.00]
; SKX-NEXT:    vmovaps %zmm2, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <16 x i32> %a to <16 x double>
  ret <16 x double> %b
}

define <8 x float> @slto8f32(<8 x i64> %a) {
; GENERIC-LABEL: slto8f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtqq2ps %zmm0, %ymm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: slto8f32:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtqq2ps %zmm0, %ymm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <8 x i64> %a to <8 x float>
  ret <8 x float> %b
}

define <16 x float> @slto16f32(<16 x i64> %a) {
; GENERIC-LABEL: slto16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtqq2ps %zmm0, %ymm0
; GENERIC-NEXT:    vcvtqq2ps %zmm1, %ymm1
; GENERIC-NEXT:    vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: slto16f32:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtqq2ps %zmm0, %ymm0 # sched: [7:1.00]
; SKX-NEXT:    vcvtqq2ps %zmm1, %ymm1 # sched: [7:1.00]
; SKX-NEXT:    vinsertf64x4 $1, %ymm1, %zmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <16 x i64> %a to <16 x float>
  ret <16 x float> %b
}

define <8 x double> @slto8f64(<8 x i64> %a) {
; GENERIC-LABEL: slto8f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtqq2pd %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: slto8f64:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtqq2pd %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <8 x i64> %a to <8 x double>
  ret <8 x double> %b
}

define <16 x double> @slto16f64(<16 x i64> %a) {
; GENERIC-LABEL: slto16f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtqq2pd %zmm0, %zmm0
; GENERIC-NEXT:    vcvtqq2pd %zmm1, %zmm1
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: slto16f64:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtqq2pd %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    vcvtqq2pd %zmm1, %zmm1 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <16 x i64> %a to <16 x double>
  ret <16 x double> %b
}

define <8 x float> @ulto8f32(<8 x i64> %a) {
; GENERIC-LABEL: ulto8f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtuqq2ps %zmm0, %ymm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ulto8f32:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtuqq2ps %zmm0, %ymm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <8 x i64> %a to <8 x float>
  ret <8 x float> %b
}

define <16 x float> @ulto16f32(<16 x i64> %a) {
; GENERIC-LABEL: ulto16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtuqq2ps %zmm0, %ymm0
; GENERIC-NEXT:    vcvtuqq2ps %zmm1, %ymm1
; GENERIC-NEXT:    vinsertf64x4 $1, %ymm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ulto16f32:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtuqq2ps %zmm0, %ymm0 # sched: [7:1.00]
; SKX-NEXT:    vcvtuqq2ps %zmm1, %ymm1 # sched: [7:1.00]
; SKX-NEXT:    vinsertf64x4 $1, %ymm1, %zmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <16 x i64> %a to <16 x float>
  ret <16 x float> %b
}

define <8 x double> @uito8f64_mask(<8 x double> %a, <8 x i32> %b, i8 %c) nounwind {
; GENERIC-LABEL: uito8f64_mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vcvtudq2pd %ymm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: uito8f64_mask:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vcvtudq2pd %ymm1, %zmm0 {%k1} # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
; VLNOBW-LABEL: uito8f64_mask:
; VLNOBW:       # BB#0:
; VLNOBW-NEXT:    kmovw %edi, %k1
; VLNOBW-NEXT:    vcvtudq2pd %ymm1, %zmm0 {%k1}
; VLNOBW-NEXT:    ret{{[l|q]}}
  %1 = bitcast i8 %c to <8 x i1>
  %2 = uitofp <8 x i32> %b to <8 x double>
  %3 = select <8 x i1> %1, <8 x double> %2, <8 x double> %a
  ret <8 x double> %3
}
define <8 x double> @uito8f64_maskz(<8 x i32> %a, i8 %b) nounwind {
; GENERIC-LABEL: uito8f64_maskz:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vcvtudq2pd %ymm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: uito8f64_maskz:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vcvtudq2pd %ymm0, %zmm0 {%k1} {z} # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %1 = bitcast i8 %b to <8 x i1>
  %2 = uitofp <8 x i32> %a to <8 x double>
  %3 = select <8 x i1> %1, <8 x double> %2, <8 x double> zeroinitializer
  ret <8 x double> %3
}

define <4 x double> @uito4f64(<4 x i32> %a) nounwind {
; GENERIC-LABEL: uito4f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtudq2pd %xmm0, %ymm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: uito4f64:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtudq2pd %xmm0, %ymm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <4 x i32> %a to <4 x double>
  ret <4 x double> %b
}

define <16 x float> @uito16f32(<16 x i32> %a) nounwind {
; GENERIC-LABEL: uito16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtudq2ps %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: uito16f32:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtudq2ps %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <16 x i32> %a to <16 x float>
  ret <16 x float> %b
}

define <8 x double> @uito8f64(<8 x i32> %a) {
; GENERIC-LABEL: uito8f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtudq2pd %ymm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: uito8f64:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtudq2pd %ymm0, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <8 x i32> %a to <8 x double>
  ret <8 x double> %b
}

define <8 x float> @uito8f32(<8 x i32> %a) nounwind {
; GENERIC-LABEL: uito8f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtudq2ps %ymm0, %ymm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: uito8f32:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtudq2ps %ymm0, %ymm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <8 x i32> %a to <8 x float>
  ret <8 x float> %b
}

define <4 x float> @uito4f32(<4 x i32> %a) nounwind {
; GENERIC-LABEL: uito4f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtudq2ps %xmm0, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: uito4f32:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtudq2ps %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <4 x i32> %a to <4 x float>
  ret <4 x float> %b
}

define i32 @fptosi(float %a) nounwind {
; GENERIC-LABEL: fptosi:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttss2si %xmm0, %eax # sched: [5:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: fptosi:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttss2si %xmm0, %eax # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = fptosi float %a to i32
  ret i32 %b
}

define i32 @fptoui(float %a) nounwind {
; GENERIC-LABEL: fptoui:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvttss2usi %xmm0, %eax
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: fptoui:
; SKX:       # BB#0:
; SKX-NEXT:    vcvttss2usi %xmm0, %eax
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = fptoui float %a to i32
  ret i32 %b
}

define float @uitof32(i32 %a) nounwind {
; GENERIC-LABEL: uitof32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtusi2ssl %edi, %xmm0, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: uitof32:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtusi2ssl %edi, %xmm0, %xmm0 # sched: [5:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp i32 %a to float
  ret float %b
}

define double @uitof64(i32 %a) nounwind {
; GENERIC-LABEL: uitof64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtusi2sdl %edi, %xmm0, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: uitof64:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtusi2sdl %edi, %xmm0, %xmm0 # sched: [5:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp i32 %a to double
  ret double %b
}

define <16 x float> @sbto16f32(<16 x i32> %a) {
; GENERIC-LABEL: sbto16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpgtd %zmm0, %zmm1, %k0
; GENERIC-NEXT:    vpmovm2d %k0, %zmm0
; GENERIC-NEXT:    vcvtdq2ps %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sbto16f32:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpcmpgtd %zmm0, %zmm1, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2d %k0, %zmm0
; SKX-NEXT:    vcvtdq2ps %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp slt <16 x i32> %a, zeroinitializer
  %1 = sitofp <16 x i1> %mask to <16 x float>
  ret <16 x float> %1
}

define <16 x float> @scto16f32(<16 x i8> %a) {
; GENERIC-LABEL: scto16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxbd %xmm0, %zmm0
; GENERIC-NEXT:    vcvtdq2ps %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: scto16f32:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxbd %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    vcvtdq2ps %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %1 = sitofp <16 x i8> %a to <16 x float>
  ret <16 x float> %1
}

define <16 x float> @ssto16f32(<16 x i16> %a) {
; GENERIC-LABEL: ssto16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxwd %ymm0, %zmm0
; GENERIC-NEXT:    vcvtdq2ps %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ssto16f32:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxwd %ymm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    vcvtdq2ps %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %1 = sitofp <16 x i16> %a to <16 x float>
  ret <16 x float> %1
}

define <8 x double> @ssto16f64(<8 x i16> %a) {
; GENERIC-LABEL: ssto16f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxwd %xmm0, %ymm0 # sched: [1:1.00]
; GENERIC-NEXT:    vcvtdq2pd %ymm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ssto16f64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxwd %xmm0, %ymm0 # sched: [3:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm0, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %1 = sitofp <8 x i16> %a to <8 x double>
  ret <8 x double> %1
}

define <8 x double> @scto8f64(<8 x i8> %a) {
; GENERIC-LABEL: scto8f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero sched: [1:1.00]
; GENERIC-NEXT:    vpslld $24, %ymm0, %ymm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpsrad $24, %ymm0, %ymm0 # sched: [1:1.00]
; GENERIC-NEXT:    vcvtdq2pd %ymm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: scto8f64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero sched: [3:1.00]
; SKX-NEXT:    vpslld $24, %ymm0, %ymm0 # sched: [1:0.50]
; SKX-NEXT:    vpsrad $24, %ymm0, %ymm0 # sched: [1:0.50]
; SKX-NEXT:    vcvtdq2pd %ymm0, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %1 = sitofp <8 x i8> %a to <8 x double>
  ret <8 x double> %1
}

define <16 x double> @scto16f64(<16 x i8> %a) {
; GENERIC-LABEL: scto16f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxbd %xmm0, %zmm1
; GENERIC-NEXT:    vcvtdq2pd %ymm1, %zmm0
; GENERIC-NEXT:    vextracti64x4 $1, %zmm1, %ymm1
; GENERIC-NEXT:    vcvtdq2pd %ymm1, %zmm1
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: scto16f64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxbd %xmm0, %zmm1 # sched: [3:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm1, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    vextracti64x4 $1, %zmm1, %ymm1 # sched: [3:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm1, %zmm1 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <16 x i8> %a to <16 x double>
  ret <16 x double> %b
}

define <16 x double> @sbto16f64(<16 x double> %a) {
; GENERIC-LABEL: sbto16f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorpd %xmm2, %xmm2, %xmm2 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpltpd %zmm1, %zmm2, %k0
; GENERIC-NEXT:    vcmpltpd %zmm0, %zmm2, %k1
; GENERIC-NEXT:    vpmovm2d %k1, %ymm0
; GENERIC-NEXT:    vcvtdq2pd %ymm0, %zmm0
; GENERIC-NEXT:    vpmovm2d %k0, %ymm1
; GENERIC-NEXT:    vcvtdq2pd %ymm1, %zmm1
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sbto16f64:
; SKX:       # BB#0:
; SKX-NEXT:    vxorpd %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vcmpltpd %zmm1, %zmm2, %k0 # sched: [3:1.00]
; SKX-NEXT:    vcmpltpd %zmm0, %zmm2, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2d %k1, %ymm0
; SKX-NEXT:    vcvtdq2pd %ymm0, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    vpmovm2d %k0, %ymm1
; SKX-NEXT:    vcvtdq2pd %ymm1, %zmm1 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %cmpres = fcmp ogt <16 x double> %a, zeroinitializer
  %1 = sitofp <16 x i1> %cmpres to <16 x double>
  ret <16 x double> %1
}

define <8 x double> @sbto8f64(<8 x double> %a) {
; GENERIC-LABEL: sbto8f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorpd %xmm1, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpltpd %zmm0, %zmm1, %k0
; GENERIC-NEXT:    vpmovm2d %k0, %ymm0
; GENERIC-NEXT:    vcvtdq2pd %ymm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sbto8f64:
; SKX:       # BB#0:
; SKX-NEXT:    vxorpd %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vcmpltpd %zmm0, %zmm1, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2d %k0, %ymm0
; SKX-NEXT:    vcvtdq2pd %ymm0, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %cmpres = fcmp ogt <8 x double> %a, zeroinitializer
  %1 = sitofp <8 x i1> %cmpres to <8 x double>
  ret <8 x double> %1
}

define <8 x float> @sbto8f32(<8 x float> %a) {
; GENERIC-LABEL: sbto8f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorps %xmm1, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpltps %ymm0, %ymm1, %k0
; GENERIC-NEXT:    vpmovm2d %k0, %ymm0
; GENERIC-NEXT:    vcvtdq2ps %ymm0, %ymm0 # sched: [3:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sbto8f32:
; SKX:       # BB#0:
; SKX-NEXT:    vxorps %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vcmpltps %ymm0, %ymm1, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2d %k0, %ymm0
; SKX-NEXT:    vcvtdq2ps %ymm0, %ymm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %cmpres = fcmp ogt <8 x float> %a, zeroinitializer
  %1 = sitofp <8 x i1> %cmpres to <8 x float>
  ret <8 x float> %1
}

define <4 x float> @sbto4f32(<4 x float> %a) {
; GENERIC-LABEL: sbto4f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorps %xmm1, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpltps %xmm0, %xmm1, %k0
; GENERIC-NEXT:    vpmovm2d %k0, %xmm0
; GENERIC-NEXT:    vcvtdq2ps %xmm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sbto4f32:
; SKX:       # BB#0:
; SKX-NEXT:    vxorps %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vcmpltps %xmm0, %xmm1, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2d %k0, %xmm0
; SKX-NEXT:    vcvtdq2ps %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %cmpres = fcmp ogt <4 x float> %a, zeroinitializer
  %1 = sitofp <4 x i1> %cmpres to <4 x float>
  ret <4 x float> %1
}

define <4 x double> @sbto4f64(<4 x double> %a) {
; GENERIC-LABEL: sbto4f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorpd %xmm1, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpltpd %ymm0, %ymm1, %k0
; GENERIC-NEXT:    vpmovm2d %k0, %xmm0
; GENERIC-NEXT:    vcvtdq2pd %xmm0, %ymm0 # sched: [4:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sbto4f64:
; SKX:       # BB#0:
; SKX-NEXT:    vxorpd %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vcmpltpd %ymm0, %ymm1, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2d %k0, %xmm0
; SKX-NEXT:    vcvtdq2pd %xmm0, %ymm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %cmpres = fcmp ogt <4 x double> %a, zeroinitializer
  %1 = sitofp <4 x i1> %cmpres to <4 x double>
  ret <4 x double> %1
}

define <2 x float> @sbto2f32(<2 x float> %a) {
; GENERIC-LABEL: sbto2f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorps %xmm1, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpltps %xmm0, %xmm1, %k0
; GENERIC-NEXT:    vpmovm2d %k0, %xmm0
; GENERIC-NEXT:    vcvtdq2ps %xmm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sbto2f32:
; SKX:       # BB#0:
; SKX-NEXT:    vxorps %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vcmpltps %xmm0, %xmm1, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2d %k0, %xmm0
; SKX-NEXT:    vcvtdq2ps %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %cmpres = fcmp ogt <2 x float> %a, zeroinitializer
  %1 = sitofp <2 x i1> %cmpres to <2 x float>
  ret <2 x float> %1
}

define <2 x double> @sbto2f64(<2 x double> %a) {
; GENERIC-LABEL: sbto2f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorpd %xmm1, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpltpd %xmm0, %xmm1, %k0
; GENERIC-NEXT:    vpmovm2q %k0, %xmm0
; GENERIC-NEXT:    vcvtqq2pd %xmm0, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sbto2f64:
; SKX:       # BB#0:
; SKX-NEXT:    vxorpd %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vcmpltpd %xmm0, %xmm1, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2q %k0, %xmm0
; SKX-NEXT:    vcvtqq2pd %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %cmpres = fcmp ogt <2 x double> %a, zeroinitializer
  %1 = sitofp <2 x i1> %cmpres to <2 x double>
  ret <2 x double> %1
}

define <16 x float> @ucto16f32(<16 x i8> %a) {
; GENERIC-LABEL: ucto16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxbd {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; GENERIC-NEXT:    vcvtdq2ps %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ucto16f32:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxbd {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero sched: [3:1.00]
; SKX-NEXT:    vcvtdq2ps %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <16 x i8> %a to <16 x float>
  ret <16 x float>%b
}

define <8 x double> @ucto8f64(<8 x i8> %a) {
; GENERIC-LABEL: ucto8f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0 # sched: [7:0.50]
; GENERIC-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero sched: [1:1.00]
; GENERIC-NEXT:    vcvtdq2pd %ymm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ucto8f64:
; SKX:       # BB#0:
; SKX-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm0 # sched: [7:0.50]
; SKX-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero sched: [3:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm0, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <8 x i8> %a to <8 x double>
  ret <8 x double> %b
}

define <16 x float> @swto16f32(<16 x i16> %a) {
; GENERIC-LABEL: swto16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxwd %ymm0, %zmm0
; GENERIC-NEXT:    vcvtdq2ps %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: swto16f32:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxwd %ymm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    vcvtdq2ps %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <16 x i16> %a to <16 x float>
  ret <16 x float> %b
}

define <8 x double> @swto8f64(<8 x i16> %a) {
; GENERIC-LABEL: swto8f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxwd %xmm0, %ymm0 # sched: [1:1.00]
; GENERIC-NEXT:    vcvtdq2pd %ymm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: swto8f64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxwd %xmm0, %ymm0 # sched: [3:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm0, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <8 x i16> %a to <8 x double>
  ret <8 x double> %b
}

define <16 x double> @swto16f64(<16 x i16> %a) {
; GENERIC-LABEL: swto16f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxwd %ymm0, %zmm1
; GENERIC-NEXT:    vcvtdq2pd %ymm1, %zmm0
; GENERIC-NEXT:    vextracti64x4 $1, %zmm1, %ymm1
; GENERIC-NEXT:    vcvtdq2pd %ymm1, %zmm1
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: swto16f64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxwd %ymm0, %zmm1 # sched: [3:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm1, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    vextracti64x4 $1, %zmm1, %ymm1 # sched: [3:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm1, %zmm1 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <16 x i16> %a to <16 x double>
  ret <16 x double> %b
}

define <16 x double> @ucto16f64(<16 x i8> %a) {
; GENERIC-LABEL: ucto16f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxbd {{.*#+}} zmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; GENERIC-NEXT:    vcvtdq2pd %ymm1, %zmm0
; GENERIC-NEXT:    vextracti64x4 $1, %zmm1, %ymm1
; GENERIC-NEXT:    vcvtdq2pd %ymm1, %zmm1
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ucto16f64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxbd {{.*#+}} zmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero sched: [3:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm1, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    vextracti64x4 $1, %zmm1, %ymm1 # sched: [3:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm1, %zmm1 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <16 x i8> %a to <16 x double>
  ret <16 x double> %b
}

define <16 x float> @uwto16f32(<16 x i16> %a) {
; GENERIC-LABEL: uwto16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxwd {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero
; GENERIC-NEXT:    vcvtdq2ps %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: uwto16f32:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxwd {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero sched: [3:1.00]
; SKX-NEXT:    vcvtdq2ps %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <16 x i16> %a to <16 x float>
  ret <16 x float> %b
}

define <8 x double> @uwto8f64(<8 x i16> %a) {
; GENERIC-LABEL: uwto8f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero sched: [1:1.00]
; GENERIC-NEXT:    vcvtdq2pd %ymm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: uwto8f64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero sched: [3:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm0, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <8 x i16> %a to <8 x double>
  ret <8 x double> %b
}

define <16 x double> @uwto16f64(<16 x i16> %a) {
; GENERIC-LABEL: uwto16f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxwd {{.*#+}} zmm1 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero
; GENERIC-NEXT:    vcvtdq2pd %ymm1, %zmm0
; GENERIC-NEXT:    vextracti64x4 $1, %zmm1, %ymm1
; GENERIC-NEXT:    vcvtdq2pd %ymm1, %zmm1
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: uwto16f64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxwd {{.*#+}} zmm1 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero sched: [3:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm1, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    vextracti64x4 $1, %zmm1, %ymm1 # sched: [3:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm1, %zmm1 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <16 x i16> %a to <16 x double>
  ret <16 x double> %b
}

define <16 x float> @sito16f32(<16 x i32> %a) {
; GENERIC-LABEL: sito16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtdq2ps %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sito16f32:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtdq2ps %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <16 x i32> %a to <16 x float>
  ret <16 x float> %b
}

define <16 x double> @sito16f64(<16 x i32> %a) {
; GENERIC-LABEL: sito16f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtdq2pd %ymm0, %zmm2
; GENERIC-NEXT:    vextractf64x4 $1, %zmm0, %ymm0
; GENERIC-NEXT:    vcvtdq2pd %ymm0, %zmm1
; GENERIC-NEXT:    vmovaps %zmm2, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sito16f64:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtdq2pd %ymm0, %zmm2 # sched: [7:1.00]
; SKX-NEXT:    vextractf64x4 $1, %zmm0, %ymm0 # sched: [3:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm0, %zmm1 # sched: [7:1.00]
; SKX-NEXT:    vmovaps %zmm2, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = sitofp <16 x i32> %a to <16 x double>
  ret <16 x double> %b
}

define <16 x float> @usto16f32(<16 x i16> %a) {
; GENERIC-LABEL: usto16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxwd {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero
; GENERIC-NEXT:    vcvtdq2ps %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: usto16f32:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxwd {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero sched: [3:1.00]
; SKX-NEXT:    vcvtdq2ps %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = uitofp <16 x i16> %a to <16 x float>
  ret <16 x float> %b
}

define <16 x float> @ubto16f32(<16 x i32> %a) {
; GENERIC-LABEL: ubto16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpgtd %zmm0, %zmm1, %k1
; GENERIC-NEXT:    vpbroadcastd {{.*}}(%rip), %zmm0 {%k1} {z}
; GENERIC-NEXT:    vcvtdq2ps %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ubto16f32:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpcmpgtd %zmm0, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpbroadcastd {{.*}}(%rip), %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    vcvtdq2ps %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp slt <16 x i32> %a, zeroinitializer
  %1 = uitofp <16 x i1> %mask to <16 x float>
  ret <16 x float> %1
}

define <16 x double> @ubto16f64(<16 x i32> %a) {
; GENERIC-LABEL: ubto16f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpgtd %zmm0, %zmm1, %k1
; GENERIC-NEXT:    movl {{.*}}(%rip), %eax # sched: [5:0.50]
; GENERIC-NEXT:    vpbroadcastd %eax, %ymm0 {%k1} {z}
; GENERIC-NEXT:    vcvtdq2pd %ymm0, %zmm0
; GENERIC-NEXT:    kshiftrw $8, %k1, %k1
; GENERIC-NEXT:    vpbroadcastd %eax, %ymm1 {%k1} {z}
; GENERIC-NEXT:    vcvtdq2pd %ymm1, %zmm1
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ubto16f64:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpcmpgtd %zmm0, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    movl {{.*}}(%rip), %eax # sched: [5:0.50]
; SKX-NEXT:    vpbroadcastd %eax, %ymm0 {%k1} {z} # sched: [3:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm0, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    kshiftrw $8, %k1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpbroadcastd %eax, %ymm1 {%k1} {z} # sched: [3:1.00]
; SKX-NEXT:    vcvtdq2pd %ymm1, %zmm1 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp slt <16 x i32> %a, zeroinitializer
  %1 = uitofp <16 x i1> %mask to <16 x double>
  ret <16 x double> %1
}

define <8 x float> @ubto8f32(<8 x i32> %a) {
; GENERIC-LABEL: ubto8f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpgtd %ymm0, %ymm1, %k1
; GENERIC-NEXT:    vpbroadcastd {{.*}}(%rip), %ymm0 {%k1} {z}
; GENERIC-NEXT:    vcvtdq2ps %ymm0, %ymm0 # sched: [3:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ubto8f32:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpcmpgtd %ymm0, %ymm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpbroadcastd {{.*}}(%rip), %ymm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    vcvtdq2ps %ymm0, %ymm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp slt <8 x i32> %a, zeroinitializer
  %1 = uitofp <8 x i1> %mask to <8 x float>
  ret <8 x float> %1
}

define <8 x double> @ubto8f64(<8 x i32> %a) {
; GENERIC-LABEL: ubto8f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpgtd %ymm0, %ymm1, %k1
; GENERIC-NEXT:    vpbroadcastd {{.*}}(%rip), %ymm0 {%k1} {z}
; GENERIC-NEXT:    vcvtdq2pd %ymm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ubto8f64:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpcmpgtd %ymm0, %ymm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpbroadcastd {{.*}}(%rip), %ymm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    vcvtdq2pd %ymm0, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp slt <8 x i32> %a, zeroinitializer
  %1 = uitofp <8 x i1> %mask to <8 x double>
  ret <8 x double> %1
}

define <4 x float> @ubto4f32(<4 x i32> %a) {
; GENERIC-LABEL: ubto4f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpgtd %xmm0, %xmm1, %k1
; GENERIC-NEXT:    vpbroadcastd {{.*}}(%rip), %xmm0 {%k1} {z}
; GENERIC-NEXT:    vcvtdq2ps %xmm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ubto4f32:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpcmpgtd %xmm0, %xmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpbroadcastd {{.*}}(%rip), %xmm0 {%k1} {z} # sched: [7:0.50]
; SKX-NEXT:    vcvtdq2ps %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp slt <4 x i32> %a, zeroinitializer
  %1 = uitofp <4 x i1> %mask to <4 x float>
  ret <4 x float> %1
}

define <4 x double> @ubto4f64(<4 x i32> %a) {
; GENERIC-LABEL: ubto4f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpgtd %xmm0, %xmm1, %k1
; GENERIC-NEXT:    vpbroadcastd {{.*}}(%rip), %xmm0 {%k1} {z}
; GENERIC-NEXT:    vcvtdq2pd %xmm0, %ymm0 # sched: [4:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ubto4f64:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpcmpgtd %xmm0, %xmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpbroadcastd {{.*}}(%rip), %xmm0 {%k1} {z} # sched: [7:0.50]
; SKX-NEXT:    vcvtdq2pd %xmm0, %ymm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp slt <4 x i32> %a, zeroinitializer
  %1 = uitofp <4 x i1> %mask to <4 x double>
  ret <4 x double> %1
}

define <2 x float> @ubto2f32(<2 x i32> %a) {
; GENERIC-LABEL: ubto2f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3] sched: [1:0.50]
; GENERIC-NEXT:    vpcmpltuq %xmm1, %xmm0, %k1
; GENERIC-NEXT:    vpbroadcastd {{.*}}(%rip), %xmm0 {%k1} {z}
; GENERIC-NEXT:    vcvtdq2ps %xmm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ubto2f32:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3] sched: [1:0.33]
; SKX-NEXT:    vpcmpltuq %xmm1, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpbroadcastd {{.*}}(%rip), %xmm0 {%k1} {z} # sched: [7:0.50]
; SKX-NEXT:    vcvtdq2ps %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ult <2 x i32> %a, zeroinitializer
  %1 = uitofp <2 x i1> %mask to <2 x float>
  ret <2 x float> %1
}

define <2 x double> @ubto2f64(<2 x i32> %a) {
; GENERIC-LABEL: ubto2f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3] sched: [1:0.50]
; GENERIC-NEXT:    vpcmpltuq %xmm1, %xmm0, %k1
; GENERIC-NEXT:    vmovdqa64 {{.*}}(%rip), %xmm0 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    vcvtqq2pd %xmm0, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ubto2f64:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3] sched: [1:0.33]
; SKX-NEXT:    vpcmpltuq %xmm1, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovdqa64 {{.*}}(%rip), %xmm0 {%k1} {z} # sched: [7:0.50]
; SKX-NEXT:    vcvtqq2pd %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ult <2 x i32> %a, zeroinitializer
  %1 = uitofp <2 x i1> %mask to <2 x double>
  ret <2 x double> %1
}

define <8 x i16> @zext_8x8mem_to_8x16(<8 x i8> *%i , <8 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_8x8mem_to_8x16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k1
; GENERIC-NEXT:    vpmovzxbw {{.*#+}} xmm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_8x8mem_to_8x16:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxbw {{.*#+}} xmm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i8>,<8 x i8> *%i,align 1
  %x   = zext <8 x i8> %a to <8 x i16>
  %ret = select <8 x i1> %mask, <8 x i16> %x, <8 x i16> zeroinitializer
  ret <8 x i16> %ret
}

define <8 x i16> @sext_8x8mem_to_8x16(<8 x i8> *%i , <8 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_8x8mem_to_8x16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k1
; GENERIC-NEXT:    vpmovsxbw (%rdi), %xmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_8x8mem_to_8x16:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovsxbw (%rdi), %xmm0 {%k1} {z} # sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i8>,<8 x i8> *%i,align 1
  %x   = sext <8 x i8> %a to <8 x i16>
  %ret = select <8 x i1> %mask, <8 x i16> %x, <8 x i16> zeroinitializer
  ret <8 x i16> %ret
}


define <16 x i16> @zext_16x8mem_to_16x16(<16 x i8> *%i , <16 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_16x8mem_to_16x16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm0, %k1
; GENERIC-NEXT:    vpmovzxbw {{.*#+}} ymm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_16x8mem_to_16x16:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxbw {{.*#+}} ymm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <16 x i8>,<16 x i8> *%i,align 1
  %x   = zext <16 x i8> %a to <16 x i16>
  %ret = select <16 x i1> %mask, <16 x i16> %x, <16 x i16> zeroinitializer
  ret <16 x i16> %ret
}

define <16 x i16> @sext_16x8mem_to_16x16(<16 x i8> *%i , <16 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_16x8mem_to_16x16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm0, %k1
; GENERIC-NEXT:    vpmovsxbw (%rdi), %ymm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_16x8mem_to_16x16:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovsxbw (%rdi), %ymm0 {%k1} {z} # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <16 x i8>,<16 x i8> *%i,align 1
  %x   = sext <16 x i8> %a to <16 x i16>
  %ret = select <16 x i1> %mask, <16 x i16> %x, <16 x i16> zeroinitializer
  ret <16 x i16> %ret
}

define <16 x i16> @zext_16x8_to_16x16(<16 x i8> %a ) nounwind readnone {
; GENERIC-LABEL: zext_16x8_to_16x16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero sched: [1:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_16x8_to_16x16:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = zext <16 x i8> %a to <16 x i16>
  ret <16 x i16> %x
}

define <16 x i16> @zext_16x8_to_16x16_mask(<16 x i8> %a ,<16 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_16x8_to_16x16_mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm1, %k1
; GENERIC-NEXT:    vpmovzxbw {{.*#+}} ymm0 {%k1} {z} = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_16x8_to_16x16_mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm1, %xmm1 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm1, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxbw {{.*#+}} ymm0 {%k1} {z} = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = zext <16 x i8> %a to <16 x i16>
  %ret = select <16 x i1> %mask, <16 x i16> %x, <16 x i16> zeroinitializer
  ret <16 x i16> %ret
}

define <16 x i16> @sext_16x8_to_16x16(<16 x i8> %a ) nounwind readnone {
; GENERIC-LABEL: sext_16x8_to_16x16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxbw %xmm0, %ymm0 # sched: [1:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_16x8_to_16x16:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxbw %xmm0, %ymm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = sext <16 x i8> %a to <16 x i16>
  ret <16 x i16> %x
}

define <16 x i16> @sext_16x8_to_16x16_mask(<16 x i8> %a ,<16 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_16x8_to_16x16_mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm1, %k1
; GENERIC-NEXT:    vpmovsxbw %xmm0, %ymm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_16x8_to_16x16_mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm1, %xmm1 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm1, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovsxbw %xmm0, %ymm0 {%k1} {z} # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = sext <16 x i8> %a to <16 x i16>
  %ret = select <16 x i1> %mask, <16 x i16> %x, <16 x i16> zeroinitializer
  ret <16 x i16> %ret
}

define <32 x i16> @zext_32x8mem_to_32x16(<32 x i8> *%i , <32 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_32x8mem_to_32x16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %ymm0, %ymm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %ymm0, %k1
; GENERIC-NEXT:    vpmovzxbw {{.*#+}} zmm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero,mem[16],zero,mem[17],zero,mem[18],zero,mem[19],zero,mem[20],zero,mem[21],zero,mem[22],zero,mem[23],zero,mem[24],zero,mem[25],zero,mem[26],zero,mem[27],zero,mem[28],zero,mem[29],zero,mem[30],zero,mem[31],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_32x8mem_to_32x16:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %ymm0, %ymm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %ymm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxbw {{.*#+}} zmm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero,mem[16],zero,mem[17],zero,mem[18],zero,mem[19],zero,mem[20],zero,mem[21],zero,mem[22],zero,mem[23],zero,mem[24],zero,mem[25],zero,mem[26],zero,mem[27],zero,mem[28],zero,mem[29],zero,mem[30],zero,mem[31],zero sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <32 x i8>,<32 x i8> *%i,align 1
  %x   = zext <32 x i8> %a to <32 x i16>
  %ret = select <32 x i1> %mask, <32 x i16> %x, <32 x i16> zeroinitializer
  ret <32 x i16> %ret
}

define <32 x i16> @sext_32x8mem_to_32x16(<32 x i8> *%i , <32 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_32x8mem_to_32x16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %ymm0, %ymm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %ymm0, %k1
; GENERIC-NEXT:    vpmovsxbw (%rdi), %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_32x8mem_to_32x16:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %ymm0, %ymm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %ymm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovsxbw (%rdi), %zmm0 {%k1} {z} # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <32 x i8>,<32 x i8> *%i,align 1
  %x   = sext <32 x i8> %a to <32 x i16>
  %ret = select <32 x i1> %mask, <32 x i16> %x, <32 x i16> zeroinitializer
  ret <32 x i16> %ret
}

define <32 x i16> @zext_32x8_to_32x16(<32 x i8> %a ) nounwind readnone {
; GENERIC-LABEL: zext_32x8_to_32x16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxbw {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero,ymm0[16],zero,ymm0[17],zero,ymm0[18],zero,ymm0[19],zero,ymm0[20],zero,ymm0[21],zero,ymm0[22],zero,ymm0[23],zero,ymm0[24],zero,ymm0[25],zero,ymm0[26],zero,ymm0[27],zero,ymm0[28],zero,ymm0[29],zero,ymm0[30],zero,ymm0[31],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_32x8_to_32x16:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxbw {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero,ymm0[16],zero,ymm0[17],zero,ymm0[18],zero,ymm0[19],zero,ymm0[20],zero,ymm0[21],zero,ymm0[22],zero,ymm0[23],zero,ymm0[24],zero,ymm0[25],zero,ymm0[26],zero,ymm0[27],zero,ymm0[28],zero,ymm0[29],zero,ymm0[30],zero,ymm0[31],zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = zext <32 x i8> %a to <32 x i16>
  ret <32 x i16> %x
}

define <32 x i16> @zext_32x8_to_32x16_mask(<32 x i8> %a ,<32 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_32x8_to_32x16_mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %ymm1, %ymm1 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %ymm1, %k1
; GENERIC-NEXT:    vpmovzxbw {{.*#+}} zmm0 {%k1} {z} = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero,ymm0[16],zero,ymm0[17],zero,ymm0[18],zero,ymm0[19],zero,ymm0[20],zero,ymm0[21],zero,ymm0[22],zero,ymm0[23],zero,ymm0[24],zero,ymm0[25],zero,ymm0[26],zero,ymm0[27],zero,ymm0[28],zero,ymm0[29],zero,ymm0[30],zero,ymm0[31],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_32x8_to_32x16_mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %ymm1, %ymm1 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %ymm1, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxbw {{.*#+}} zmm0 {%k1} {z} = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero,ymm0[16],zero,ymm0[17],zero,ymm0[18],zero,ymm0[19],zero,ymm0[20],zero,ymm0[21],zero,ymm0[22],zero,ymm0[23],zero,ymm0[24],zero,ymm0[25],zero,ymm0[26],zero,ymm0[27],zero,ymm0[28],zero,ymm0[29],zero,ymm0[30],zero,ymm0[31],zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = zext <32 x i8> %a to <32 x i16>
  %ret = select <32 x i1> %mask, <32 x i16> %x, <32 x i16> zeroinitializer
  ret <32 x i16> %ret
}

define <32 x i16> @sext_32x8_to_32x16(<32 x i8> %a ) nounwind readnone {
; GENERIC-LABEL: sext_32x8_to_32x16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxbw %ymm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_32x8_to_32x16:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxbw %ymm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = sext <32 x i8> %a to <32 x i16>
  ret <32 x i16> %x
}

define <32 x i16> @sext_32x8_to_32x16_mask(<32 x i8> %a ,<32 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_32x8_to_32x16_mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %ymm1, %ymm1 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %ymm1, %k1
; GENERIC-NEXT:    vpmovsxbw %ymm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_32x8_to_32x16_mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %ymm1, %ymm1 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %ymm1, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovsxbw %ymm0, %zmm0 {%k1} {z} # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = sext <32 x i8> %a to <32 x i16>
  %ret = select <32 x i1> %mask, <32 x i16> %x, <32 x i16> zeroinitializer
  ret <32 x i16> %ret
}

define <4 x i32> @zext_4x8mem_to_4x32(<4 x i8> *%i , <4 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_4x8mem_to_4x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovzxbd {{.*#+}} xmm0 {%k1} {z} = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_4x8mem_to_4x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovzxbd {{.*#+}} xmm0 {%k1} {z} = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <4 x i8>,<4 x i8> *%i,align 1
  %x   = zext <4 x i8> %a to <4 x i32>
  %ret = select <4 x i1> %mask, <4 x i32> %x, <4 x i32> zeroinitializer
  ret <4 x i32> %ret
}

define <4 x i32> @sext_4x8mem_to_4x32(<4 x i8> *%i , <4 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_4x8mem_to_4x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovsxbd (%rdi), %xmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_4x8mem_to_4x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovsxbd (%rdi), %xmm0 {%k1} {z} # sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <4 x i8>,<4 x i8> *%i,align 1
  %x   = sext <4 x i8> %a to <4 x i32>
  %ret = select <4 x i1> %mask, <4 x i32> %x, <4 x i32> zeroinitializer
  ret <4 x i32> %ret
}

define <8 x i32> @zext_8x8mem_to_8x32(<8 x i8> *%i , <8 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_8x8mem_to_8x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k1
; GENERIC-NEXT:    vpmovzxbd {{.*#+}} ymm0 {%k1} {z} = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_8x8mem_to_8x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxbd {{.*#+}} ymm0 {%k1} {z} = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i8>,<8 x i8> *%i,align 1
  %x   = zext <8 x i8> %a to <8 x i32>
  %ret = select <8 x i1> %mask, <8 x i32> %x, <8 x i32> zeroinitializer
  ret <8 x i32> %ret
}

define <8 x i32> @sext_8x8mem_to_8x32(<8 x i8> *%i , <8 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_8x8mem_to_8x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k1
; GENERIC-NEXT:    vpmovsxbd (%rdi), %ymm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_8x8mem_to_8x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovsxbd (%rdi), %ymm0 {%k1} {z} # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i8>,<8 x i8> *%i,align 1
  %x   = sext <8 x i8> %a to <8 x i32>
  %ret = select <8 x i1> %mask, <8 x i32> %x, <8 x i32> zeroinitializer
  ret <8 x i32> %ret
}

define <16 x i32> @zext_16x8mem_to_16x32(<16 x i8> *%i , <16 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_16x8mem_to_16x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm0, %k1
; GENERIC-NEXT:    vpmovzxbd {{.*#+}} zmm0 {%k1} {z} = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_16x8mem_to_16x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxbd {{.*#+}} zmm0 {%k1} {z} = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero,mem[8],zero,zero,zero,mem[9],zero,zero,zero,mem[10],zero,zero,zero,mem[11],zero,zero,zero,mem[12],zero,zero,zero,mem[13],zero,zero,zero,mem[14],zero,zero,zero,mem[15],zero,zero,zero sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <16 x i8>,<16 x i8> *%i,align 1
  %x   = zext <16 x i8> %a to <16 x i32>
  %ret = select <16 x i1> %mask, <16 x i32> %x, <16 x i32> zeroinitializer
  ret <16 x i32> %ret
}

define <16 x i32> @sext_16x8mem_to_16x32(<16 x i8> *%i , <16 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_16x8mem_to_16x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm0, %k1
; GENERIC-NEXT:    vpmovsxbd (%rdi), %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_16x8mem_to_16x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovsxbd (%rdi), %zmm0 {%k1} {z} # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <16 x i8>,<16 x i8> *%i,align 1
  %x   = sext <16 x i8> %a to <16 x i32>
  %ret = select <16 x i1> %mask, <16 x i32> %x, <16 x i32> zeroinitializer
  ret <16 x i32> %ret
}

define <16 x i32> @zext_16x8_to_16x32_mask(<16 x i8> %a , <16 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_16x8_to_16x32_mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm1, %k1
; GENERIC-NEXT:    vpmovzxbd {{.*#+}} zmm0 {%k1} {z} = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_16x8_to_16x32_mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm1, %xmm1 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm1, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxbd {{.*#+}} zmm0 {%k1} {z} = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = zext <16 x i8> %a to <16 x i32>
  %ret = select <16 x i1> %mask, <16 x i32> %x, <16 x i32> zeroinitializer
  ret <16 x i32> %ret
}

define <16 x i32> @sext_16x8_to_16x32_mask(<16 x i8> %a , <16 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_16x8_to_16x32_mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm1, %k1
; GENERIC-NEXT:    vpmovsxbd %xmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_16x8_to_16x32_mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm1, %xmm1 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm1, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovsxbd %xmm0, %zmm0 {%k1} {z} # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = sext <16 x i8> %a to <16 x i32>
  %ret = select <16 x i1> %mask, <16 x i32> %x, <16 x i32> zeroinitializer
  ret <16 x i32> %ret
}

define <16 x i32> @zext_16x8_to_16x32(<16 x i8> %i) nounwind readnone {
; GENERIC-LABEL: zext_16x8_to_16x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxbd {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_16x8_to_16x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxbd {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = zext <16 x i8> %i to <16 x i32>
  ret <16 x i32> %x
}

define <16 x i32> @sext_16x8_to_16x32(<16 x i8> %i) nounwind readnone {
; GENERIC-LABEL: sext_16x8_to_16x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxbd %xmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_16x8_to_16x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxbd %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = sext <16 x i8> %i to <16 x i32>
  ret <16 x i32> %x
}

define <2 x i64> @zext_2x8mem_to_2x64(<2 x i8> *%i , <2 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_2x8mem_to_2x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmq %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovzxbq {{.*#+}} xmm0 {%k1} {z} = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_2x8mem_to_2x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmq %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovzxbq {{.*#+}} xmm0 {%k1} {z} = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <2 x i8>,<2 x i8> *%i,align 1
  %x   = zext <2 x i8> %a to <2 x i64>
  %ret = select <2 x  i1> %mask, <2 x i64> %x, <2 x i64> zeroinitializer
  ret <2 x i64> %ret
}
define <2 x i64> @sext_2x8mem_to_2x64mask(<2 x i8> *%i , <2 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_2x8mem_to_2x64mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmq %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovsxbq (%rdi), %xmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_2x8mem_to_2x64mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmq %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovsxbq (%rdi), %xmm0 {%k1} {z} # sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <2 x i8>,<2 x i8> *%i,align 1
  %x   = sext <2 x i8> %a to <2 x i64>
  %ret = select <2 x i1> %mask, <2 x i64> %x, <2 x i64> zeroinitializer
  ret <2 x i64> %ret
}
define <2 x i64> @sext_2x8mem_to_2x64(<2 x i8> *%i) nounwind readnone {
; GENERIC-LABEL: sext_2x8mem_to_2x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxbq (%rdi), %xmm0 # sched: [7:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_2x8mem_to_2x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxbq (%rdi), %xmm0 # sched: [6:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <2 x i8>,<2 x i8> *%i,align 1
  %x   = sext <2 x i8> %a to <2 x i64>
  ret <2 x i64> %x
}

define <4 x i64> @zext_4x8mem_to_4x64(<4 x i8> *%i , <4 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_4x8mem_to_4x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovzxbq {{.*#+}} ymm0 {%k1} {z} = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero,mem[2],zero,zero,zero,zero,zero,zero,zero,mem[3],zero,zero,zero,zero,zero,zero,zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_4x8mem_to_4x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovzxbq {{.*#+}} ymm0 {%k1} {z} = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero,mem[2],zero,zero,zero,zero,zero,zero,zero,mem[3],zero,zero,zero,zero,zero,zero,zero sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <4 x i8>,<4 x i8> *%i,align 1
  %x   = zext <4 x i8> %a to <4 x i64>
  %ret = select <4 x  i1> %mask, <4 x i64> %x, <4 x i64> zeroinitializer
  ret <4 x i64> %ret
}

define <4 x i64> @sext_4x8mem_to_4x64mask(<4 x i8> *%i , <4 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_4x8mem_to_4x64mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovsxbq (%rdi), %ymm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_4x8mem_to_4x64mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovsxbq (%rdi), %ymm0 {%k1} {z} # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <4 x i8>,<4 x i8> *%i,align 1
  %x   = sext <4 x i8> %a to <4 x i64>
  %ret = select <4 x i1> %mask, <4 x i64> %x, <4 x i64> zeroinitializer
  ret <4 x i64> %ret
}

define <4 x i64> @sext_4x8mem_to_4x64(<4 x i8> *%i) nounwind readnone {
; GENERIC-LABEL: sext_4x8mem_to_4x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxbq (%rdi), %ymm0 # sched: [5:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_4x8mem_to_4x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxbq (%rdi), %ymm0 # sched: [8:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <4 x i8>,<4 x i8> *%i,align 1
  %x   = sext <4 x i8> %a to <4 x i64>
  ret <4 x i64> %x
}

define <8 x i64> @zext_8x8mem_to_8x64(<8 x i8> *%i , <8 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_8x8mem_to_8x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k1
; GENERIC-NEXT:    vpmovzxbq {{.*#+}} zmm0 {%k1} {z} = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero,mem[2],zero,zero,zero,zero,zero,zero,zero,mem[3],zero,zero,zero,zero,zero,zero,zero,mem[4],zero,zero,zero,zero,zero,zero,zero,mem[5],zero,zero,zero,zero,zero,zero,zero,mem[6],zero,zero,zero,zero,zero,zero,zero,mem[7],zero,zero,zero,zero,zero,zero,zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_8x8mem_to_8x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxbq {{.*#+}} zmm0 {%k1} {z} = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero,mem[2],zero,zero,zero,zero,zero,zero,zero,mem[3],zero,zero,zero,zero,zero,zero,zero,mem[4],zero,zero,zero,zero,zero,zero,zero,mem[5],zero,zero,zero,zero,zero,zero,zero,mem[6],zero,zero,zero,zero,zero,zero,zero,mem[7],zero,zero,zero,zero,zero,zero,zero sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i8>,<8 x i8> *%i,align 1
  %x   = zext <8 x i8> %a to <8 x i64>
  %ret = select <8 x  i1> %mask, <8 x i64> %x, <8 x i64> zeroinitializer
  ret <8 x i64> %ret
}

define <8 x i64> @sext_8x8mem_to_8x64mask(<8 x i8> *%i , <8 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_8x8mem_to_8x64mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k1
; GENERIC-NEXT:    vpmovsxbq (%rdi), %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_8x8mem_to_8x64mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovsxbq (%rdi), %zmm0 {%k1} {z} # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i8>,<8 x i8> *%i,align 1
  %x   = sext <8 x i8> %a to <8 x i64>
  %ret = select <8 x i1> %mask, <8 x i64> %x, <8 x i64> zeroinitializer
  ret <8 x i64> %ret
}

define <8 x i64> @sext_8x8mem_to_8x64(<8 x i8> *%i) nounwind readnone {
; GENERIC-LABEL: sext_8x8mem_to_8x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxbq (%rdi), %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_8x8mem_to_8x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxbq (%rdi), %zmm0 # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i8>,<8 x i8> *%i,align 1
  %x   = sext <8 x i8> %a to <8 x i64>
  ret <8 x i64> %x
}

define <4 x i32> @zext_4x16mem_to_4x32(<4 x i16> *%i , <4 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_4x16mem_to_4x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovzxwd {{.*#+}} xmm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_4x16mem_to_4x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovzxwd {{.*#+}} xmm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <4 x i16>,<4 x i16> *%i,align 1
  %x   = zext <4 x i16> %a to <4 x i32>
  %ret = select <4 x i1> %mask, <4 x i32> %x, <4 x i32> zeroinitializer
  ret <4 x i32> %ret
}

define <4 x i32> @sext_4x16mem_to_4x32mask(<4 x i16> *%i , <4 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_4x16mem_to_4x32mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovsxwd (%rdi), %xmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_4x16mem_to_4x32mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovsxwd (%rdi), %xmm0 {%k1} {z} # sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <4 x i16>,<4 x i16> *%i,align 1
  %x   = sext <4 x i16> %a to <4 x i32>
  %ret = select <4 x i1> %mask, <4 x i32> %x, <4 x i32> zeroinitializer
  ret <4 x i32> %ret
}

define <4 x i32> @sext_4x16mem_to_4x32(<4 x i16> *%i) nounwind readnone {
; GENERIC-LABEL: sext_4x16mem_to_4x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxwd (%rdi), %xmm0 # sched: [7:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_4x16mem_to_4x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxwd (%rdi), %xmm0 # sched: [6:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <4 x i16>,<4 x i16> *%i,align 1
  %x   = sext <4 x i16> %a to <4 x i32>
  ret <4 x i32> %x
}


define <8 x i32> @zext_8x16mem_to_8x32(<8 x i16> *%i , <8 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_8x16mem_to_8x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k1
; GENERIC-NEXT:    vpmovzxwd {{.*#+}} ymm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_8x16mem_to_8x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxwd {{.*#+}} ymm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i16>,<8 x i16> *%i,align 1
  %x   = zext <8 x i16> %a to <8 x i32>
  %ret = select <8 x i1> %mask, <8 x i32> %x, <8 x i32> zeroinitializer
  ret <8 x i32> %ret
}

define <8 x i32> @sext_8x16mem_to_8x32mask(<8 x i16> *%i , <8 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_8x16mem_to_8x32mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k1
; GENERIC-NEXT:    vpmovsxwd (%rdi), %ymm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_8x16mem_to_8x32mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovsxwd (%rdi), %ymm0 {%k1} {z} # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i16>,<8 x i16> *%i,align 1
  %x   = sext <8 x i16> %a to <8 x i32>
  %ret = select <8 x i1> %mask, <8 x i32> %x, <8 x i32> zeroinitializer
  ret <8 x i32> %ret
}

define <8 x i32> @sext_8x16mem_to_8x32(<8 x i16> *%i) nounwind readnone {
; GENERIC-LABEL: sext_8x16mem_to_8x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxwd (%rdi), %ymm0 # sched: [5:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_8x16mem_to_8x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxwd (%rdi), %ymm0 # sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i16>,<8 x i16> *%i,align 1
  %x   = sext <8 x i16> %a to <8 x i32>
  ret <8 x i32> %x
}

define <8 x i32> @zext_8x16_to_8x32mask(<8 x i16> %a , <8 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_8x16_to_8x32mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm1, %k1
; GENERIC-NEXT:    vpmovzxwd {{.*#+}} ymm0 {%k1} {z} = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_8x16_to_8x32mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm1, %xmm1 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm1, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxwd {{.*#+}} ymm0 {%k1} {z} = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = zext <8 x i16> %a to <8 x i32>
  %ret = select <8 x i1> %mask, <8 x i32> %x, <8 x i32> zeroinitializer
  ret <8 x i32> %ret
}

define <8 x i32> @zext_8x16_to_8x32(<8 x i16> %a ) nounwind readnone {
; GENERIC-LABEL: zext_8x16_to_8x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero sched: [1:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_8x16_to_8x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = zext <8 x i16> %a to <8 x i32>
  ret <8 x i32> %x
}

define <16 x i32> @zext_16x16mem_to_16x32(<16 x i16> *%i , <16 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_16x16mem_to_16x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm0, %k1
; GENERIC-NEXT:    vpmovzxwd {{.*#+}} zmm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_16x16mem_to_16x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxwd {{.*#+}} zmm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero,mem[8],zero,mem[9],zero,mem[10],zero,mem[11],zero,mem[12],zero,mem[13],zero,mem[14],zero,mem[15],zero sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <16 x i16>,<16 x i16> *%i,align 1
  %x   = zext <16 x i16> %a to <16 x i32>
  %ret = select <16 x i1> %mask, <16 x i32> %x, <16 x i32> zeroinitializer
  ret <16 x i32> %ret
}

define <16 x i32> @sext_16x16mem_to_16x32mask(<16 x i16> *%i , <16 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_16x16mem_to_16x32mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm0, %k1
; GENERIC-NEXT:    vpmovsxwd (%rdi), %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_16x16mem_to_16x32mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovsxwd (%rdi), %zmm0 {%k1} {z} # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <16 x i16>,<16 x i16> *%i,align 1
  %x   = sext <16 x i16> %a to <16 x i32>
  %ret = select <16 x i1> %mask, <16 x i32> %x, <16 x i32> zeroinitializer
  ret <16 x i32> %ret
}

define <16 x i32> @sext_16x16mem_to_16x32(<16 x i16> *%i) nounwind readnone {
; GENERIC-LABEL: sext_16x16mem_to_16x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxwd (%rdi), %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_16x16mem_to_16x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxwd (%rdi), %zmm0 # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <16 x i16>,<16 x i16> *%i,align 1
  %x   = sext <16 x i16> %a to <16 x i32>
  ret <16 x i32> %x
}
define <16 x i32> @zext_16x16_to_16x32mask(<16 x i16> %a , <16 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_16x16_to_16x32mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm1, %k1
; GENERIC-NEXT:    vpmovzxwd {{.*#+}} zmm0 {%k1} {z} = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_16x16_to_16x32mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm1, %xmm1 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm1, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxwd {{.*#+}} zmm0 {%k1} {z} = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = zext <16 x i16> %a to <16 x i32>
  %ret = select <16 x i1> %mask, <16 x i32> %x, <16 x i32> zeroinitializer
  ret <16 x i32> %ret
}

define <16 x i32> @zext_16x16_to_16x32(<16 x i16> %a ) nounwind readnone {
; GENERIC-LABEL: zext_16x16_to_16x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxwd {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_16x16_to_16x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxwd {{.*#+}} zmm0 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = zext <16 x i16> %a to <16 x i32>
  ret <16 x i32> %x
}

define <2 x i64> @zext_2x16mem_to_2x64(<2 x i16> *%i , <2 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_2x16mem_to_2x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmq %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovzxwq {{.*#+}} xmm0 {%k1} {z} = mem[0],zero,zero,zero,mem[1],zero,zero,zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_2x16mem_to_2x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmq %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovzxwq {{.*#+}} xmm0 {%k1} {z} = mem[0],zero,zero,zero,mem[1],zero,zero,zero sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <2 x i16>,<2 x i16> *%i,align 1
  %x   = zext <2 x i16> %a to <2 x i64>
  %ret = select <2 x  i1> %mask, <2 x i64> %x, <2 x i64> zeroinitializer
  ret <2 x i64> %ret
}

define <2 x i64> @sext_2x16mem_to_2x64mask(<2 x i16> *%i , <2 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_2x16mem_to_2x64mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmq %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovsxwq (%rdi), %xmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_2x16mem_to_2x64mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmq %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovsxwq (%rdi), %xmm0 {%k1} {z} # sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <2 x i16>,<2 x i16> *%i,align 1
  %x   = sext <2 x i16> %a to <2 x i64>
  %ret = select <2 x i1> %mask, <2 x i64> %x, <2 x i64> zeroinitializer
  ret <2 x i64> %ret
}

define <2 x i64> @sext_2x16mem_to_2x64(<2 x i16> *%i) nounwind readnone {
; GENERIC-LABEL: sext_2x16mem_to_2x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxwq (%rdi), %xmm0 # sched: [7:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_2x16mem_to_2x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxwq (%rdi), %xmm0 # sched: [6:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <2 x i16>,<2 x i16> *%i,align 1
  %x   = sext <2 x i16> %a to <2 x i64>
  ret <2 x i64> %x
}

define <4 x i64> @zext_4x16mem_to_4x64(<4 x i16> *%i , <4 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_4x16mem_to_4x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovzxwq {{.*#+}} ymm0 {%k1} {z} = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_4x16mem_to_4x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovzxwq {{.*#+}} ymm0 {%k1} {z} = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <4 x i16>,<4 x i16> *%i,align 1
  %x   = zext <4 x i16> %a to <4 x i64>
  %ret = select <4 x  i1> %mask, <4 x i64> %x, <4 x i64> zeroinitializer
  ret <4 x i64> %ret
}

define <4 x i64> @sext_4x16mem_to_4x64mask(<4 x i16> *%i , <4 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_4x16mem_to_4x64mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovsxwq (%rdi), %ymm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_4x16mem_to_4x64mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovsxwq (%rdi), %ymm0 {%k1} {z} # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <4 x i16>,<4 x i16> *%i,align 1
  %x   = sext <4 x i16> %a to <4 x i64>
  %ret = select <4 x i1> %mask, <4 x i64> %x, <4 x i64> zeroinitializer
  ret <4 x i64> %ret
}

define <4 x i64> @sext_4x16mem_to_4x64(<4 x i16> *%i) nounwind readnone {
; GENERIC-LABEL: sext_4x16mem_to_4x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxwq (%rdi), %ymm0 # sched: [5:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_4x16mem_to_4x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxwq (%rdi), %ymm0 # sched: [8:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <4 x i16>,<4 x i16> *%i,align 1
  %x   = sext <4 x i16> %a to <4 x i64>
  ret <4 x i64> %x
}

define <8 x i64> @zext_8x16mem_to_8x64(<8 x i16> *%i , <8 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_8x16mem_to_8x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k1
; GENERIC-NEXT:    vpmovzxwq {{.*#+}} zmm0 {%k1} {z} = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_8x16mem_to_8x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxwq {{.*#+}} zmm0 {%k1} {z} = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero,mem[4],zero,zero,zero,mem[5],zero,zero,zero,mem[6],zero,zero,zero,mem[7],zero,zero,zero sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i16>,<8 x i16> *%i,align 1
  %x   = zext <8 x i16> %a to <8 x i64>
  %ret = select <8 x  i1> %mask, <8 x i64> %x, <8 x i64> zeroinitializer
  ret <8 x i64> %ret
}

define <8 x i64> @sext_8x16mem_to_8x64mask(<8 x i16> *%i , <8 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_8x16mem_to_8x64mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k1
; GENERIC-NEXT:    vpmovsxwq (%rdi), %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_8x16mem_to_8x64mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovsxwq (%rdi), %zmm0 {%k1} {z} # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i16>,<8 x i16> *%i,align 1
  %x   = sext <8 x i16> %a to <8 x i64>
  %ret = select <8 x i1> %mask, <8 x i64> %x, <8 x i64> zeroinitializer
  ret <8 x i64> %ret
}

define <8 x i64> @sext_8x16mem_to_8x64(<8 x i16> *%i) nounwind readnone {
; GENERIC-LABEL: sext_8x16mem_to_8x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxwq (%rdi), %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_8x16mem_to_8x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxwq (%rdi), %zmm0 # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i16>,<8 x i16> *%i,align 1
  %x   = sext <8 x i16> %a to <8 x i64>
  ret <8 x i64> %x
}

define <8 x i64> @zext_8x16_to_8x64mask(<8 x i16> %a , <8 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_8x16_to_8x64mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm1, %k1
; GENERIC-NEXT:    vpmovzxwq {{.*#+}} zmm0 {%k1} {z} = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_8x16_to_8x64mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm1, %xmm1 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm1, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxwq {{.*#+}} zmm0 {%k1} {z} = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = zext <8 x i16> %a to <8 x i64>
  %ret = select <8 x  i1> %mask, <8 x i64> %x, <8 x i64> zeroinitializer
  ret <8 x i64> %ret
}

define <8 x i64> @zext_8x16_to_8x64(<8 x i16> %a) nounwind readnone {
; GENERIC-LABEL: zext_8x16_to_8x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxwq {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_8x16_to_8x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxwq {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %ret   = zext <8 x i16> %a to <8 x i64>
  ret <8 x i64> %ret
}

define <2 x i64> @zext_2x32mem_to_2x64(<2 x i32> *%i , <2 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_2x32mem_to_2x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmq %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovzxdq {{.*#+}} xmm0 {%k1} {z} = mem[0],zero,mem[1],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_2x32mem_to_2x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmq %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovzxdq {{.*#+}} xmm0 {%k1} {z} = mem[0],zero,mem[1],zero sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <2 x i32>,<2 x i32> *%i,align 1
  %x   = zext <2 x i32> %a to <2 x i64>
  %ret = select <2 x  i1> %mask, <2 x i64> %x, <2 x i64> zeroinitializer
  ret <2 x i64> %ret
}

define <2 x i64> @sext_2x32mem_to_2x64mask(<2 x i32> *%i , <2 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_2x32mem_to_2x64mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmq %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovsxdq (%rdi), %xmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_2x32mem_to_2x64mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmq %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovsxdq (%rdi), %xmm0 {%k1} {z} # sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <2 x i32>,<2 x i32> *%i,align 1
  %x   = sext <2 x i32> %a to <2 x i64>
  %ret = select <2 x i1> %mask, <2 x i64> %x, <2 x i64> zeroinitializer
  ret <2 x i64> %ret
}

define <2 x i64> @sext_2x32mem_to_2x64(<2 x i32> *%i) nounwind readnone {
; GENERIC-LABEL: sext_2x32mem_to_2x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxdq (%rdi), %xmm0 # sched: [7:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_2x32mem_to_2x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxdq (%rdi), %xmm0 # sched: [6:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <2 x i32>,<2 x i32> *%i,align 1
  %x   = sext <2 x i32> %a to <2 x i64>
  ret <2 x i64> %x
}

define <4 x i64> @zext_4x32mem_to_4x64(<4 x i32> *%i , <4 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_4x32mem_to_4x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovzxdq {{.*#+}} ymm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_4x32mem_to_4x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovzxdq {{.*#+}} ymm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <4 x i32>,<4 x i32> *%i,align 1
  %x   = zext <4 x i32> %a to <4 x i64>
  %ret = select <4 x  i1> %mask, <4 x i64> %x, <4 x i64> zeroinitializer
  ret <4 x i64> %ret
}

define <4 x i64> @sext_4x32mem_to_4x64mask(<4 x i32> *%i , <4 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_4x32mem_to_4x64mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpmovsxdq (%rdi), %ymm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_4x32mem_to_4x64mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovsxdq (%rdi), %ymm0 {%k1} {z} # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <4 x i32>,<4 x i32> *%i,align 1
  %x   = sext <4 x i32> %a to <4 x i64>
  %ret = select <4 x i1> %mask, <4 x i64> %x, <4 x i64> zeroinitializer
  ret <4 x i64> %ret
}

define <4 x i64> @sext_4x32mem_to_4x64(<4 x i32> *%i) nounwind readnone {
; GENERIC-LABEL: sext_4x32mem_to_4x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxdq (%rdi), %ymm0 # sched: [5:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_4x32mem_to_4x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxdq (%rdi), %ymm0 # sched: [9:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <4 x i32>,<4 x i32> *%i,align 1
  %x   = sext <4 x i32> %a to <4 x i64>
  ret <4 x i64> %x
}

define <4 x i64> @sext_4x32_to_4x64(<4 x i32> %a) nounwind readnone {
; GENERIC-LABEL: sext_4x32_to_4x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxdq %xmm0, %ymm0 # sched: [1:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_4x32_to_4x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxdq %xmm0, %ymm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = sext <4 x i32> %a to <4 x i64>
  ret <4 x i64> %x
}

define <4 x i64> @zext_4x32_to_4x64mask(<4 x i32> %a , <4 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_4x32_to_4x64mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm1, %xmm1, %k1
; GENERIC-NEXT:    vpmovzxdq {{.*#+}} ymm0 {%k1} {z} = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_4x32_to_4x64mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %xmm1, %xmm1 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm1, %xmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpmovzxdq {{.*#+}} ymm0 {%k1} {z} = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = zext <4 x i32> %a to <4 x i64>
  %ret = select <4 x  i1> %mask, <4 x i64> %x, <4 x i64> zeroinitializer
  ret <4 x i64> %ret
}

define <8 x i64> @zext_8x32mem_to_8x64(<8 x i32> *%i , <8 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_8x32mem_to_8x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k1
; GENERIC-NEXT:    vpmovzxdq {{.*#+}} zmm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_8x32mem_to_8x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxdq {{.*#+}} zmm0 {%k1} {z} = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i32>,<8 x i32> *%i,align 1
  %x   = zext <8 x i32> %a to <8 x i64>
  %ret = select <8 x  i1> %mask, <8 x i64> %x, <8 x i64> zeroinitializer
  ret <8 x i64> %ret
}

define <8 x i64> @sext_8x32mem_to_8x64mask(<8 x i32> *%i , <8 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: sext_8x32mem_to_8x64mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k1
; GENERIC-NEXT:    vpmovsxdq (%rdi), %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_8x32mem_to_8x64mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovsxdq (%rdi), %zmm0 {%k1} {z} # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i32>,<8 x i32> *%i,align 1
  %x   = sext <8 x i32> %a to <8 x i64>
  %ret = select <8 x i1> %mask, <8 x i64> %x, <8 x i64> zeroinitializer
  ret <8 x i64> %ret
}

define <8 x i64> @sext_8x32mem_to_8x64(<8 x i32> *%i) nounwind readnone {
; GENERIC-LABEL: sext_8x32mem_to_8x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxdq (%rdi), %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_8x32mem_to_8x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxdq (%rdi), %zmm0 # sched: [10:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a   = load <8 x i32>,<8 x i32> *%i,align 1
  %x   = sext <8 x i32> %a to <8 x i64>
  ret <8 x i64> %x
}

define <8 x i64> @sext_8x32_to_8x64(<8 x i32> %a) nounwind readnone {
; GENERIC-LABEL: sext_8x32_to_8x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxdq %ymm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_8x32_to_8x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxdq %ymm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = sext <8 x i32> %a to <8 x i64>
  ret <8 x i64> %x
}

define <8 x i64> @zext_8x32_to_8x64mask(<8 x i32> %a , <8 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: zext_8x32_to_8x64mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm1, %k1
; GENERIC-NEXT:    vpmovzxdq {{.*#+}} zmm0 {%k1} {z} = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_8x32_to_8x64mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm1, %xmm1 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm1, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxdq {{.*#+}} zmm0 {%k1} {z} = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = zext <8 x i32> %a to <8 x i64>
  %ret = select <8 x  i1> %mask, <8 x i64> %x, <8 x i64> zeroinitializer
  ret <8 x i64> %ret
}
define <8 x float> @fptrunc_test(<8 x double> %a) nounwind readnone {
; GENERIC-LABEL: fptrunc_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtpd2ps %zmm0, %ymm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: fptrunc_test:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtpd2ps %zmm0, %ymm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = fptrunc <8 x double> %a to <8 x float>
  ret <8 x float> %b
}

define <8 x double> @fpext_test(<8 x float> %a) nounwind readnone {
; GENERIC-LABEL: fpext_test:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vcvtps2pd %ymm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: fpext_test:
; SKX:       # BB#0:
; SKX-NEXT:    vcvtps2pd %ymm0, %zmm0 # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = fpext <8 x float> %a to <8 x double>
  ret <8 x double> %b
}

define   <16 x i32> @zext_16i1_to_16xi32(i16 %b) {
; GENERIC-LABEL: zext_16i1_to_16xi32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vpbroadcastd {{.*}}(%rip), %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_16i1_to_16xi32:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpbroadcastd {{.*}}(%rip), %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = bitcast i16 %b to <16 x i1>
  %c = zext <16 x i1> %a to <16 x i32>
  ret <16 x i32> %c
}

define   <8 x i64> @zext_8i1_to_8xi64(i8 %b) {
; GENERIC-LABEL: zext_8i1_to_8xi64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vpbroadcastq {{.*}}(%rip), %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_8i1_to_8xi64:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpbroadcastq {{.*}}(%rip), %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = bitcast i8 %b to <8 x i1>
  %c = zext <8 x i1> %a to <8 x i64>
  ret <8 x i64> %c
}

define i16 @trunc_16i8_to_16i1(<16 x i8> %a) {
; GENERIC-LABEL: trunc_16i8_to_16i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: trunc_16i8_to_16i1:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask_b = trunc <16 x i8>%a to <16 x i1>
  %mask = bitcast <16 x i1> %mask_b to i16
  ret i16 %mask
}

define i16 @trunc_16i32_to_16i1(<16 x i32> %a) {
; GENERIC-LABEL: trunc_16i32_to_16i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %zmm0, %zmm0
; GENERIC-NEXT:    vptestmd %zmm0, %zmm0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: trunc_16i32_to_16i1:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %zmm0, %zmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %zmm0, %zmm0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask_b = trunc <16 x i32>%a to <16 x i1>
  %mask = bitcast <16 x i1> %mask_b to i16
  ret i16 %mask
}

define <4 x i32> @trunc_4i32_to_4i1(<4 x i32> %a, <4 x i32> %b) {
; GENERIC-LABEL: trunc_4i32_to_4i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm0, %xmm0, %k1
; GENERIC-NEXT:    vpslld $31, %xmm1, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm0, %xmm0, %k0 {%k1}
; GENERIC-NEXT:    vpmovm2d %k0, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: trunc_4i32_to_4i1:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm0, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpslld $31, %xmm1, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm0, %xmm0, %k0 {%k1} # sched: [3:1.00]
; SKX-NEXT:    vpmovm2d %k0, %xmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask_a = trunc <4 x i32>%a to <4 x i1>
  %mask_b = trunc <4 x i32>%b to <4 x i1>
  %a_and_b = and <4 x i1>%mask_a, %mask_b
  %res = sext <4 x i1>%a_and_b to <4 x i32>
  ret <4 x i32>%res
}


define i8 @trunc_8i16_to_8i1(<8 x i16> %a) {
; GENERIC-LABEL: trunc_8i16_to_8i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    # kill: %AL<def> %AL<kill> %EAX<kill>
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: trunc_8i16_to_8i1:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    # kill: %AL<def> %AL<kill> %EAX<kill>
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask_b = trunc <8 x i16>%a to <8 x i1>
  %mask = bitcast <8 x i1> %mask_b to i8
  ret i8 %mask
}

define <8 x i32> @sext_8i1_8i32(<8 x i32> %a1, <8 x i32> %a2) nounwind {
; GENERIC-LABEL: sext_8i1_8i32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpcmpled %ymm0, %ymm1, %k0
; GENERIC-NEXT:    vpmovm2d %k0, %ymm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_8i1_8i32:
; SKX:       # BB#0:
; SKX-NEXT:    vpcmpled %ymm0, %ymm1, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2d %k0, %ymm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = icmp slt <8 x i32> %a1, %a2
  %x1 = xor <8 x i1>%x, <i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>
  %y = sext <8 x i1> %x1 to <8 x i32>
  ret <8 x i32> %y
}


define i16 @trunc_i32_to_i1(i32 %a) {
; GENERIC-LABEL: trunc_i32_to_i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    movw $-4, %ax # sched: [1:0.33]
; GENERIC-NEXT:    kmovd %eax, %k0
; GENERIC-NEXT:    kshiftrw $1, %k0, %k0
; GENERIC-NEXT:    kshiftlw $1, %k0, %k0
; GENERIC-NEXT:    andl $1, %edi # sched: [1:0.33]
; GENERIC-NEXT:    kmovw %edi, %k1
; GENERIC-NEXT:    korw %k1, %k0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: trunc_i32_to_i1:
; SKX:       # BB#0:
; SKX-NEXT:    movw $-4, %ax # sched: [1:0.25]
; SKX-NEXT:    kmovd %eax, %k0 # sched: [1:1.00]
; SKX-NEXT:    kshiftrw $1, %k0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kshiftlw $1, %k0, %k0 # sched: [3:1.00]
; SKX-NEXT:    andl $1, %edi # sched: [1:0.25]
; SKX-NEXT:    kmovw %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    korw %k1, %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; SKX-NEXT:    retq # sched: [7:1.00]
  %a_i = trunc i32 %a to i1
  %maskv = insertelement <16 x i1> <i1 true, i1 false, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true, i1 true>, i1 %a_i, i32 0
  %res = bitcast <16 x i1> %maskv to i16
  ret i16 %res
}

define <8 x i16> @sext_8i1_8i16(<8 x i32> %a1, <8 x i32> %a2) nounwind {
; GENERIC-LABEL: sext_8i1_8i16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpcmpgtd %ymm0, %ymm1, %k0
; GENERIC-NEXT:    vpmovm2w %k0, %xmm0
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_8i1_8i16:
; SKX:       # BB#0:
; SKX-NEXT:    vpcmpgtd %ymm0, %ymm1, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2w %k0, %xmm0
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = icmp slt <8 x i32> %a1, %a2
  %y = sext <8 x i1> %x to <8 x i16>
  ret <8 x i16> %y
}

define <16 x i32> @sext_16i1_16i32(<16 x i32> %a1, <16 x i32> %a2) nounwind {
; GENERIC-LABEL: sext_16i1_16i32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpcmpgtd %zmm0, %zmm1, %k0
; GENERIC-NEXT:    vpmovm2d %k0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_16i1_16i32:
; SKX:       # BB#0:
; SKX-NEXT:    vpcmpgtd %zmm0, %zmm1, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2d %k0, %zmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = icmp slt <16 x i32> %a1, %a2
  %y = sext <16 x i1> %x to <16 x i32>
  ret <16 x i32> %y
}

define <8 x i64> @sext_8i1_8i64(<8 x i32> %a1, <8 x i32> %a2) nounwind {
; GENERIC-LABEL: sext_8i1_8i64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpcmpgtd %ymm0, %ymm1, %k0
; GENERIC-NEXT:    vpmovm2q %k0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: sext_8i1_8i64:
; SKX:       # BB#0:
; SKX-NEXT:    vpcmpgtd %ymm0, %ymm1, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2q %k0, %zmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = icmp slt <8 x i32> %a1, %a2
  %y = sext <8 x i1> %x to <8 x i64>
  ret <8 x i64> %y
}

define void @extload_v8i64(<8 x i8>* %a, <8 x i64>* %res) {
; GENERIC-LABEL: extload_v8i64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovsxbq (%rdi), %zmm0
; GENERIC-NEXT:    vmovdqa64 %zmm0, (%rsi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: extload_v8i64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovsxbq (%rdi), %zmm0 # sched: [10:1.00]
; SKX-NEXT:    vmovdqa64 %zmm0, (%rsi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %sign_load = load <8 x i8>, <8 x i8>* %a
  %c = sext <8 x i8> %sign_load to <8 x i64>
  store <8 x i64> %c, <8 x i64>* %res
  ret void
}

define <64 x i16> @test21(<64 x i16> %x , <64 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: test21:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %zmm2, %zmm2
; GENERIC-NEXT:    vpmovb2m %zmm2, %k1
; GENERIC-NEXT:    vmovdqu16 %zmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    kshiftrq $32, %k1, %k1
; GENERIC-NEXT:    vmovdqu16 %zmm1, %zmm1 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test21:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %zmm2, %zmm2 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %zmm2, %k1 # sched: [1:1.00]
; SKX-NEXT:    vmovdqu16 %zmm0, %zmm0 {%k1} {z}
; SKX-NEXT:    kshiftrq $32, %k1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovdqu16 %zmm1, %zmm1 {%k1} {z}
; SKX-NEXT:    retq # sched: [7:1.00]
  %ret = select <64 x i1> %mask, <64 x i16> %x, <64 x i16> zeroinitializer
  ret <64 x i16> %ret
}

define <16 x i16> @shuffle_zext_16x8_to_16x16(<16 x i8> %a) nounwind readnone {
; GENERIC-LABEL: shuffle_zext_16x8_to_16x16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero sched: [1:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: shuffle_zext_16x8_to_16x16:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %1 = shufflevector <16 x i8> %a, <16 x i8> zeroinitializer, <32 x i32> <i32 0, i32 16, i32 1, i32 16, i32 2, i32 16, i32 3, i32 16, i32 4, i32 16, i32 5, i32 16, i32 6, i32 16, i32 7, i32 16, i32 8, i32 16, i32 9, i32 16, i32 10, i32 16, i32 11, i32 16, i32 12, i32 16, i32 13, i32 16, i32 14, i32 16, i32 15, i32 16>
  %2 = bitcast <32 x i8> %1 to <16 x i16>
  ret <16 x i16> %2
}

define <16 x i16> @shuffle_zext_16x8_to_16x16_mask(<16 x i8> %a, <16 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: shuffle_zext_16x8_to_16x16_mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm1, %k1
; GENERIC-NEXT:    vpmovzxbw {{.*#+}} ymm0 {%k1} {z} = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: shuffle_zext_16x8_to_16x16_mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm1, %xmm1 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm1, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovzxbw {{.*#+}} ymm0 {%k1} {z} = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x   = shufflevector <16 x i8> %a, <16 x i8> zeroinitializer, <32 x i32> <i32 0, i32 16, i32 1, i32 16, i32 2, i32 16, i32 3, i32 16, i32 4, i32 16, i32 5, i32 16, i32 6, i32 16, i32 7, i32 16, i32 8, i32 16, i32 9, i32 16, i32 10, i32 16, i32 11, i32 16, i32 12, i32 16, i32 13, i32 16, i32 14, i32 16, i32 15, i32 16>
  %bc  = bitcast <32 x i8> %x to <16 x i16>
  %ret = select <16 x i1> %mask, <16 x i16> %bc, <16 x i16> zeroinitializer
  ret <16 x i16> %ret
}

define <16 x i16> @zext_32x8_to_16x16(<32 x i8> %a) {
; GENERIC-LABEL: zext_32x8_to_16x16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero sched: [1:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_32x8_to_16x16:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxbw {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %1 = shufflevector <32 x i8> %a, <32 x i8> zeroinitializer, <32 x i32> <i32 0, i32 32, i32 1, i32 32, i32 2, i32 32, i32 3, i32 32, i32 4, i32 32, i32 5, i32 32, i32 6, i32 32, i32 7, i32 32, i32 8, i32 32, i32 9, i32 32, i32 10, i32 32, i32 11, i32 32, i32 12, i32 32, i32 13, i32 32, i32 14, i32 32, i32 15, i32 32>
  %2 = bitcast <32 x i8> %1 to <16 x i16>
  ret <16 x i16> %2
}

define <8 x i32> @zext_32x8_to_8x32(<32 x i8> %a) {
; GENERIC-LABEL: zext_32x8_to_8x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxbd {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero sched: [1:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_32x8_to_8x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxbd {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %1 = shufflevector <32 x i8> %a, <32 x i8> zeroinitializer, <32 x i32> <i32 0, i32 32, i32 32, i32 32, i32 1, i32 32, i32 32, i32 32, i32 2, i32 32, i32 32, i32 32, i32 3, i32 32, i32 32, i32 32, i32 4, i32 32, i32 32, i32 32, i32 5, i32 32, i32 32, i32 32, i32 6, i32 32, i32 32, i32 32, i32 7, i32 32, i32 32, i32 32>
  %2 = bitcast <32 x i8> %1 to <8 x i32>
  ret <8 x i32> %2
}

define <4 x i64> @zext_32x8_to_4x64(<32 x i8> %a) {
; GENERIC-LABEL: zext_32x8_to_4x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxbq {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero,xmm0[2],zero,zero,zero,zero,zero,zero,zero,xmm0[3],zero,zero,zero,zero,zero,zero,zero sched: [1:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_32x8_to_4x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxbq {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero,xmm0[2],zero,zero,zero,zero,zero,zero,zero,xmm0[3],zero,zero,zero,zero,zero,zero,zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %1 = shufflevector <32 x i8> %a, <32 x i8> zeroinitializer, <32 x i32> <i32 0, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 1, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 2, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 3, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32, i32 32>
  %2 = bitcast <32 x i8> %1 to <4 x i64>
  ret <4 x i64> %2
}

define <8 x i32> @zext_16x16_to_8x32(<16 x i16> %a) {
; GENERIC-LABEL: zext_16x16_to_8x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero sched: [1:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_16x16_to_8x32:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %1 = shufflevector <16 x i16> %a, <16 x i16> zeroinitializer, <16 x i32> <i32 0, i32 16, i32 1, i32 16, i32 2, i32 16, i32 3, i32 16, i32 4, i32 16, i32 5, i32 16, i32 6, i32 16, i32 7, i32 16>
  %2 = bitcast <16 x i16> %1 to <8 x i32>
  ret <8 x i32> %2
}

define <4 x i64> @zext_16x16_to_4x64(<16 x i16> %a) {
; GENERIC-LABEL: zext_16x16_to_4x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxwq {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero sched: [1:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_16x16_to_4x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxwq {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %1 = shufflevector <16 x i16> %a, <16 x i16> zeroinitializer, <16 x i32> <i32 0, i32 16, i32 16, i32 16, i32 1, i32 16, i32 16, i32 16, i32 2, i32 16, i32 16, i32 16, i32 3, i32 16, i32 16, i32 16>
  %2 = bitcast <16 x i16> %1 to <4 x i64>
  ret <4 x i64> %2
}

define <4 x i64> @zext_8x32_to_4x64(<8 x i32> %a) {
; GENERIC-LABEL: zext_8x32_to_4x64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpmovzxdq {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero sched: [1:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_8x32_to_4x64:
; SKX:       # BB#0:
; SKX-NEXT:    vpmovzxdq {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %1 = shufflevector <8 x i32> %a, <8 x i32> zeroinitializer, <8 x i32> <i32 0, i32 8, i32 1, i32 8, i32 2, i32 8, i32 3, i32 8>
  %2 = bitcast <8 x i32> %1 to <4 x i64>
  ret <4 x i64> %2
}

define <64 x i8> @zext_64xi1_to_64xi8(<64 x i8> %x, <64 x i8> %y) #0 {
; GENERIC-LABEL: zext_64xi1_to_64xi8:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpcmpeqb %zmm1, %zmm0, %k1
; GENERIC-NEXT:    vmovdqu8 {{.*}}(%rip), %zmm0 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_64xi1_to_64xi8:
; SKX:       # BB#0:
; SKX-NEXT:    vpcmpeqb %zmm1, %zmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovdqu8 {{.*}}(%rip), %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp eq <64 x i8> %x, %y
  %1 = zext <64 x i1> %mask to <64 x i8>
  ret <64 x i8> %1
}

define <32 x i16> @zext_32xi1_to_32xi16(<32 x i16> %x, <32 x i16> %y) #0 {
; GENERIC-LABEL: zext_32xi1_to_32xi16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpcmpeqw %zmm1, %zmm0, %k1
; GENERIC-NEXT:    vmovdqu16 {{.*}}(%rip), %zmm0 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_32xi1_to_32xi16:
; SKX:       # BB#0:
; SKX-NEXT:    vpcmpeqw %zmm1, %zmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovdqu16 {{.*}}(%rip), %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp eq <32 x i16> %x, %y
  %1 = zext <32 x i1> %mask to <32 x i16>
  ret <32 x i16> %1
}

define <16 x i16> @zext_16xi1_to_16xi16(<16 x i16> %x, <16 x i16> %y) #0 {
; GENERIC-LABEL: zext_16xi1_to_16xi16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpcmpeqw %ymm1, %ymm0, %k1
; GENERIC-NEXT:    vmovdqu16 {{.*}}(%rip), %ymm0 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_16xi1_to_16xi16:
; SKX:       # BB#0:
; SKX-NEXT:    vpcmpeqw %ymm1, %ymm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovdqu16 {{.*}}(%rip), %ymm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp eq <16 x i16> %x, %y
  %1 = zext <16 x i1> %mask to <16 x i16>
  ret <16 x i16> %1
}


define <32 x i8> @zext_32xi1_to_32xi8(<32 x i16> %x, <32 x i16> %y) #0 {
; GENERIC-LABEL: zext_32xi1_to_32xi8:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpcmpeqw %zmm1, %zmm0, %k1
; GENERIC-NEXT:    vmovdqu8 {{.*}}(%rip), %ymm0 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_32xi1_to_32xi8:
; SKX:       # BB#0:
; SKX-NEXT:    vpcmpeqw %zmm1, %zmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovdqu8 {{.*}}(%rip), %ymm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp eq <32 x i16> %x, %y
  %1 = zext <32 x i1> %mask to <32 x i8>
  ret <32 x i8> %1
}

define <4 x i32> @zext_4xi1_to_4x32(<4 x i8> %x, <4 x i8> %y) #0 {
; GENERIC-LABEL: zext_4xi1_to_4x32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0] sched: [6:0.50]
; GENERIC-NEXT:    vpand %xmm2, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpand %xmm2, %xmm0, %xmm0 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpeqd %xmm1, %xmm0, %k1
; GENERIC-NEXT:    vpbroadcastd {{.*}}(%rip), %xmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_4xi1_to_4x32:
; SKX:       # BB#0:
; SKX-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0] sched: [6:0.50]
; SKX-NEXT:    vpand %xmm2, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpand %xmm2, %xmm0, %xmm0 # sched: [1:0.33]
; SKX-NEXT:    vpcmpeqd %xmm1, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vpbroadcastd {{.*}}(%rip), %xmm0 {%k1} {z} # sched: [7:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp eq <4 x i8> %x, %y
  %1 = zext <4 x i1> %mask to <4 x i32>
  ret <4 x i32> %1
}

define <2 x i64> @zext_2xi1_to_2xi64(<2 x i8> %x, <2 x i8> %y) #0 {
; GENERIC-LABEL: zext_2xi1_to_2xi64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,0,0,0,0,0,0,0,255,0,0,0,0,0,0,0] sched: [6:0.50]
; GENERIC-NEXT:    vpand %xmm2, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpand %xmm2, %xmm0, %xmm0 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpeqq %xmm1, %xmm0, %k1
; GENERIC-NEXT:    vmovdqa64 {{.*}}(%rip), %xmm0 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_2xi1_to_2xi64:
; SKX:       # BB#0:
; SKX-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,0,0,0,0,0,0,0,255,0,0,0,0,0,0,0] sched: [6:0.50]
; SKX-NEXT:    vpand %xmm2, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpand %xmm2, %xmm0, %xmm0 # sched: [1:0.33]
; SKX-NEXT:    vpcmpeqq %xmm1, %xmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovdqa64 {{.*}}(%rip), %xmm0 {%k1} {z} # sched: [7:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp eq <2 x i8> %x, %y
  %1 = zext <2 x i1> %mask to <2 x i64>
  ret <2 x i64> %1
}

define <16 x float> @test_x86_fmadd_ps_z(<16 x float> %a0, <16 x float> %a1, <16 x float> %a2) {
; GENERIC-LABEL: test_x86_fmadd_ps_z:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmulps %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    vaddps %zmm2, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_x86_fmadd_ps_z:
; SKX:       # BB#0:
; SKX-NEXT:    vmulps %zmm1, %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    vaddps %zmm2, %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = fmul <16 x float> %a0, %a1
  %res = fadd <16 x float> %x, %a2
  ret <16 x float> %res
}

define <16 x float> @test_x86_fmsub_ps_z(<16 x float> %a0, <16 x float> %a1, <16 x float> %a2) {
; GENERIC-LABEL: test_x86_fmsub_ps_z:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmulps %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    vsubps %zmm2, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_x86_fmsub_ps_z:
; SKX:       # BB#0:
; SKX-NEXT:    vmulps %zmm1, %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    vsubps %zmm2, %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = fmul <16 x float> %a0, %a1
  %res = fsub <16 x float> %x, %a2
  ret <16 x float> %res
}

define <16 x float> @test_x86_fnmadd_ps_z(<16 x float> %a0, <16 x float> %a1, <16 x float> %a2) {
; GENERIC-LABEL: test_x86_fnmadd_ps_z:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmulps %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    vsubps %zmm0, %zmm2, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_x86_fnmadd_ps_z:
; SKX:       # BB#0:
; SKX-NEXT:    vmulps %zmm1, %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    vsubps %zmm0, %zmm2, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = fmul <16 x float> %a0, %a1
  %res = fsub <16 x float> %a2, %x
  ret <16 x float> %res
}

define <16 x float> @test_x86_fnmsub_ps_z(<16 x float> %a0, <16 x float> %a1, <16 x float> %a2) {
; GENERIC-LABEL: test_x86_fnmsub_ps_z:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmulps %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    vxorps {{.*}}(%rip){1to16}, %zmm0, %zmm0
; GENERIC-NEXT:    vsubps %zmm2, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_x86_fnmsub_ps_z:
; SKX:       # BB#0:
; SKX-NEXT:    vmulps %zmm1, %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    vxorps {{.*}}(%rip){1to16}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    vsubps %zmm2, %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = fmul <16 x float> %a0, %a1
  %y = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00,
                          float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00,
                          float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00,
                          float -0.000000e+00>, %x
  %res = fsub <16 x float> %y, %a2
  ret <16 x float> %res
}

define <8 x double> @test_x86_fmadd_pd_z(<8 x double> %a0, <8 x double> %a1, <8 x double> %a2) {
; GENERIC-LABEL: test_x86_fmadd_pd_z:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    vaddpd %zmm2, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_x86_fmadd_pd_z:
; SKX:       # BB#0:
; SKX-NEXT:    vmulpd %zmm1, %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    vaddpd %zmm2, %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = fmul <8 x double> %a0, %a1
  %res = fadd <8 x double> %x, %a2
  ret <8 x double> %res
}

define <8 x double> @test_x86_fmsub_pd_z(<8 x double> %a0, <8 x double> %a1, <8 x double> %a2) {
; GENERIC-LABEL: test_x86_fmsub_pd_z:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    vsubpd %zmm2, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_x86_fmsub_pd_z:
; SKX:       # BB#0:
; SKX-NEXT:    vmulpd %zmm1, %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    vsubpd %zmm2, %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = fmul <8 x double> %a0, %a1
  %res = fsub <8 x double> %x, %a2
  ret <8 x double> %res
}

define double @test_x86_fmsub_213(double %a0, double %a1, double %a2) {
; GENERIC-LABEL: test_x86_fmsub_213:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmulsd %xmm1, %xmm0, %xmm0 # sched: [5:1.00]
; GENERIC-NEXT:    vsubsd %xmm2, %xmm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_x86_fmsub_213:
; SKX:       # BB#0:
; SKX-NEXT:    vmulsd %xmm1, %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    vsubsd %xmm2, %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = fmul double %a0, %a1
  %res = fsub double %x, %a2
  ret double %res
}

define double @test_x86_fmsub_213_m(double %a0, double %a1, double * %a2_ptr) {
; GENERIC-LABEL: test_x86_fmsub_213_m:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmulsd %xmm1, %xmm0, %xmm0 # sched: [5:1.00]
; GENERIC-NEXT:    vsubsd (%rdi), %xmm0, %xmm0 # sched: [9:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_x86_fmsub_213_m:
; SKX:       # BB#0:
; SKX-NEXT:    vmulsd %xmm1, %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    vsubsd (%rdi), %xmm0, %xmm0 # sched: [9:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a2 = load double , double *%a2_ptr
  %x = fmul double %a0, %a1
  %res = fsub double %x, %a2
  ret double %res
}

define double @test_x86_fmsub_231_m(double %a0, double %a1, double * %a2_ptr) {
; GENERIC-LABEL: test_x86_fmsub_231_m:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmulsd (%rdi), %xmm0, %xmm0 # sched: [11:1.00]
; GENERIC-NEXT:    vsubsd %xmm1, %xmm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_x86_fmsub_231_m:
; SKX:       # BB#0:
; SKX-NEXT:    vmulsd (%rdi), %xmm0, %xmm0 # sched: [9:0.50]
; SKX-NEXT:    vsubsd %xmm1, %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a2 = load double , double *%a2_ptr
  %x = fmul double %a0, %a2
  %res = fsub double %x, %a1
  ret double %res
}

define <16 x float> @test231_br(<16 x float> %a1, <16 x float> %a2) nounwind {
; GENERIC-LABEL: test231_br:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmulps {{.*}}(%rip){1to16}, %zmm0, %zmm0
; GENERIC-NEXT:    vaddps %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test231_br:
; SKX:       # BB#0:
; SKX-NEXT:    vmulps {{.*}}(%rip){1to16}, %zmm0, %zmm0 # sched: [11:0.50]
; SKX-NEXT:    vaddps %zmm1, %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b1 = fmul <16 x float> %a1, <float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000>
  %b2 = fadd <16 x float> %b1, %a2
  ret <16 x float> %b2
}

define <16 x float> @test213_br(<16 x float> %a1, <16 x float> %a2) nounwind {
; GENERIC-LABEL: test213_br:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmulps %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    vaddps {{.*}}(%rip){1to16}, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test213_br:
; SKX:       # BB#0:
; SKX-NEXT:    vmulps %zmm1, %zmm0, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    vaddps {{.*}}(%rip){1to16}, %zmm0, %zmm0 # sched: [11:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b1 = fmul <16 x float> %a1, %a2
  %b2 = fadd <16 x float> %b1, <float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000>
  ret <16 x float> %b2
}

;mask (a*c+b , a)
define <16 x float> @test_x86_fmadd132_ps(<16 x float> %a0, <16 x float> %a1, <16 x float> *%a2_ptrt, <16 x i1> %mask) {
; GENERIC-LABEL: test_x86_fmadd132_ps:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm2, %xmm2 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm2, %k1
; GENERIC-NEXT:    vmulps (%rdi), %zmm0, %zmm2
; GENERIC-NEXT:    vaddps %zmm1, %zmm2, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_x86_fmadd132_ps:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm2, %xmm2 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm2, %k1 # sched: [1:1.00]
; SKX-NEXT:    vmulps (%rdi), %zmm0, %zmm2 # sched: [11:0.50]
; SKX-NEXT:    vaddps %zmm1, %zmm2, %zmm0 {%k1} # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a2   = load <16 x float>,<16 x float> *%a2_ptrt,align 1
  %x = fmul <16 x float> %a0, %a2
  %y = fadd <16 x float> %x, %a1
  %res = select <16 x i1> %mask, <16 x float> %y, <16 x float> %a0
  ret <16 x float> %res
}

;mask (a*c+b , b)
define <16 x float> @test_x86_fmadd231_ps(<16 x float> %a0, <16 x float> %a1, <16 x float> *%a2_ptrt, <16 x i1> %mask) {
; GENERIC-LABEL: test_x86_fmadd231_ps:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm2, %xmm2 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm2, %k1
; GENERIC-NEXT:    vmulps (%rdi), %zmm0, %zmm0
; GENERIC-NEXT:    vaddps %zmm1, %zmm0, %zmm1 {%k1}
; GENERIC-NEXT:    vmovaps %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_x86_fmadd231_ps:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm2, %xmm2 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm2, %k1 # sched: [1:1.00]
; SKX-NEXT:    vmulps (%rdi), %zmm0, %zmm0 # sched: [11:0.50]
; SKX-NEXT:    vaddps %zmm1, %zmm0, %zmm1 {%k1} # sched: [4:0.33]
; SKX-NEXT:    vmovaps %zmm1, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a2   = load <16 x float>,<16 x float> *%a2_ptrt,align 1
  %x = fmul <16 x float> %a0, %a2
  %y = fadd <16 x float> %x, %a1
  %res = select <16 x i1> %mask, <16 x float> %y, <16 x float> %a1
  ret <16 x float> %res
}

;mask (b*a+c , b)
define <16 x float> @test_x86_fmadd213_ps(<16 x float> %a0, <16 x float> %a1, <16 x float> *%a2_ptrt, <16 x i1> %mask) {
; GENERIC-LABEL: test_x86_fmadd213_ps:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm2, %xmm2 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm2, %k1
; GENERIC-NEXT:    vmulps %zmm0, %zmm1, %zmm0
; GENERIC-NEXT:    vaddps (%rdi), %zmm0, %zmm1 {%k1}
; GENERIC-NEXT:    vmovaps %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_x86_fmadd213_ps:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm2, %xmm2 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm2, %k1 # sched: [1:1.00]
; SKX-NEXT:    vmulps %zmm0, %zmm1, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    vaddps (%rdi), %zmm0, %zmm1 {%k1} # sched: [11:0.50]
; SKX-NEXT:    vmovaps %zmm1, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a2   = load <16 x float>,<16 x float> *%a2_ptrt,align 1
  %x = fmul <16 x float> %a1, %a0
  %y = fadd <16 x float> %x, %a2
  %res = select <16 x i1> %mask, <16 x float> %y, <16 x float> %a1
  ret <16 x float> %res
}

define <16 x i32> @vpandd(<16 x i32> %a, <16 x i32> %b) nounwind uwtable readnone ssp {
; GENERIC-LABEL: vpandd:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vpaddd {{.*}}(%rip){1to16}, %zmm0, %zmm0
; GENERIC-NEXT:    vpandq %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpandd:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vpaddd {{.*}}(%rip){1to16}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    vpandq %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  ; Force the execution domain with an add.
  %a2 = add <16 x i32> %a, <i32 2, i32 2, i32 2, i32 2, i32 2, i32 2, i32 2, i32 2,
                            i32 2, i32 2, i32 2, i32 2, i32 2, i32 2, i32 2, i32 2>
  %x = and <16 x i32> %a2, %b
  ret <16 x i32> %x
}

define <16 x i32> @vpandnd(<16 x i32> %a, <16 x i32> %b) nounwind uwtable readnone ssp {
; GENERIC-LABEL: vpandnd:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vpaddd {{.*}}(%rip){1to16}, %zmm0, %zmm0
; GENERIC-NEXT:    vpandnq %zmm0, %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpandnd:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vpaddd {{.*}}(%rip){1to16}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    vpandnq %zmm0, %zmm1, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  ; Force the execution domain with an add.
  %a2 = add <16 x i32> %a, <i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3,
                            i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3>
  %b2 = xor <16 x i32> %b, <i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1,
                            i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1>
  %x = and <16 x i32> %a2, %b2
  ret <16 x i32> %x
}

define <16 x i32> @vpord(<16 x i32> %a, <16 x i32> %b) nounwind uwtable readnone ssp {
; GENERIC-LABEL: vpord:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vpaddd {{.*}}(%rip){1to16}, %zmm0, %zmm0
; GENERIC-NEXT:    vporq %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpord:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vpaddd {{.*}}(%rip){1to16}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    vporq %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  ; Force the execution domain with an add.
  %a2 = add <16 x i32> %a, <i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4,
                            i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4>
  %x = or <16 x i32> %a2, %b
  ret <16 x i32> %x
}

define <16 x i32> @vpxord(<16 x i32> %a, <16 x i32> %b) nounwind uwtable readnone ssp {
; GENERIC-LABEL: vpxord:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vpaddd {{.*}}(%rip){1to16}, %zmm0, %zmm0
; GENERIC-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpxord:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vpaddd {{.*}}(%rip){1to16}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    vpxorq %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  ; Force the execution domain with an add.
  %a2 = add <16 x i32> %a, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5,
                            i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
  %x = xor <16 x i32> %a2, %b
  ret <16 x i32> %x
}

define <8 x i64> @vpandq(<8 x i64> %a, <8 x i64> %b) nounwind uwtable readnone ssp {
; GENERIC-LABEL: vpandq:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm0
; GENERIC-NEXT:    vpandq %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpandq:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    vpandq %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  ; Force the execution domain with an add.
  %a2 = add <8 x i64> %a, <i64 6, i64 6, i64 6, i64 6, i64 6, i64 6, i64 6, i64 6>
  %x = and <8 x i64> %a2, %b
  ret <8 x i64> %x
}

define <8 x i64> @vpandnq(<8 x i64> %a, <8 x i64> %b) nounwind uwtable readnone ssp {
; GENERIC-LABEL: vpandnq:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm0
; GENERIC-NEXT:    vpandnq %zmm0, %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpandnq:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    vpandnq %zmm0, %zmm1, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  ; Force the execution domain with an add.
  %a2 = add <8 x i64> %a, <i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7, i64 7>
  %b2 = xor <8 x i64> %b, <i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1>
  %x = and <8 x i64> %a2, %b2
  ret <8 x i64> %x
}

define <8 x i64> @vporq(<8 x i64> %a, <8 x i64> %b) nounwind uwtable readnone ssp {
; GENERIC-LABEL: vporq:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm0
; GENERIC-NEXT:    vporq %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vporq:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    vporq %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  ; Force the execution domain with an add.
  %a2 = add <8 x i64> %a, <i64 8, i64 8, i64 8, i64 8, i64 8, i64 8, i64 8, i64 8>
  %x = or <8 x i64> %a2, %b
  ret <8 x i64> %x
}

define <8 x i64> @vpxorq(<8 x i64> %a, <8 x i64> %b) nounwind uwtable readnone ssp {
; GENERIC-LABEL: vpxorq:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm0
; GENERIC-NEXT:    vpxorq %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpxorq:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vpaddq {{.*}}(%rip){1to8}, %zmm0, %zmm0 # sched: [8:0.50]
; SKX-NEXT:    vpxorq %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  ; Force the execution domain with an add.
  %a2 = add <8 x i64> %a, <i64 9, i64 9, i64 9, i64 9, i64 9, i64 9, i64 9, i64 9>
  %x = xor <8 x i64> %a2, %b
  ret <8 x i64> %x
}

define <64 x i8> @and_v64i8(<64 x i8> %a, <64 x i8> %b) {
; GENERIC-LABEL: and_v64i8:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vandps %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: and_v64i8:
; SKX:       # BB#0:
; SKX-NEXT:    vandps %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = and <64 x i8> %a, %b
  ret <64 x i8> %res
}

define <64 x i8> @andn_v64i8(<64 x i8> %a, <64 x i8> %b) {
; GENERIC-LABEL: andn_v64i8:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vandnps %zmm0, %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: andn_v64i8:
; SKX:       # BB#0:
; SKX-NEXT:    vandnps %zmm0, %zmm1, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b2 = xor <64 x i8> %b, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1,
                           i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1,
                           i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1,
                           i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
  %res = and <64 x i8> %a, %b2
  ret <64 x i8> %res
}

define <64 x i8> @or_v64i8(<64 x i8> %a, <64 x i8> %b) {
; GENERIC-LABEL: or_v64i8:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vorps %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: or_v64i8:
; SKX:       # BB#0:
; SKX-NEXT:    vorps %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = or <64 x i8> %a, %b
  ret <64 x i8> %res
}

define <64 x i8> @xor_v64i8(<64 x i8> %a, <64 x i8> %b) {
; GENERIC-LABEL: xor_v64i8:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorps %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: xor_v64i8:
; SKX:       # BB#0:
; SKX-NEXT:    vxorps %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = xor <64 x i8> %a, %b
  ret <64 x i8> %res
}

define <32 x i16> @and_v32i16(<32 x i16> %a, <32 x i16> %b) {
; GENERIC-LABEL: and_v32i16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vandps %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: and_v32i16:
; SKX:       # BB#0:
; SKX-NEXT:    vandps %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = and <32 x i16> %a, %b
  ret <32 x i16> %res
}

define <32 x i16> @andn_v32i16(<32 x i16> %a, <32 x i16> %b) {
; GENERIC-LABEL: andn_v32i16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vandnps %zmm0, %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: andn_v32i16:
; SKX:       # BB#0:
; SKX-NEXT:    vandnps %zmm0, %zmm1, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b2 = xor <32 x i16> %b, <i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1,
                            i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1>
  %res = and <32 x i16> %a, %b2
  ret <32 x i16> %res
}

define <32 x i16> @or_v32i16(<32 x i16> %a, <32 x i16> %b) {
; GENERIC-LABEL: or_v32i16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vorps %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: or_v32i16:
; SKX:       # BB#0:
; SKX-NEXT:    vorps %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = or <32 x i16> %a, %b
  ret <32 x i16> %res
}

define <32 x i16> @xor_v32i16(<32 x i16> %a, <32 x i16> %b) {
; GENERIC-LABEL: xor_v32i16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorps %zmm1, %zmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: xor_v32i16:
; SKX:       # BB#0:
; SKX-NEXT:    vxorps %zmm1, %zmm0, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = xor <32 x i16> %a, %b
  ret <32 x i16> %res
}

define <16 x float> @masked_and_v16f32(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask, <16 x float> %c) {
; GENERIC-LABEL: masked_and_v16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vandps %zmm1, %zmm0, %zmm2 {%k1}
; GENERIC-NEXT:    vaddps %zmm2, %zmm3, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: masked_and_v16f32:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vandps %zmm1, %zmm0, %zmm2 {%k1} # sched: [1:0.33]
; SKX-NEXT:    vaddps %zmm2, %zmm3, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a1 = bitcast <16 x float> %a to <16 x i32>
  %b1 = bitcast <16 x float> %b to <16 x i32>
  %passThru1 = bitcast <16 x float> %passThru to <16 x i32>
  %mask1 = bitcast i16 %mask to <16 x i1>
  %op = and <16 x i32> %a1, %b1
  %select = select <16 x i1> %mask1, <16 x i32> %op, <16 x i32> %passThru1
  %cast = bitcast <16 x i32> %select to <16 x float>
  %add = fadd <16 x float> %c, %cast
  ret <16 x float> %add
}

define <16 x float> @masked_or_v16f32(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask, <16 x float> %c) {
; GENERIC-LABEL: masked_or_v16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vandps %zmm1, %zmm0, %zmm2 {%k1}
; GENERIC-NEXT:    vaddps %zmm2, %zmm3, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: masked_or_v16f32:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vandps %zmm1, %zmm0, %zmm2 {%k1} # sched: [1:0.33]
; SKX-NEXT:    vaddps %zmm2, %zmm3, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a1 = bitcast <16 x float> %a to <16 x i32>
  %b1 = bitcast <16 x float> %b to <16 x i32>
  %passThru1 = bitcast <16 x float> %passThru to <16 x i32>
  %mask1 = bitcast i16 %mask to <16 x i1>
  %op = and <16 x i32> %a1, %b1
  %select = select <16 x i1> %mask1, <16 x i32> %op, <16 x i32> %passThru1
  %cast = bitcast <16 x i32> %select to <16 x float>
  %add = fadd <16 x float> %c, %cast
  ret <16 x float> %add
}

define <16 x float> @masked_xor_v16f32(<16 x float> %a, <16 x float> %b, <16 x float> %passThru, i16 %mask, <16 x float> %c) {
; GENERIC-LABEL: masked_xor_v16f32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vandps %zmm1, %zmm0, %zmm2 {%k1}
; GENERIC-NEXT:    vaddps %zmm2, %zmm3, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: masked_xor_v16f32:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vandps %zmm1, %zmm0, %zmm2 {%k1} # sched: [1:0.33]
; SKX-NEXT:    vaddps %zmm2, %zmm3, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a1 = bitcast <16 x float> %a to <16 x i32>
  %b1 = bitcast <16 x float> %b to <16 x i32>
  %passThru1 = bitcast <16 x float> %passThru to <16 x i32>
  %mask1 = bitcast i16 %mask to <16 x i1>
  %op = and <16 x i32> %a1, %b1
  %select = select <16 x i1> %mask1, <16 x i32> %op, <16 x i32> %passThru1
  %cast = bitcast <16 x i32> %select to <16 x float>
  %add = fadd <16 x float> %c, %cast
  ret <16 x float> %add
}

define <8 x double> @masked_and_v8f64(<8 x double> %a, <8 x double> %b, <8 x double> %passThru, i8 %mask, <8 x double> %c) {
; GENERIC-LABEL: masked_and_v8f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vandpd %zmm1, %zmm0, %zmm2 {%k1}
; GENERIC-NEXT:    vaddpd %zmm2, %zmm3, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: masked_and_v8f64:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vandpd %zmm1, %zmm0, %zmm2 {%k1} # sched: [1:0.33]
; SKX-NEXT:    vaddpd %zmm2, %zmm3, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a1 = bitcast <8 x double> %a to <8 x i64>
  %b1 = bitcast <8 x double> %b to <8 x i64>
  %passThru1 = bitcast <8 x double> %passThru to <8 x i64>
  %mask1 = bitcast i8 %mask to <8 x i1>
  %op = and <8 x i64> %a1, %b1
  %select = select <8 x i1> %mask1, <8 x i64> %op, <8 x i64> %passThru1
  %cast = bitcast <8 x i64> %select to <8 x double>
  %add = fadd <8 x double> %c, %cast
  ret <8 x double> %add
}

define <8 x double> @masked_or_v8f64(<8 x double> %a, <8 x double> %b, <8 x double> %passThru, i8 %mask, <8 x double> %c) {
; GENERIC-LABEL: masked_or_v8f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vandpd %zmm1, %zmm0, %zmm2 {%k1}
; GENERIC-NEXT:    vaddpd %zmm2, %zmm3, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: masked_or_v8f64:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vandpd %zmm1, %zmm0, %zmm2 {%k1} # sched: [1:0.33]
; SKX-NEXT:    vaddpd %zmm2, %zmm3, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a1 = bitcast <8 x double> %a to <8 x i64>
  %b1 = bitcast <8 x double> %b to <8 x i64>
  %passThru1 = bitcast <8 x double> %passThru to <8 x i64>
  %mask1 = bitcast i8 %mask to <8 x i1>
  %op = and <8 x i64> %a1, %b1
  %select = select <8 x i1> %mask1, <8 x i64> %op, <8 x i64> %passThru1
  %cast = bitcast <8 x i64> %select to <8 x double>
  %add = fadd <8 x double> %c, %cast
  ret <8 x double> %add
}

define <8 x double> @masked_xor_v8f64(<8 x double> %a, <8 x double> %b, <8 x double> %passThru, i8 %mask, <8 x double> %c) {
; GENERIC-LABEL: masked_xor_v8f64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vandpd %zmm1, %zmm0, %zmm2 {%k1}
; GENERIC-NEXT:    vaddpd %zmm2, %zmm3, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: masked_xor_v8f64:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vandpd %zmm1, %zmm0, %zmm2 {%k1} # sched: [1:0.33]
; SKX-NEXT:    vaddpd %zmm2, %zmm3, %zmm0 # sched: [4:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a1 = bitcast <8 x double> %a to <8 x i64>
  %b1 = bitcast <8 x double> %b to <8 x i64>
  %passThru1 = bitcast <8 x double> %passThru to <8 x i64>
  %mask1 = bitcast i8 %mask to <8 x i1>
  %op = and <8 x i64> %a1, %b1
  %select = select <8 x i1> %mask1, <8 x i64> %op, <8 x i64> %passThru1
  %cast = bitcast <8 x i64> %select to <8 x double>
  %add = fadd <8 x double> %c, %cast
  ret <8 x double> %add
}

define <8 x i64> @test_mm512_mask_and_epi32(<8 x i64> %__src, i16 zeroext %__k, <8 x i64> %__a, <8 x i64> %__b) {
; GENERIC-LABEL: test_mm512_mask_and_epi32:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vandps %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_mask_and_epi32:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vandps %zmm2, %zmm1, %zmm0 {%k1} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %and1.i.i = and <8 x i64> %__a, %__b
  %0 = bitcast <8 x i64> %and1.i.i to <16 x i32>
  %1 = bitcast <8 x i64> %__src to <16 x i32>
  %2 = bitcast i16 %__k to <16 x i1>
  %3 = select <16 x i1> %2, <16 x i32> %0, <16 x i32> %1
  %4 = bitcast <16 x i32> %3 to <8 x i64>
  ret <8 x i64> %4
}

define <8 x i64> @test_mm512_mask_or_epi32(<8 x i64> %__src, i16 zeroext %__k, <8 x i64> %__a, <8 x i64> %__b) {
; GENERIC-LABEL: test_mm512_mask_or_epi32:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vorps %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_mask_or_epi32:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vorps %zmm2, %zmm1, %zmm0 {%k1} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %or1.i.i = or <8 x i64> %__a, %__b
  %0 = bitcast <8 x i64> %or1.i.i to <16 x i32>
  %1 = bitcast <8 x i64> %__src to <16 x i32>
  %2 = bitcast i16 %__k to <16 x i1>
  %3 = select <16 x i1> %2, <16 x i32> %0, <16 x i32> %1
  %4 = bitcast <16 x i32> %3 to <8 x i64>
  ret <8 x i64> %4
}

define <8 x i64> @test_mm512_mask_xor_epi32(<8 x i64> %__src, i16 zeroext %__k, <8 x i64> %__a, <8 x i64> %__b) {
; GENERIC-LABEL: test_mm512_mask_xor_epi32:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vxorps %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_mask_xor_epi32:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vxorps %zmm2, %zmm1, %zmm0 {%k1} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %xor1.i.i = xor <8 x i64> %__a, %__b
  %0 = bitcast <8 x i64> %xor1.i.i to <16 x i32>
  %1 = bitcast <8 x i64> %__src to <16 x i32>
  %2 = bitcast i16 %__k to <16 x i1>
  %3 = select <16 x i1> %2, <16 x i32> %0, <16 x i32> %1
  %4 = bitcast <16 x i32> %3 to <8 x i64>
  ret <8 x i64> %4
}

define <8 x double> @test_mm512_mask_xor_pd(<8 x double> %__W, i8 zeroext %__U, <8 x double> %__A, <8 x double> %__B) {
; GENERIC-LABEL: test_mm512_mask_xor_pd:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vxorpd %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_mask_xor_pd:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vxorpd %zmm2, %zmm1, %zmm0 {%k1} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <8 x double> %__A to <8 x i64>
  %1 = bitcast <8 x double> %__B to <8 x i64>
  %xor.i.i = xor <8 x i64> %0, %1
  %2 = bitcast <8 x i64> %xor.i.i to <8 x double>
  %3 = bitcast i8 %__U to <8 x i1>
  %4 = select <8 x i1> %3, <8 x double> %2, <8 x double> %__W
  ret <8 x double> %4
}

define <8 x double> @test_mm512_maskz_xor_pd(i8 zeroext %__U, <8 x double> %__A, <8 x double> %__B) {
; GENERIC-LABEL: test_mm512_maskz_xor_pd:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vxorpd %zmm1, %zmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_maskz_xor_pd:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vxorpd %zmm1, %zmm0, %zmm0 {%k1} {z} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <8 x double> %__A to <8 x i64>
  %1 = bitcast <8 x double> %__B to <8 x i64>
  %xor.i.i = xor <8 x i64> %0, %1
  %2 = bitcast <8 x i64> %xor.i.i to <8 x double>
  %3 = bitcast i8 %__U to <8 x i1>
  %4 = select <8 x i1> %3, <8 x double> %2, <8 x double> zeroinitializer
  ret <8 x double> %4
}

define <16 x float> @test_mm512_mask_xor_ps(<16 x float> %__W, i16 zeroext %__U, <16 x float> %__A, <16 x float> %__B) {
; GENERIC-LABEL: test_mm512_mask_xor_ps:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vxorps %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_mask_xor_ps:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vxorps %zmm2, %zmm1, %zmm0 {%k1} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <16 x float> %__A to <16 x i32>
  %1 = bitcast <16 x float> %__B to <16 x i32>
  %xor.i.i = xor <16 x i32> %0, %1
  %2 = bitcast <16 x i32> %xor.i.i to <16 x float>
  %3 = bitcast i16 %__U to <16 x i1>
  %4 = select <16 x i1> %3, <16 x float> %2, <16 x float> %__W
  ret <16 x float> %4
}

define <16 x float> @test_mm512_maskz_xor_ps(i16 zeroext %__U, <16 x float> %__A, <16 x float> %__B) {
; GENERIC-LABEL: test_mm512_maskz_xor_ps:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vxorps %zmm1, %zmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_maskz_xor_ps:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vxorps %zmm1, %zmm0, %zmm0 {%k1} {z} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <16 x float> %__A to <16 x i32>
  %1 = bitcast <16 x float> %__B to <16 x i32>
  %xor.i.i = xor <16 x i32> %0, %1
  %2 = bitcast <16 x i32> %xor.i.i to <16 x float>
  %3 = bitcast i16 %__U to <16 x i1>
  %4 = select <16 x i1> %3, <16 x float> %2, <16 x float> zeroinitializer
  ret <16 x float> %4
}

define <8 x double> @test_mm512_mask_or_pd(<8 x double> %__W, i8 zeroext %__U, <8 x double> %__A, <8 x double> %__B) {
; GENERIC-LABEL: test_mm512_mask_or_pd:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vorpd %zmm1, %zmm2, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_mask_or_pd:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vorpd %zmm1, %zmm2, %zmm0 {%k1} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <8 x double> %__A to <8 x i64>
  %1 = bitcast <8 x double> %__B to <8 x i64>
  %or.i.i = or <8 x i64> %1, %0
  %2 = bitcast <8 x i64> %or.i.i to <8 x double>
  %3 = bitcast i8 %__U to <8 x i1>
  %4 = select <8 x i1> %3, <8 x double> %2, <8 x double> %__W
  ret <8 x double> %4
}

define <8 x double> @test_mm512_maskz_or_pd(i8 zeroext %__U, <8 x double> %__A, <8 x double> %__B) {
; GENERIC-LABEL: test_mm512_maskz_or_pd:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vorpd %zmm0, %zmm1, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_maskz_or_pd:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vorpd %zmm0, %zmm1, %zmm0 {%k1} {z} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <8 x double> %__A to <8 x i64>
  %1 = bitcast <8 x double> %__B to <8 x i64>
  %or.i.i = or <8 x i64> %1, %0
  %2 = bitcast <8 x i64> %or.i.i to <8 x double>
  %3 = bitcast i8 %__U to <8 x i1>
  %4 = select <8 x i1> %3, <8 x double> %2, <8 x double> zeroinitializer
  ret <8 x double> %4
}

define <16 x float> @test_mm512_mask_or_ps(<16 x float> %__W, i16 zeroext %__U, <16 x float> %__A, <16 x float> %__B) {
; GENERIC-LABEL: test_mm512_mask_or_ps:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vorps %zmm1, %zmm2, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_mask_or_ps:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vorps %zmm1, %zmm2, %zmm0 {%k1} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <16 x float> %__A to <16 x i32>
  %1 = bitcast <16 x float> %__B to <16 x i32>
  %or.i.i = or <16 x i32> %1, %0
  %2 = bitcast <16 x i32> %or.i.i to <16 x float>
  %3 = bitcast i16 %__U to <16 x i1>
  %4 = select <16 x i1> %3, <16 x float> %2, <16 x float> %__W
  ret <16 x float> %4
}

define <16 x float> @test_mm512_maskz_or_ps(i16 zeroext %__U, <16 x float> %__A, <16 x float> %__B) {
; GENERIC-LABEL: test_mm512_maskz_or_ps:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vorps %zmm0, %zmm1, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_maskz_or_ps:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vorps %zmm0, %zmm1, %zmm0 {%k1} {z} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <16 x float> %__A to <16 x i32>
  %1 = bitcast <16 x float> %__B to <16 x i32>
  %or.i.i = or <16 x i32> %1, %0
  %2 = bitcast <16 x i32> %or.i.i to <16 x float>
  %3 = bitcast i16 %__U to <16 x i1>
  %4 = select <16 x i1> %3, <16 x float> %2, <16 x float> zeroinitializer
  ret <16 x float> %4
}

define <8 x double> @test_mm512_mask_and_pd(<8 x double> %__W, i8 zeroext %__U, <8 x double> %__A, <8 x double> %__B) {
; GENERIC-LABEL: test_mm512_mask_and_pd:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vandpd %zmm1, %zmm2, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_mask_and_pd:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vandpd %zmm1, %zmm2, %zmm0 {%k1} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <8 x double> %__A to <8 x i64>
  %1 = bitcast <8 x double> %__B to <8 x i64>
  %and.i.i = and <8 x i64> %1, %0
  %2 = bitcast <8 x i64> %and.i.i to <8 x double>
  %3 = bitcast i8 %__U to <8 x i1>
  %4 = select <8 x i1> %3, <8 x double> %2, <8 x double> %__W
  ret <8 x double> %4
}

define <8 x double> @test_mm512_maskz_and_pd(i8 zeroext %__U, <8 x double> %__A, <8 x double> %__B) {
; GENERIC-LABEL: test_mm512_maskz_and_pd:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vandpd %zmm0, %zmm1, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_maskz_and_pd:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vandpd %zmm0, %zmm1, %zmm0 {%k1} {z} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <8 x double> %__A to <8 x i64>
  %1 = bitcast <8 x double> %__B to <8 x i64>
  %and.i.i = and <8 x i64> %1, %0
  %2 = bitcast <8 x i64> %and.i.i to <8 x double>
  %3 = bitcast i8 %__U to <8 x i1>
  %4 = select <8 x i1> %3, <8 x double> %2, <8 x double> zeroinitializer
  ret <8 x double> %4
}

define <16 x float> @test_mm512_mask_and_ps(<16 x float> %__W, i16 zeroext %__U, <16 x float> %__A, <16 x float> %__B) {
; GENERIC-LABEL: test_mm512_mask_and_ps:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vandps %zmm1, %zmm2, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_mask_and_ps:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vandps %zmm1, %zmm2, %zmm0 {%k1} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <16 x float> %__A to <16 x i32>
  %1 = bitcast <16 x float> %__B to <16 x i32>
  %and.i.i = and <16 x i32> %1, %0
  %2 = bitcast <16 x i32> %and.i.i to <16 x float>
  %3 = bitcast i16 %__U to <16 x i1>
  %4 = select <16 x i1> %3, <16 x float> %2, <16 x float> %__W
  ret <16 x float> %4
}

define <16 x float> @test_mm512_maskz_and_ps(i16 zeroext %__U, <16 x float> %__A, <16 x float> %__B) {
; GENERIC-LABEL: test_mm512_maskz_and_ps:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vandps %zmm0, %zmm1, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_maskz_and_ps:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vandps %zmm0, %zmm1, %zmm0 {%k1} {z} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <16 x float> %__A to <16 x i32>
  %1 = bitcast <16 x float> %__B to <16 x i32>
  %and.i.i = and <16 x i32> %1, %0
  %2 = bitcast <16 x i32> %and.i.i to <16 x float>
  %3 = bitcast i16 %__U to <16 x i1>
  %4 = select <16 x i1> %3, <16 x float> %2, <16 x float> zeroinitializer
  ret <16 x float> %4
}

define <8 x double> @test_mm512_mask_andnot_pd(<8 x double> %__W, i8 zeroext %__U, <8 x double> %__A, <8 x double> %__B) {
; GENERIC-LABEL: test_mm512_mask_andnot_pd:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vandnpd %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_mask_andnot_pd:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vandnpd %zmm2, %zmm1, %zmm0 {%k1} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <8 x double> %__A to <8 x i64>
  %neg.i.i = xor <8 x i64> %0, <i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1>
  %1 = bitcast <8 x double> %__B to <8 x i64>
  %and.i.i = and <8 x i64> %1, %neg.i.i
  %2 = bitcast <8 x i64> %and.i.i to <8 x double>
  %3 = bitcast i8 %__U to <8 x i1>
  %4 = select <8 x i1> %3, <8 x double> %2, <8 x double> %__W
  ret <8 x double> %4
}

define <8 x double> @test_mm512_maskz_andnot_pd(i8 zeroext %__U, <8 x double> %__A, <8 x double> %__B) {
; GENERIC-LABEL: test_mm512_maskz_andnot_pd:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vandnpd %zmm1, %zmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_maskz_andnot_pd:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vandnpd %zmm1, %zmm0, %zmm0 {%k1} {z} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <8 x double> %__A to <8 x i64>
  %neg.i.i = xor <8 x i64> %0, <i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1>
  %1 = bitcast <8 x double> %__B to <8 x i64>
  %and.i.i = and <8 x i64> %1, %neg.i.i
  %2 = bitcast <8 x i64> %and.i.i to <8 x double>
  %3 = bitcast i8 %__U to <8 x i1>
  %4 = select <8 x i1> %3, <8 x double> %2, <8 x double> zeroinitializer
  ret <8 x double> %4
}

define <16 x float> @test_mm512_mask_andnot_ps(<16 x float> %__W, i16 zeroext %__U, <16 x float> %__A, <16 x float> %__B) {
; GENERIC-LABEL: test_mm512_mask_andnot_ps:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vandnps %zmm2, %zmm1, %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_mask_andnot_ps:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vandnps %zmm2, %zmm1, %zmm0 {%k1} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <16 x float> %__A to <16 x i32>
  %neg.i.i = xor <16 x i32> %0, <i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1>
  %1 = bitcast <16 x float> %__B to <16 x i32>
  %and.i.i = and <16 x i32> %1, %neg.i.i
  %2 = bitcast <16 x i32> %and.i.i to <16 x float>
  %3 = bitcast i16 %__U to <16 x i1>
  %4 = select <16 x i1> %3, <16 x float> %2, <16 x float> %__W
  ret <16 x float> %4
}

define <16 x float> @test_mm512_maskz_andnot_ps(i16 zeroext %__U, <16 x float> %__A, <16 x float> %__B) {
; GENERIC-LABEL: test_mm512_maskz_andnot_ps:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    vandnps %zmm1, %zmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_maskz_andnot_ps:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    vandnps %zmm1, %zmm0, %zmm0 {%k1} {z} # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = bitcast <16 x float> %__A to <16 x i32>
  %neg.i.i = xor <16 x i32> %0, <i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1>
  %1 = bitcast <16 x float> %__B to <16 x i32>
  %and.i.i = and <16 x i32> %1, %neg.i.i
  %2 = bitcast <16 x i32> %and.i.i to <16 x float>
  %3 = bitcast i16 %__U to <16 x i1>
  %4 = select <16 x i1> %3, <16 x float> %2, <16 x float> zeroinitializer
  ret <16 x float> %4
}

define i32 @mov_test1(float %x) {
; GENERIC-LABEL: mov_test1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovd %xmm0, %eax # sched: [1:0.33]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test1:
; SKX:       # BB#0:
; SKX-NEXT:    vmovd %xmm0, %eax # sched: [1:0.25]
; SKX-NEXT:    retq # sched: [7:1.00]
   %res = bitcast float %x to i32
   ret i32 %res
}

define <4 x i32> @mov_test2(i32 %x) {
; GENERIC-LABEL: mov_test2:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovd %edi, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test2:
; SKX:       # BB#0:
; SKX-NEXT:    vmovd %edi, %xmm0 # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
   %res = insertelement <4 x i32>undef, i32 %x, i32 0
   ret <4 x i32>%res
}

define <2 x i64> @mov_test3(i64 %x) {
; GENERIC-LABEL: mov_test3:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovq %rdi, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test3:
; SKX:       # BB#0:
; SKX-NEXT:    vmovq %rdi, %xmm0 # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
   %res = insertelement <2 x i64>undef, i64 %x, i32 0
   ret <2 x i64>%res
}

define <4 x i32> @mov_test4(i32* %x) {
; GENERIC-LABEL: mov_test4:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero sched: [6:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test4:
; SKX:       # BB#0:
; SKX-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero sched: [5:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
   %y = load i32, i32* %x
   %res = insertelement <4 x i32>undef, i32 %y, i32 0
   ret <4 x i32>%res
}

define void @mov_test5(float %x, float* %y) {
; GENERIC-LABEL: mov_test5:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovss %xmm0, (%rdi) # sched: [5:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test5:
; SKX:       # BB#0:
; SKX-NEXT:    vmovss %xmm0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
   store float %x, float* %y, align 4
   ret void
}

define void @mov_test6(double %x, double* %y) {
; GENERIC-LABEL: mov_test6:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovsd %xmm0, (%rdi) # sched: [5:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test6:
; SKX:       # BB#0:
; SKX-NEXT:    vmovsd %xmm0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
   store double %x, double* %y, align 8
   ret void
}

define float @mov_test7(i32* %x) {
; GENERIC-LABEL: mov_test7:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero sched: [6:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test7:
; SKX:       # BB#0:
; SKX-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero sched: [5:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
   %y = load i32, i32* %x
   %res = bitcast i32 %y to float
   ret float %res
}

define i32 @mov_test8(<4 x i32> %x) {
; GENERIC-LABEL: mov_test8:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovd %xmm0, %eax # sched: [2:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test8:
; SKX:       # BB#0:
; SKX-NEXT:    vmovd %xmm0, %eax # sched: [2:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
   %res = extractelement <4 x i32> %x, i32 0
   ret i32 %res
}

define i64 @mov_test9(<2 x i64> %x) {
; GENERIC-LABEL: mov_test9:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovq %xmm0, %rax # sched: [2:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test9:
; SKX:       # BB#0:
; SKX-NEXT:    vmovq %xmm0, %rax # sched: [2:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
   %res = extractelement <2 x i64> %x, i32 0
   ret i64 %res
}

define <4 x i32> @mov_test10(i32* %x) {
; GENERIC-LABEL: mov_test10:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero sched: [6:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test10:
; SKX:       # BB#0:
; SKX-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero sched: [5:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
   %y = load i32, i32* %x, align 4
   %res = insertelement <4 x i32>zeroinitializer, i32 %y, i32 0
   ret <4 x i32>%res
}

define <4 x float> @mov_test11(float* %x) {
; GENERIC-LABEL: mov_test11:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero sched: [6:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test11:
; SKX:       # BB#0:
; SKX-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero sched: [5:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
   %y = load float, float* %x, align 4
   %res = insertelement <4 x float>zeroinitializer, float %y, i32 0
   ret <4 x float>%res
}

define <2 x double> @mov_test12(double* %x) {
; GENERIC-LABEL: mov_test12:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovsd {{.*#+}} xmm0 = mem[0],zero sched: [6:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test12:
; SKX:       # BB#0:
; SKX-NEXT:    vmovsd {{.*#+}} xmm0 = mem[0],zero sched: [5:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
   %y = load double, double* %x, align 8
   %res = insertelement <2 x double>zeroinitializer, double %y, i32 0
   ret <2 x double>%res
}

define <2 x i64> @mov_test13(i64 %x) {
; GENERIC-LABEL: mov_test13:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovq %rdi, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test13:
; SKX:       # BB#0:
; SKX-NEXT:    vmovq %rdi, %xmm0 # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
   %res = insertelement <2 x i64>zeroinitializer, i64 %x, i32 0
   ret <2 x i64>%res
}

define <4 x i32> @mov_test14(i32 %x) {
; GENERIC-LABEL: mov_test14:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovd %edi, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test14:
; SKX:       # BB#0:
; SKX-NEXT:    vmovd %edi, %xmm0 # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
   %res = insertelement <4 x i32>zeroinitializer, i32 %x, i32 0
   ret <4 x i32>%res
}

define <4 x i32> @mov_test15(i32* %x) {
; GENERIC-LABEL: mov_test15:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero sched: [6:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test15:
; SKX:       # BB#0:
; SKX-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero sched: [5:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
   %y = load i32, i32* %x, align 4
   %res = insertelement <4 x i32>zeroinitializer, i32 %y, i32 0
   ret <4 x i32>%res
}

define <16 x i32> @mov_test16(i8 * %addr) {
; GENERIC-LABEL: mov_test16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovups (%rdi), %zmm0 # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test16:
; SKX:       # BB#0:
; SKX-NEXT:    vmovups (%rdi), %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <16 x i32>*
  %res = load <16 x i32>, <16 x i32>* %vaddr, align 1
  ret <16 x i32>%res
}

define <16 x i32> @mov_test17(i8 * %addr) {
; GENERIC-LABEL: mov_test17:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovaps (%rdi), %zmm0 # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test17:
; SKX:       # BB#0:
; SKX-NEXT:    vmovaps (%rdi), %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <16 x i32>*
  %res = load <16 x i32>, <16 x i32>* %vaddr, align 64
  ret <16 x i32>%res
}

define void @mov_test18(i8 * %addr, <8 x i64> %data) {
; GENERIC-LABEL: mov_test18:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovaps %zmm0, (%rdi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test18:
; SKX:       # BB#0:
; SKX-NEXT:    vmovaps %zmm0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <8 x i64>*
  store <8 x i64>%data, <8 x i64>* %vaddr, align 64
  ret void
}

define void @mov_test19(i8 * %addr, <16 x i32> %data) {
; GENERIC-LABEL: mov_test19:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovups %zmm0, (%rdi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test19:
; SKX:       # BB#0:
; SKX-NEXT:    vmovups %zmm0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <16 x i32>*
  store <16 x i32>%data, <16 x i32>* %vaddr, align 1
  ret void
}

define void @mov_test20(i8 * %addr, <16 x i32> %data) {
; GENERIC-LABEL: mov_test20:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovaps %zmm0, (%rdi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test20:
; SKX:       # BB#0:
; SKX-NEXT:    vmovaps %zmm0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <16 x i32>*
  store <16 x i32>%data, <16 x i32>* %vaddr, align 64
  ret void
}

define  <8 x i64> @mov_test21(i8 * %addr) {
; GENERIC-LABEL: mov_test21:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovaps (%rdi), %zmm0 # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test21:
; SKX:       # BB#0:
; SKX-NEXT:    vmovaps (%rdi), %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <8 x i64>*
  %res = load <8 x i64>, <8 x i64>* %vaddr, align 64
  ret <8 x i64>%res
}

define void @mov_test22(i8 * %addr, <8 x i64> %data) {
; GENERIC-LABEL: mov_test22:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovups %zmm0, (%rdi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test22:
; SKX:       # BB#0:
; SKX-NEXT:    vmovups %zmm0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <8 x i64>*
  store <8 x i64>%data, <8 x i64>* %vaddr, align 1
  ret void
}

define <8 x i64> @mov_test23(i8 * %addr) {
; GENERIC-LABEL: mov_test23:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovups (%rdi), %zmm0 # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test23:
; SKX:       # BB#0:
; SKX-NEXT:    vmovups (%rdi), %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <8 x i64>*
  %res = load <8 x i64>, <8 x i64>* %vaddr, align 1
  ret <8 x i64>%res
}

define void @mov_test24(i8 * %addr, <8 x double> %data) {
; GENERIC-LABEL: mov_test24:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovaps %zmm0, (%rdi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test24:
; SKX:       # BB#0:
; SKX-NEXT:    vmovaps %zmm0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <8 x double>*
  store <8 x double>%data, <8 x double>* %vaddr, align 64
  ret void
}

define <8 x double> @mov_test25(i8 * %addr) {
; GENERIC-LABEL: mov_test25:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovaps (%rdi), %zmm0 # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test25:
; SKX:       # BB#0:
; SKX-NEXT:    vmovaps (%rdi), %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <8 x double>*
  %res = load <8 x double>, <8 x double>* %vaddr, align 64
  ret <8 x double>%res
}

define void @mov_test26(i8 * %addr, <16 x float> %data) {
; GENERIC-LABEL: mov_test26:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovaps %zmm0, (%rdi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test26:
; SKX:       # BB#0:
; SKX-NEXT:    vmovaps %zmm0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <16 x float>*
  store <16 x float>%data, <16 x float>* %vaddr, align 64
  ret void
}

define <16 x float> @mov_test27(i8 * %addr) {
; GENERIC-LABEL: mov_test27:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovaps (%rdi), %zmm0 # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test27:
; SKX:       # BB#0:
; SKX-NEXT:    vmovaps (%rdi), %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <16 x float>*
  %res = load <16 x float>, <16 x float>* %vaddr, align 64
  ret <16 x float>%res
}

define void @mov_test28(i8 * %addr, <8 x double> %data) {
; GENERIC-LABEL: mov_test28:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovups %zmm0, (%rdi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test28:
; SKX:       # BB#0:
; SKX-NEXT:    vmovups %zmm0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <8 x double>*
  store <8 x double>%data, <8 x double>* %vaddr, align 1
  ret void
}

define <8 x double> @mov_test29(i8 * %addr) {
; GENERIC-LABEL: mov_test29:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovups (%rdi), %zmm0 # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test29:
; SKX:       # BB#0:
; SKX-NEXT:    vmovups (%rdi), %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <8 x double>*
  %res = load <8 x double>, <8 x double>* %vaddr, align 1
  ret <8 x double>%res
}

define void @mov_test30(i8 * %addr, <16 x float> %data) {
; GENERIC-LABEL: mov_test30:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovups %zmm0, (%rdi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test30:
; SKX:       # BB#0:
; SKX-NEXT:    vmovups %zmm0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <16 x float>*
  store <16 x float>%data, <16 x float>* %vaddr, align 1
  ret void
}

define <16 x float> @mov_test31(i8 * %addr) {
; GENERIC-LABEL: mov_test31:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovups (%rdi), %zmm0 # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test31:
; SKX:       # BB#0:
; SKX-NEXT:    vmovups (%rdi), %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %vaddr = bitcast i8* %addr to <16 x float>*
  %res = load <16 x float>, <16 x float>* %vaddr, align 1
  ret <16 x float>%res
}

define <16 x i32> @mov_test32(i8 * %addr, <16 x i32> %old, <16 x i32> %mask1) {
; GENERIC-LABEL: mov_test32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vmovdqa32 (%rdi), %zmm0 {%k1} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test32:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovdqa32 (%rdi), %zmm0 {%k1} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <16 x i32>*
  %r = load <16 x i32>, <16 x i32>* %vaddr, align 64
  %res = select <16 x i1> %mask, <16 x i32> %r, <16 x i32> %old
  ret <16 x i32>%res
}

define <16 x i32> @mov_test33(i8 * %addr, <16 x i32> %old, <16 x i32> %mask1) {
; GENERIC-LABEL: mov_test33:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vmovdqu32 (%rdi), %zmm0 {%k1} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test33:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovdqu32 (%rdi), %zmm0 {%k1} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <16 x i32>*
  %r = load <16 x i32>, <16 x i32>* %vaddr, align 1
  %res = select <16 x i1> %mask, <16 x i32> %r, <16 x i32> %old
  ret <16 x i32>%res
}

define <16 x i32> @mov_test34(i8 * %addr, <16 x i32> %mask1) {
; GENERIC-LABEL: mov_test34:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm1, %zmm0, %k1
; GENERIC-NEXT:    vmovdqa32 (%rdi), %zmm0 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test34:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm1, %zmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovdqa32 (%rdi), %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <16 x i32>*
  %r = load <16 x i32>, <16 x i32>* %vaddr, align 64
  %res = select <16 x i1> %mask, <16 x i32> %r, <16 x i32> zeroinitializer
  ret <16 x i32>%res
}

define <16 x i32> @mov_test35(i8 * %addr, <16 x i32> %mask1) {
; GENERIC-LABEL: mov_test35:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm1, %zmm0, %k1
; GENERIC-NEXT:    vmovdqu32 (%rdi), %zmm0 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test35:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm1, %zmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovdqu32 (%rdi), %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <16 x i32>*
  %r = load <16 x i32>, <16 x i32>* %vaddr, align 1
  %res = select <16 x i1> %mask, <16 x i32> %r, <16 x i32> zeroinitializer
  ret <16 x i32>%res
}

define <8 x i64> @mov_test36(i8 * %addr, <8 x i64> %old, <8 x i64> %mask1) {
; GENERIC-LABEL: mov_test36:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqq %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vmovdqa64 (%rdi), %zmm0 {%k1} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test36:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqq %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovdqa64 (%rdi), %zmm0 {%k1} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <8 x i64> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <8 x i64>*
  %r = load <8 x i64>, <8 x i64>* %vaddr, align 64
  %res = select <8 x i1> %mask, <8 x i64> %r, <8 x i64> %old
  ret <8 x i64>%res
}

define <8 x i64> @mov_test37(i8 * %addr, <8 x i64> %old, <8 x i64> %mask1) {
; GENERIC-LABEL: mov_test37:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqq %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vmovdqu64 (%rdi), %zmm0 {%k1} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test37:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqq %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovdqu64 (%rdi), %zmm0 {%k1} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <8 x i64> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <8 x i64>*
  %r = load <8 x i64>, <8 x i64>* %vaddr, align 1
  %res = select <8 x i1> %mask, <8 x i64> %r, <8 x i64> %old
  ret <8 x i64>%res
}

define <8 x i64> @mov_test38(i8 * %addr, <8 x i64> %mask1) {
; GENERIC-LABEL: mov_test38:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqq %zmm1, %zmm0, %k1
; GENERIC-NEXT:    vmovdqa64 (%rdi), %zmm0 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test38:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqq %zmm1, %zmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovdqa64 (%rdi), %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <8 x i64> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <8 x i64>*
  %r = load <8 x i64>, <8 x i64>* %vaddr, align 64
  %res = select <8 x i1> %mask, <8 x i64> %r, <8 x i64> zeroinitializer
  ret <8 x i64>%res
}

define <8 x i64> @mov_test39(i8 * %addr, <8 x i64> %mask1) {
; GENERIC-LABEL: mov_test39:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqq %zmm1, %zmm0, %k1
; GENERIC-NEXT:    vmovdqu64 (%rdi), %zmm0 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test39:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqq %zmm1, %zmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovdqu64 (%rdi), %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <8 x i64> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <8 x i64>*
  %r = load <8 x i64>, <8 x i64>* %vaddr, align 1
  %res = select <8 x i1> %mask, <8 x i64> %r, <8 x i64> zeroinitializer
  ret <8 x i64>%res
}

define <16 x float> @mov_test40(i8 * %addr, <16 x float> %old, <16 x float> %mask1) {
; GENERIC-LABEL: mov_test40:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorps %xmm2, %xmm2, %xmm2 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpneq_oqps %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vmovaps (%rdi), %zmm0 {%k1} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test40:
; SKX:       # BB#0:
; SKX-NEXT:    vxorps %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vcmpneq_oqps %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovaps (%rdi), %zmm0 {%k1} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = fcmp one <16 x float> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <16 x float>*
  %r = load <16 x float>, <16 x float>* %vaddr, align 64
  %res = select <16 x i1> %mask, <16 x float> %r, <16 x float> %old
  ret <16 x float>%res
}

define <16 x float> @mov_test41(i8 * %addr, <16 x float> %old, <16 x float> %mask1) {
; GENERIC-LABEL: mov_test41:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorps %xmm2, %xmm2, %xmm2 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpneq_oqps %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vmovups (%rdi), %zmm0 {%k1} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test41:
; SKX:       # BB#0:
; SKX-NEXT:    vxorps %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vcmpneq_oqps %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovups (%rdi), %zmm0 {%k1} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = fcmp one <16 x float> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <16 x float>*
  %r = load <16 x float>, <16 x float>* %vaddr, align 1
  %res = select <16 x i1> %mask, <16 x float> %r, <16 x float> %old
  ret <16 x float>%res
}

define <16 x float> @mov_test42(i8 * %addr, <16 x float> %mask1) {
; GENERIC-LABEL: mov_test42:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorps %xmm1, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpneq_oqps %zmm1, %zmm0, %k1
; GENERIC-NEXT:    vmovaps (%rdi), %zmm0 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test42:
; SKX:       # BB#0:
; SKX-NEXT:    vxorps %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vcmpneq_oqps %zmm1, %zmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovaps (%rdi), %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = fcmp one <16 x float> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <16 x float>*
  %r = load <16 x float>, <16 x float>* %vaddr, align 64
  %res = select <16 x i1> %mask, <16 x float> %r, <16 x float> zeroinitializer
  ret <16 x float>%res
}

define <16 x float> @mov_test43(i8 * %addr, <16 x float> %mask1) {
; GENERIC-LABEL: mov_test43:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorps %xmm1, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpneq_oqps %zmm1, %zmm0, %k1
; GENERIC-NEXT:    vmovups (%rdi), %zmm0 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test43:
; SKX:       # BB#0:
; SKX-NEXT:    vxorps %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vcmpneq_oqps %zmm1, %zmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovups (%rdi), %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = fcmp one <16 x float> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <16 x float>*
  %r = load <16 x float>, <16 x float>* %vaddr, align 1
  %res = select <16 x i1> %mask, <16 x float> %r, <16 x float> zeroinitializer
  ret <16 x float>%res
}

define <8 x double> @mov_test44(i8 * %addr, <8 x double> %old, <8 x double> %mask1) {
; GENERIC-LABEL: mov_test44:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorpd %xmm2, %xmm2, %xmm2 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpneq_oqpd %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vmovapd (%rdi), %zmm0 {%k1} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test44:
; SKX:       # BB#0:
; SKX-NEXT:    vxorpd %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vcmpneq_oqpd %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovapd (%rdi), %zmm0 {%k1} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = fcmp one <8 x double> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <8 x double>*
  %r = load <8 x double>, <8 x double>* %vaddr, align 64
  %res = select <8 x i1> %mask, <8 x double> %r, <8 x double> %old
  ret <8 x double>%res
}

define <8 x double> @mov_test45(i8 * %addr, <8 x double> %old, <8 x double> %mask1) {
; GENERIC-LABEL: mov_test45:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorpd %xmm2, %xmm2, %xmm2 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpneq_oqpd %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vmovupd (%rdi), %zmm0 {%k1} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test45:
; SKX:       # BB#0:
; SKX-NEXT:    vxorpd %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vcmpneq_oqpd %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovupd (%rdi), %zmm0 {%k1} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = fcmp one <8 x double> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <8 x double>*
  %r = load <8 x double>, <8 x double>* %vaddr, align 1
  %res = select <8 x i1> %mask, <8 x double> %r, <8 x double> %old
  ret <8 x double>%res
}

define <8 x double> @mov_test46(i8 * %addr, <8 x double> %mask1) {
; GENERIC-LABEL: mov_test46:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorpd %xmm1, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpneq_oqpd %zmm1, %zmm0, %k1
; GENERIC-NEXT:    vmovapd (%rdi), %zmm0 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test46:
; SKX:       # BB#0:
; SKX-NEXT:    vxorpd %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vcmpneq_oqpd %zmm1, %zmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovapd (%rdi), %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = fcmp one <8 x double> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <8 x double>*
  %r = load <8 x double>, <8 x double>* %vaddr, align 64
  %res = select <8 x i1> %mask, <8 x double> %r, <8 x double> zeroinitializer
  ret <8 x double>%res
}

define <8 x double> @mov_test47(i8 * %addr, <8 x double> %mask1) {
; GENERIC-LABEL: mov_test47:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vxorpd %xmm1, %xmm1, %xmm1 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpneq_oqpd %zmm1, %zmm0, %k1
; GENERIC-NEXT:    vmovupd (%rdi), %zmm0 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mov_test47:
; SKX:       # BB#0:
; SKX-NEXT:    vxorpd %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vcmpneq_oqpd %zmm1, %zmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovupd (%rdi), %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = fcmp one <8 x double> %mask1, zeroinitializer
  %vaddr = bitcast i8* %addr to <8 x double>*
  %r = load <8 x double>, <8 x double>* %vaddr, align 1
  %res = select <8 x i1> %mask, <8 x double> %r, <8 x double> zeroinitializer
  ret <8 x double>%res
}

define i16 @mask16(i16 %x) {
; GENERIC-LABEL: mask16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k0
; GENERIC-NEXT:    knotw %k0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mask16:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k0 # sched: [1:1.00]
; SKX-NEXT:    knotw %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; SKX-NEXT:    retq # sched: [7:1.00]
  %m0 = bitcast i16 %x to <16 x i1>
  %m1 = xor <16 x i1> %m0, <i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1>
  %ret = bitcast <16 x i1> %m1 to i16
  ret i16 %ret
}

define i32 @mask16_zext(i16 %x) {
; GENERIC-LABEL: mask16_zext:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k0
; GENERIC-NEXT:    knotw %k0, %k0
; GENERIC-NEXT:    kmovw %k0, %eax
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mask16_zext:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k0 # sched: [1:1.00]
; SKX-NEXT:    knotw %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovw %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %m0 = bitcast i16 %x to <16 x i1>
  %m1 = xor <16 x i1> %m0, <i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1>
  %m2 = bitcast <16 x i1> %m1 to i16
  %ret = zext i16 %m2 to i32
  ret i32 %ret
}

define i8 @mask8(i8 %x) {
; GENERIC-LABEL: mask8:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k0
; GENERIC-NEXT:    knotb %k0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    # kill: %AL<def> %AL<kill> %EAX<kill>
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mask8:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k0 # sched: [1:1.00]
; SKX-NEXT:    knotb %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    # kill: %AL<def> %AL<kill> %EAX<kill>
; SKX-NEXT:    retq # sched: [7:1.00]
  %m0 = bitcast i8 %x to <8 x i1>
  %m1 = xor <8 x i1> %m0, <i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1>
  %ret = bitcast <8 x i1> %m1 to i8
  ret i8 %ret
}

define i32 @mask8_zext(i8 %x) {
; GENERIC-LABEL: mask8_zext:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k0
; GENERIC-NEXT:    knotb %k0, %k0
; GENERIC-NEXT:    kmovb %k0, %eax
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mask8_zext:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k0 # sched: [1:1.00]
; SKX-NEXT:    knotb %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovb %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %m0 = bitcast i8 %x to <8 x i1>
  %m1 = xor <8 x i1> %m0, <i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1>
  %m2 = bitcast <8 x i1> %m1 to i8
  %ret = zext i8 %m2 to i32
  ret i32 %ret
}

define void @mask16_mem(i16* %ptr) {
; GENERIC-LABEL: mask16_mem:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovw (%rdi), %k0
; GENERIC-NEXT:    knotw %k0, %k0
; GENERIC-NEXT:    kmovw %k0, (%rdi)
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mask16_mem:
; SKX:       # BB#0:
; SKX-NEXT:    kmovw (%rdi), %k0 # sched: [7:1.00]
; SKX-NEXT:    knotw %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovw %k0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = load i16, i16* %ptr, align 4
  %m0 = bitcast i16 %x to <16 x i1>
  %m1 = xor <16 x i1> %m0, <i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1>
  %ret = bitcast <16 x i1> %m1 to i16
  store i16 %ret, i16* %ptr, align 4
  ret void
}

define void @mask8_mem(i8* %ptr) {
; GENERIC-LABEL: mask8_mem:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovb (%rdi), %k0
; GENERIC-NEXT:    knotb %k0, %k0
; GENERIC-NEXT:    kmovb %k0, (%rdi)
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mask8_mem:
; SKX:       # BB#0:
; SKX-NEXT:    kmovb (%rdi), %k0 # sched: [7:1.00]
; SKX-NEXT:    knotb %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovb %k0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = load i8, i8* %ptr, align 4
  %m0 = bitcast i8 %x to <8 x i1>
  %m1 = xor <8 x i1> %m0, <i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1, i1 -1>
  %ret = bitcast <8 x i1> %m1 to i8
  store i8 %ret, i8* %ptr, align 4
  ret void
}

define i16 @mand16(i16 %x, i16 %y) {
; GENERIC-LABEL: mand16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    movl %edi, %eax # sched: [1:0.33]
; GENERIC-NEXT:    xorl %esi, %eax # sched: [1:0.33]
; GENERIC-NEXT:    andl %esi, %edi # sched: [1:0.33]
; GENERIC-NEXT:    orl %eax, %edi # sched: [1:0.33]
; GENERIC-NEXT:    movl %edi, %eax # sched: [1:0.33]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mand16:
; SKX:       # BB#0:
; SKX-NEXT:    movl %edi, %eax # sched: [1:0.25]
; SKX-NEXT:    xorl %esi, %eax # sched: [1:0.25]
; SKX-NEXT:    andl %esi, %edi # sched: [1:0.25]
; SKX-NEXT:    orl %eax, %edi # sched: [1:0.25]
; SKX-NEXT:    movl %edi, %eax # sched: [1:0.25]
; SKX-NEXT:    retq # sched: [7:1.00]
  %ma = bitcast i16 %x to <16 x i1>
  %mb = bitcast i16 %y to <16 x i1>
  %mc = and <16 x i1> %ma, %mb
  %md = xor <16 x i1> %ma, %mb
  %me = or <16 x i1> %mc, %md
  %ret = bitcast <16 x i1> %me to i16
  ret i16 %ret
}

define i16 @mand16_mem(<16 x i1>* %x, <16 x i1>* %y) {
; GENERIC-LABEL: mand16_mem:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovw (%rdi), %k0
; GENERIC-NEXT:    kmovw (%rsi), %k1
; GENERIC-NEXT:    kandw %k1, %k0, %k2
; GENERIC-NEXT:    kxorw %k1, %k0, %k0
; GENERIC-NEXT:    korw %k0, %k2, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: mand16_mem:
; SKX:       # BB#0:
; SKX-NEXT:    kmovw (%rdi), %k0 # sched: [7:1.00]
; SKX-NEXT:    kmovw (%rsi), %k1 # sched: [7:1.00]
; SKX-NEXT:    kandw %k1, %k0, %k2 # sched: [1:1.00]
; SKX-NEXT:    kxorw %k1, %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    korw %k0, %k2, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; SKX-NEXT:    retq # sched: [7:1.00]
  %ma = load <16 x i1>, <16 x i1>* %x
  %mb = load <16 x i1>, <16 x i1>* %y
  %mc = and <16 x i1> %ma, %mb
  %md = xor <16 x i1> %ma, %mb
  %me = or <16 x i1> %mc, %md
  %ret = bitcast <16 x i1> %me to i16
  ret i16 %ret
}

define i8 @shuf_test1(i16 %v) nounwind {
; GENERIC-LABEL: shuf_test1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k0
; GENERIC-NEXT:    kshiftrw $8, %k0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    # kill: %AL<def> %AL<kill> %EAX<kill>
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: shuf_test1:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k0 # sched: [1:1.00]
; SKX-NEXT:    kshiftrw $8, %k0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    # kill: %AL<def> %AL<kill> %EAX<kill>
; SKX-NEXT:    retq # sched: [7:1.00]
   %v1 = bitcast i16 %v to <16 x i1>
   %mask = shufflevector <16 x i1> %v1, <16 x i1> undef, <8 x i32> <i32 8, i32 9, i32 10, i32 11, i32 12, i32 13, i32 14, i32 15>
   %mask1 = bitcast <8 x i1> %mask to i8
   ret i8 %mask1
}

define i32 @zext_test1(<16 x i32> %a, <16 x i32> %b) {
; GENERIC-LABEL: zext_test1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpcmpnleud %zmm1, %zmm0, %k0
; GENERIC-NEXT:    kshiftlw $10, %k0, %k0
; GENERIC-NEXT:    kshiftrw $15, %k0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    andl $1, %eax # sched: [1:0.33]
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_test1:
; SKX:       # BB#0:
; SKX-NEXT:    vpcmpnleud %zmm1, %zmm0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kshiftlw $10, %k0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kshiftrw $15, %k0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    andl $1, %eax # sched: [1:0.25]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %cmp_res = icmp ugt <16 x i32> %a, %b
  %cmp_res.i1 = extractelement <16 x i1> %cmp_res, i32 5
  %res = zext i1 %cmp_res.i1 to i32
  ret i32 %res
}

define i16 @zext_test2(<16 x i32> %a, <16 x i32> %b) {
; GENERIC-LABEL: zext_test2:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpcmpnleud %zmm1, %zmm0, %k0
; GENERIC-NEXT:    kshiftlw $10, %k0, %k0
; GENERIC-NEXT:    kshiftrw $15, %k0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    andl $1, %eax # sched: [1:0.33]
; GENERIC-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_test2:
; SKX:       # BB#0:
; SKX-NEXT:    vpcmpnleud %zmm1, %zmm0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kshiftlw $10, %k0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kshiftrw $15, %k0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    andl $1, %eax # sched: [1:0.25]
; SKX-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %cmp_res = icmp ugt <16 x i32> %a, %b
  %cmp_res.i1 = extractelement <16 x i1> %cmp_res, i32 5
  %res = zext i1 %cmp_res.i1 to i16
  ret i16 %res
}

define i8 @zext_test3(<16 x i32> %a, <16 x i32> %b) {
; GENERIC-LABEL: zext_test3:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpcmpnleud %zmm1, %zmm0, %k0
; GENERIC-NEXT:    kshiftlw $10, %k0, %k0
; GENERIC-NEXT:    kshiftrw $15, %k0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    andb $1, %al # sched: [1:0.33]
; GENERIC-NEXT:    # kill: %AL<def> %AL<kill> %EAX<kill>
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: zext_test3:
; SKX:       # BB#0:
; SKX-NEXT:    vpcmpnleud %zmm1, %zmm0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kshiftlw $10, %k0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kshiftrw $15, %k0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    andb $1, %al # sched: [1:0.25]
; SKX-NEXT:    # kill: %AL<def> %AL<kill> %EAX<kill>
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %cmp_res = icmp ugt <16 x i32> %a, %b
  %cmp_res.i1 = extractelement <16 x i1> %cmp_res, i32 5
  %res = zext i1 %cmp_res.i1 to i8
  ret i8 %res
}

define i8 @conv1(<8 x i1>* %R) {
; GENERIC-LABEL: conv1:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    kxnorw %k0, %k0, %k0
; GENERIC-NEXT:    kmovb %k0, (%rdi)
; GENERIC-NEXT:    movb $-2, -{{[0-9]+}}(%rsp) # sched: [5:1.00]
; GENERIC-NEXT:    movb $-2, %al # sched: [1:0.33]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: conv1:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    kxnorw %k0, %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovb %k0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    movb $-2, -{{[0-9]+}}(%rsp) # sched: [1:1.00]
; SKX-NEXT:    movb $-2, %al # sched: [1:0.25]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  store <8 x i1> <i1 1, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1>, <8 x i1>* %R

  %maskPtr = alloca <8 x i1>
  store <8 x i1> <i1 0, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1>, <8 x i1>* %maskPtr
  %mask = load <8 x i1>, <8 x i1>* %maskPtr
  %mask_convert = bitcast <8 x i1> %mask to i8
  ret i8 %mask_convert
}

define <4 x i32> @test4(<4 x i64> %x, <4 x i64> %y, <4 x i64> %x1, <4 x i64> %y1) {
; GENERIC-LABEL: test4:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpcmpgtq %ymm1, %ymm0, %k0
; GENERIC-NEXT:    vpcmpgtq %ymm3, %ymm2, %k1
; GENERIC-NEXT:    kandnw %k0, %k1, %k0
; GENERIC-NEXT:    vpmovm2d %k0, %xmm0
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test4:
; SKX:       # BB#0:
; SKX-NEXT:    vpcmpgtq %ymm1, %ymm0, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpcmpgtq %ymm3, %ymm2, %k1 # sched: [3:1.00]
; SKX-NEXT:    kandnw %k0, %k1, %k0 # sched: [1:1.00]
; SKX-NEXT:    vpmovm2d %k0, %xmm0
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x_gt_y = icmp sgt <4 x i64> %x, %y
  %x1_gt_y1 = icmp sgt <4 x i64> %x1, %y1
  %res = icmp sgt <4 x i1>%x_gt_y, %x1_gt_y1
  %resse = sext <4 x i1>%res to <4 x i32>
  ret <4 x i32> %resse
}

define <2 x i64> @vcmp_test5(<2 x i64> %x, <2 x i64> %y, <2 x i64> %x1, <2 x i64> %y1) {
; GENERIC-LABEL: vcmp_test5:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpcmpgtq %xmm0, %xmm1, %k0
; GENERIC-NEXT:    vpcmpgtq %xmm3, %xmm2, %k1
; GENERIC-NEXT:    kandnw %k1, %k0, %k0
; GENERIC-NEXT:    vpmovm2q %k0, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vcmp_test5:
; SKX:       # BB#0:
; SKX-NEXT:    vpcmpgtq %xmm0, %xmm1, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpcmpgtq %xmm3, %xmm2, %k1 # sched: [3:1.00]
; SKX-NEXT:    kandnw %k1, %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    vpmovm2q %k0, %xmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %x_gt_y = icmp slt <2 x i64> %x, %y
  %x1_gt_y1 = icmp sgt <2 x i64> %x1, %y1
  %res = icmp slt <2 x i1>%x_gt_y, %x1_gt_y1
  %resse = sext <2 x i1>%res to <2 x i64>
  ret <2 x i64> %resse
}define void @vcmp_test6(<16 x i1> %mask)  {
allocas:
  %a= and <16 x i1> %mask, <i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false>
  %b = bitcast <16 x i1> %a to i16
  %c = icmp eq i16 %b, 0
  br i1 %c, label %true, label %false

true:
  ret void

false:
  ret void
}
define void @vcmp_test7(<8 x i1> %mask)  {
; GENERIC-LABEL: vcmp_test7:
; GENERIC:       # BB#0: # %allocas
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k0
; GENERIC-NEXT:    movb $85, %al # sched: [1:0.33]
; GENERIC-NEXT:    kmovd %eax, %k1
; GENERIC-NEXT:    korb %k1, %k0, %k0
; GENERIC-NEXT:    ktestb %k0, %k0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vcmp_test7:
; SKX:       # BB#0: # %allocas
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k0 # sched: [1:1.00]
; SKX-NEXT:    movb $85, %al # sched: [1:0.25]
; SKX-NEXT:    kmovd %eax, %k1 # sched: [1:1.00]
; SKX-NEXT:    korb %k1, %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    ktestb %k0, %k0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
allocas:
  %a= or <8 x i1> %mask, <i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false>
  %b = bitcast <8 x i1> %a to i8
  %c = icmp eq i8 %b, 0
  br i1 %c, label %true, label %false

true:
  ret void

false:
  ret void
}
define <16 x i8> @vcmp_test8(<16 x i32>%a, <16 x i32>%b, i32 %a1, i32 %b1) {
; GENERIC-LABEL: vcmp_test8:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; GENERIC-NEXT:    cmpl %esi, %edi # sched: [1:0.33]
; GENERIC-NEXT:    jg .LBB386_1 # sched: [1:1.00]
; GENERIC-NEXT:  # BB#2:
; GENERIC-NEXT:    vpcmpltud %zmm2, %zmm1, %k0
; GENERIC-NEXT:    vpmovm2b %k0, %xmm0
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
; GENERIC-NEXT:  .LBB386_1:
; GENERIC-NEXT:    vpcmpgtd %zmm2, %zmm0, %k0
; GENERIC-NEXT:    vpmovm2b %k0, %xmm0
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vcmp_test8:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    cmpl %esi, %edi # sched: [1:0.25]
; SKX-NEXT:    jg .LBB386_1 # sched: [1:0.50]
; SKX-NEXT:  # BB#2:
; SKX-NEXT:    vpcmpltud %zmm2, %zmm1, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2b %k0, %xmm0
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
; SKX-NEXT:  .LBB386_1:
; SKX-NEXT:    vpcmpgtd %zmm2, %zmm0, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2b %k0, %xmm0
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %cond = icmp sgt i32 %a1, %b1
  %cmp1 = icmp sgt <16 x i32> %a, zeroinitializer
  %cmp2 = icmp ult <16 x i32> %b, zeroinitializer
  %mix = select i1 %cond, <16 x i1> %cmp1, <16 x i1> %cmp2
  %res = sext <16 x i1> %mix to <16 x i8>
  ret <16 x i8> %res
}
define <16 x i1> @vpmov_test9(<16 x i1>%a, <16 x i1>%b, i32 %a1, i32 %b1) {
; GENERIC-LABEL: vpmov_test9:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    cmpl %esi, %edi # sched: [1:0.33]
; GENERIC-NEXT:    jg .LBB387_1 # sched: [1:1.00]
; GENERIC-NEXT:  # BB#2:
; GENERIC-NEXT:    vpsllw $7, %xmm1, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    jmp .LBB387_3 # sched: [1:1.00]
; GENERIC-NEXT:  .LBB387_1:
; GENERIC-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:  .LBB387_3:
; GENERIC-NEXT:    vpmovb2m %xmm0, %k0
; GENERIC-NEXT:    vpmovm2b %k0, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vpmov_test9:
; SKX:       # BB#0:
; SKX-NEXT:    cmpl %esi, %edi # sched: [1:0.25]
; SKX-NEXT:    jg .LBB387_1 # sched: [1:0.50]
; SKX-NEXT:  # BB#2:
; SKX-NEXT:    vpsllw $7, %xmm1, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    jmp .LBB387_3 # sched: [1:0.50]
; SKX-NEXT:  .LBB387_1:
; SKX-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:  .LBB387_3:
; SKX-NEXT:    vpmovb2m %xmm0, %k0 # sched: [1:1.00]
; SKX-NEXT:    vpmovm2b %k0, %xmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp sgt i32 %a1, %b1
  %c = select i1 %mask, <16 x i1>%a, <16 x i1>%b
  ret <16 x i1>%c
}define <8 x i1> @vpmov_test10(<8 x i1>%a, <8 x i1>%b, i32 %a1, i32 %b1) {
  %mask = icmp sgt i32 %a1, %b1
  %c = select i1 %mask, <8 x i1>%a, <8 x i1>%b
  ret <8 x i1>%c
}

define <4 x i1> @vmov_test11(<4 x i1>%a, <4 x i1>%b, i32 %a1, i32 %b1) {
; GENERIC-LABEL: vmov_test11:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    cmpl %esi, %edi # sched: [1:0.33]
; GENERIC-NEXT:    jg .LBB389_1 # sched: [1:1.00]
; GENERIC-NEXT:  # BB#2:
; GENERIC-NEXT:    vpslld $31, %xmm1, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    jmp .LBB389_3 # sched: [1:1.00]
; GENERIC-NEXT:  .LBB389_1:
; GENERIC-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:  .LBB389_3:
; GENERIC-NEXT:    vptestmd %xmm0, %xmm0, %k0
; GENERIC-NEXT:    vpmovm2d %k0, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vmov_test11:
; SKX:       # BB#0:
; SKX-NEXT:    cmpl %esi, %edi # sched: [1:0.25]
; SKX-NEXT:    jg .LBB389_1 # sched: [1:0.50]
; SKX-NEXT:  # BB#2:
; SKX-NEXT:    vpslld $31, %xmm1, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    jmp .LBB389_3 # sched: [1:0.50]
; SKX-NEXT:  .LBB389_1:
; SKX-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:  .LBB389_3:
; SKX-NEXT:    vptestmd %xmm0, %xmm0, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2d %k0, %xmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp sgt i32 %a1, %b1
  %c = select i1 %mask, <4 x i1>%a, <4 x i1>%b
  ret <4 x i1>%c
}

define i32 @vmov_test12(i32 %x, i32 %y)  {
; GENERIC-LABEL: vmov_test12:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    movl %edi, %eax # sched: [1:0.33]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vmov_test12:
; SKX:       # BB#0:
; SKX-NEXT:    movl %edi, %eax # sched: [1:0.25]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = bitcast i16 21845 to <16 x i1>
  %b = extractelement <16 x i1> %a, i32 0
  %c = select i1 %b, i32 %x, i32 %y
  ret i32 %c
}

define i32 @vmov_test13(i32 %x, i32 %y)  {
; GENERIC-LABEL: vmov_test13:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    movl %esi, %eax # sched: [1:0.33]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vmov_test13:
; SKX:       # BB#0:
; SKX-NEXT:    movl %esi, %eax # sched: [1:0.25]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = bitcast i16 21845 to <16 x i1>
  %b = extractelement <16 x i1> %a, i32 3
  %c = select i1 %b, i32 %x, i32 %y
  ret i32 %c
}define <4 x i1> @vmov_test14()  {
  %a = bitcast i16 21845 to <16 x i1>
  %b = extractelement <16 x i1> %a, i32 2
  %c = insertelement <4 x i1> <i1 true, i1 false, i1 false, i1 true>, i1 %b, i32 1
  ret <4 x i1> %c
}

define <16 x i1> @vmov_test15(i32 %x, i32 %y)  {
; GENERIC-LABEL: vmov_test15:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    cmpl %esi, %edi # sched: [1:0.33]
; GENERIC-NEXT:    movw $21845, %ax # imm = 0x5555
; GENERIC-NEXT:    # sched: [1:0.33]
; GENERIC-NEXT:    movw $1, %cx # sched: [1:0.33]
; GENERIC-NEXT:    cmovgw %ax, %cx # sched: [2:0.67]
; GENERIC-NEXT:    kmovd %ecx, %k0
; GENERIC-NEXT:    vpmovm2b %k0, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vmov_test15:
; SKX:       # BB#0:
; SKX-NEXT:    cmpl %esi, %edi # sched: [1:0.25]
; SKX-NEXT:    movw $21845, %ax # imm = 0x5555
; SKX-NEXT:    # sched: [1:0.25]
; SKX-NEXT:    movw $1, %cx # sched: [1:0.25]
; SKX-NEXT:    cmovgw %ax, %cx # sched: [1:0.50]
; SKX-NEXT:    kmovd %ecx, %k0 # sched: [1:1.00]
; SKX-NEXT:    vpmovm2b %k0, %xmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = bitcast i16 21845 to <16 x i1>
  %b = bitcast i16 1 to <16 x i1>
  %mask = icmp sgt i32 %x, %y
  %c = select i1 %mask, <16 x i1> %a, <16 x i1> %b
  ret <16 x i1> %c
}

define <64 x i8> @vmov_test16(i64 %x) {
;
; GENERIC-LABEL: vmov_test16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovq %rdi, %k0
; GENERIC-NEXT:    movb $1, %al # sched: [1:0.33]
; GENERIC-NEXT:    kmovd %eax, %k1
; GENERIC-NEXT:    vpmovm2b %k1, %zmm0
; GENERIC-NEXT:    vpsllq $40, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovm2b %k0, %zmm1
; GENERIC-NEXT:    movl $32, %eax # sched: [1:0.33]
; GENERIC-NEXT:    kmovd %eax, %k1
; GENERIC-NEXT:    vpblendmb %ymm0, %ymm1, %ymm0 {%k1}
; GENERIC-NEXT:    vshufi64x2 {{.*#+}} zmm0 = zmm0[0,1,2,3],zmm1[4,5,6,7]
; GENERIC-NEXT:    vpmovb2m %zmm0, %k0
; GENERIC-NEXT:    vpmovm2b %k0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vmov_test16:
; SKX:       # BB#0:
; SKX-NEXT:    kmovq %rdi, %k0 # sched: [1:1.00]
; SKX-NEXT:    movb $1, %al # sched: [1:0.25]
; SKX-NEXT:    kmovd %eax, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovm2b %k1, %zmm0
; SKX-NEXT:    vpsllq $40, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovm2b %k0, %zmm1
; SKX-NEXT:    movl $32, %eax # sched: [1:0.25]
; SKX-NEXT:    kmovd %eax, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpblendmb %ymm0, %ymm1, %ymm0 {%k1} # sched: [1:0.33]
; SKX-NEXT:    vshufi64x2 {{.*#+}} zmm0 = zmm0[0,1,2,3],zmm1[4,5,6,7] sched: [3:1.00]
; SKX-NEXT:    vpmovb2m %zmm0, %k0 # sched: [1:1.00]
; SKX-NEXT:    vpmovm2b %k0, %zmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = bitcast i64 %x to <64 x i1>
  %b = insertelement <64 x i1>%a, i1 true, i32 5
  %c = sext <64 x i1>%b to <64 x i8>
  ret <64 x i8>%c
}

define <64 x i8> @vmov_test17(i64 %x, i32 %y, i32 %z) {
;
; GENERIC-LABEL: vmov_test17:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovq %rdi, %k0
; GENERIC-NEXT:    cmpl %edx, %esi # sched: [1:0.33]
; GENERIC-NEXT:    setg %al # sched: [1:0.50]
; GENERIC-NEXT:    kmovd %eax, %k1
; GENERIC-NEXT:    vpmovm2b %k1, %zmm0
; GENERIC-NEXT:    vpsllq $40, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovm2b %k0, %zmm1
; GENERIC-NEXT:    movl $32, %eax # sched: [1:0.33]
; GENERIC-NEXT:    kmovd %eax, %k1
; GENERIC-NEXT:    vpblendmb %ymm0, %ymm1, %ymm0 {%k1}
; GENERIC-NEXT:    vshufi64x2 {{.*#+}} zmm0 = zmm0[0,1,2,3],zmm1[4,5,6,7]
; GENERIC-NEXT:    vpmovb2m %zmm0, %k0
; GENERIC-NEXT:    vpmovm2b %k0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vmov_test17:
; SKX:       # BB#0:
; SKX-NEXT:    kmovq %rdi, %k0 # sched: [1:1.00]
; SKX-NEXT:    cmpl %edx, %esi # sched: [1:0.25]
; SKX-NEXT:    setg %al # sched: [1:0.50]
; SKX-NEXT:    kmovd %eax, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpmovm2b %k1, %zmm0
; SKX-NEXT:    vpsllq $40, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovm2b %k0, %zmm1
; SKX-NEXT:    movl $32, %eax # sched: [1:0.25]
; SKX-NEXT:    kmovd %eax, %k1 # sched: [1:1.00]
; SKX-NEXT:    vpblendmb %ymm0, %ymm1, %ymm0 {%k1} # sched: [1:0.33]
; SKX-NEXT:    vshufi64x2 {{.*#+}} zmm0 = zmm0[0,1,2,3],zmm1[4,5,6,7] sched: [3:1.00]
; SKX-NEXT:    vpmovb2m %zmm0, %k0 # sched: [1:1.00]
; SKX-NEXT:    vpmovm2b %k0, %zmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = bitcast i64 %x to <64 x i1>
  %b = icmp sgt i32 %y, %z
  %c = insertelement <64 x i1>%a, i1 %b, i32 5
  %d = sext <64 x i1>%c to <64 x i8>
  ret <64 x i8>%d
}

define <8 x i1> @vmov_test18(i8 %a, i16 %y) {
; GENERIC-LABEL: vmov_test18:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k1
; GENERIC-NEXT:    kmovd %esi, %k2
; GENERIC-NEXT:    kshiftlw $7, %k2, %k0
; GENERIC-NEXT:    kshiftrw $15, %k0, %k0
; GENERIC-NEXT:    kshiftlw $6, %k2, %k2
; GENERIC-NEXT:    kshiftrw $15, %k2, %k2
; GENERIC-NEXT:    vpmovm2q %k1, %zmm0
; GENERIC-NEXT:    vpmovm2q %k2, %zmm1
; GENERIC-NEXT:    vmovdqa64 {{.*#+}} zmm2 = [0,1,2,3,4,5,8,7] sched: [4:0.50]
; GENERIC-NEXT:    vpermi2q %zmm1, %zmm0, %zmm2
; GENERIC-NEXT:    vpmovq2m %zmm2, %k1
; GENERIC-NEXT:    kshiftlb $1, %k1, %k1
; GENERIC-NEXT:    kshiftrb $1, %k1, %k1
; GENERIC-NEXT:    kshiftlb $7, %k0, %k0
; GENERIC-NEXT:    korb %k0, %k1, %k0
; GENERIC-NEXT:    vpmovm2w %k0, %xmm0
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vmov_test18:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k1 # sched: [1:1.00]
; SKX-NEXT:    kmovd %esi, %k2 # sched: [1:1.00]
; SKX-NEXT:    kshiftlw $7, %k2, %k0 # sched: [3:1.00]
; SKX-NEXT:    kshiftrw $15, %k0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kshiftlw $6, %k2, %k2 # sched: [3:1.00]
; SKX-NEXT:    kshiftrw $15, %k2, %k2 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2q %k1, %zmm0
; SKX-NEXT:    vpmovm2q %k2, %zmm1
; SKX-NEXT:    vmovdqa64 {{.*#+}} zmm2 = [0,1,2,3,4,5,8,7] sched: [8:0.50]
; SKX-NEXT:    vpermi2q %zmm1, %zmm0, %zmm2 # sched: [3:1.00]
; SKX-NEXT:    vpmovq2m %zmm2, %k1 # sched: [1:1.00]
; SKX-NEXT:    kshiftlb $1, %k1, %k1 # sched: [3:1.00]
; SKX-NEXT:    kshiftrb $1, %k1, %k1 # sched: [3:1.00]
; SKX-NEXT:    kshiftlb $7, %k0, %k0 # sched: [3:1.00]
; SKX-NEXT:    korb %k0, %k1, %k0 # sched: [1:1.00]
; SKX-NEXT:    vpmovm2w %k0, %xmm0
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = bitcast i8 %a to <8 x i1>
  %b1 = bitcast i16 %y to <16 x i1>
  %el1 = extractelement <16 x i1>%b1, i32 8
  %el2 = extractelement <16 x i1>%b1, i32 9
  %c = insertelement <8 x i1>%b, i1 %el1, i32 7
  %d = insertelement <8 x i1>%c, i1 %el2, i32 6
  ret <8 x i1>%d
}
define <32 x i16> @vmov_test21(<32 x i16> %x , <32 x i1> %mask) nounwind readnone {
; GENERIC-LABEL: vmov_test21:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %ymm1, %ymm1 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %ymm1, %k1
; GENERIC-NEXT:    vmovdqu16 %zmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vmov_test21:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %ymm1, %ymm1 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %ymm1, %k1 # sched: [1:1.00]
; SKX-NEXT:    vmovdqu16 %zmm0, %zmm0 {%k1} {z}
; SKX-NEXT:    retq # sched: [7:1.00]
  %ret = select <32 x i1> %mask, <32 x i16> %x, <32 x i16> zeroinitializer
  ret <32 x i16> %ret
}

define void @vmov_test22(<4 x i1> %a, <4 x i1>* %addr) {
; GENERIC-LABEL: vmov_test22:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm0, %xmm0, %k0
; GENERIC-NEXT:    kmovb %k0, (%rdi)
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vmov_test22:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm0, %xmm0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kmovb %k0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  store <4 x i1> %a, <4 x i1>* %addr
  ret void
}

define void @vmov_test23(<2 x i1> %a, <2 x i1>* %addr) {
; GENERIC-LABEL: vmov_test23:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmq %xmm0, %xmm0, %k0
; GENERIC-NEXT:    kmovb %k0, (%rdi)
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: vmov_test23:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmq %xmm0, %xmm0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kmovb %k0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  store <2 x i1> %a, <2 x i1>* %addr
  ret void
}

define void @store_v1i1(<1 x i1> %c , <1 x i1>* %ptr) {
; GENERIC-LABEL: store_v1i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k0
; GENERIC-NEXT:    kxnorw %k0, %k0, %k1
; GENERIC-NEXT:    kxorw %k1, %k0, %k0
; GENERIC-NEXT:    kmovb %k0, (%rsi)
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: store_v1i1:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k0 # sched: [1:1.00]
; SKX-NEXT:    kxnorw %k0, %k0, %k1 # sched: [1:1.00]
; SKX-NEXT:    kxorw %k1, %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovb %k0, (%rsi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = xor <1 x i1> %c, <i1 1>
  store <1 x i1> %x, <1 x i1>*  %ptr, align 4
  ret void
}

define void @store_v2i1(<2 x i1> %c , <2 x i1>* %ptr) {
; GENERIC-LABEL: store_v2i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmq %xmm0, %xmm0, %k0
; GENERIC-NEXT:    knotw %k0, %k0
; GENERIC-NEXT:    kmovb %k0, (%rdi)
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: store_v2i1:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllq $63, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmq %xmm0, %xmm0, %k0 # sched: [3:1.00]
; SKX-NEXT:    knotw %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovb %k0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = xor <2 x i1> %c, <i1 1, i1 1>
  store <2 x i1> %x, <2 x i1>*  %ptr, align 4
  ret void
}

define void @store_v4i1(<4 x i1> %c , <4 x i1>* %ptr) {
; GENERIC-LABEL: store_v4i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vptestmd %xmm0, %xmm0, %k0
; GENERIC-NEXT:    knotw %k0, %k0
; GENERIC-NEXT:    kmovb %k0, (%rdi)
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: store_v4i1:
; SKX:       # BB#0:
; SKX-NEXT:    vpslld $31, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vptestmd %xmm0, %xmm0, %k0 # sched: [3:1.00]
; SKX-NEXT:    knotw %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovb %k0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = xor <4 x i1> %c, <i1 1, i1 1, i1 1, i1 1>
  store <4 x i1> %x, <4 x i1>*  %ptr, align 4
  ret void
}

define void @store_v8i1(<8 x i1> %c , <8 x i1>* %ptr) {
; GENERIC-LABEL: store_v8i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k0
; GENERIC-NEXT:    knotb %k0, %k0
; GENERIC-NEXT:    kmovb %k0, (%rdi)
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: store_v8i1:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k0 # sched: [1:1.00]
; SKX-NEXT:    knotb %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovb %k0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = xor <8 x i1> %c, <i1 1, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1>
  store <8 x i1> %x, <8 x i1>*  %ptr, align 4
  ret void
}

define void @store_v16i1(<16 x i1> %c , <16 x i1>* %ptr) {
; GENERIC-LABEL: store_v16i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm0, %k0
; GENERIC-NEXT:    knotw %k0, %k0
; GENERIC-NEXT:    kmovw %k0, (%rdi)
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: store_v16i1:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm0, %k0 # sched: [1:1.00]
; SKX-NEXT:    knotw %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovw %k0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %x = xor <16 x i1> %c, <i1 1, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1, i1 1>
  store <16 x i1> %x, <16 x i1>*  %ptr, align 4
  ret void
}

;void f2(int);
;void f1(int c)
;{
;  static int v = 0;
;  if (v == 0)
;    v = 1;
;  else
;    v = 0;
;  f2(v);
;}

@f1.v = internal unnamed_addr global i1 false, align 4

define void @f1(i32 %c) {
; GENERIC-LABEL: f1:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    movzbl {{.*}}(%rip), %edi # sched: [5:0.50]
; GENERIC-NEXT:    xorl $1, %edi # sched: [1:0.33]
; GENERIC-NEXT:    movb %dil, {{.*}}(%rip) # sched: [5:1.00]
; GENERIC-NEXT:    jmp f2 # TAILCALL
;
; SKX-LABEL: f1:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    movzbl {{.*}}(%rip), %edi # sched: [5:0.50]
; SKX-NEXT:    xorl $1, %edi # sched: [1:0.25]
; SKX-NEXT:    movb %dil, {{.*}}(%rip) # sched: [1:1.00]
; SKX-NEXT:    jmp f2 # TAILCALL
entry:
  %.b1 = load i1, i1* @f1.v, align 4
  %not..b1 = xor i1 %.b1, true
  store i1 %not..b1, i1* @f1.v, align 4
  %0 = zext i1 %not..b1 to i32
  tail call void @f2(i32 %0) #2
  ret void
}

declare void @f2(i32) #1

define void @store_i16_i1(i16 %x, i1 *%y) {
; GENERIC-LABEL: store_i16_i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    andl $1, %edi # sched: [1:0.33]
; GENERIC-NEXT:    movb %dil, (%rsi) # sched: [5:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: store_i16_i1:
; SKX:       # BB#0:
; SKX-NEXT:    andl $1, %edi # sched: [1:0.25]
; SKX-NEXT:    movb %dil, (%rsi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %c = trunc i16 %x to i1
  store i1 %c, i1* %y
  ret void
}

define void @store_i8_i1(i8 %x, i1 *%y) {
; GENERIC-LABEL: store_i8_i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    andl $1, %edi # sched: [1:0.33]
; GENERIC-NEXT:    movb %dil, (%rsi) # sched: [5:1.00]
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: store_i8_i1:
; SKX:       # BB#0:
; SKX-NEXT:    andl $1, %edi # sched: [1:0.25]
; SKX-NEXT:    movb %dil, (%rsi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %c = trunc i8 %x to i1
  store i1 %c, i1* %y
  ret void
}

define <32 x i16> @test_build_vec_v32i1(<32 x i16> %x) {
; GENERIC-LABEL: test_build_vec_v32i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    movl $1497715861, %eax # imm = 0x59455495
; GENERIC-NEXT:    # sched: [1:0.33]
; GENERIC-NEXT:    kmovd %eax, %k1
; GENERIC-NEXT:    vmovdqu16 %zmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_build_vec_v32i1:
; SKX:       # BB#0:
; SKX-NEXT:    movl $1497715861, %eax # imm = 0x59455495
; SKX-NEXT:    # sched: [1:0.25]
; SKX-NEXT:    kmovd %eax, %k1 # sched: [1:1.00]
; SKX-NEXT:    vmovdqu16 %zmm0, %zmm0 {%k1} {z}
; SKX-NEXT:    retq # sched: [7:1.00]
  %ret = select <32 x i1> <i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 false, i1 true, i1 false, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 true, i1 false, i1 false, i1 true, i1 true, i1 false, i1 true, i1 false>, <32 x i16> %x, <32 x i16> zeroinitializer
  ret <32 x i16> %ret
}

define <64 x i8> @test_build_vec_v64i1(<64 x i8> %x) {
; GENERIC-LABEL: test_build_vec_v64i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpshufb {{.*#+}} zmm0 = zero,zero,zmm0[2],zero,zero,zero,zmm0[6],zero,zmm0[8],zero,zmm0[10],zero,zmm0[12],zero,zero,zmm0[15],zero,zero,zmm0[18],zero,zmm0[20],zero,zmm0[22],zero,zmm0[24],zero,zero,zmm0[27],zero,zero,zmm0[30],zero,zmm0[32],zero,zmm0[34],zero,zero,zero,zmm0[38],zero,zmm0[40],zero,zero,zmm0[43,44],zero,zmm0[46],zero,zmm0[48],zero,zmm0[50],zero,zero,zero,zmm0[54],zero,zmm0[56],zero,zero,zmm0[59,60],zero,zmm0[62],zero
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_build_vec_v64i1:
; SKX:       # BB#0:
; SKX-NEXT:    vpshufb {{.*#+}} zmm0 = zero,zero,zmm0[2],zero,zero,zero,zmm0[6],zero,zmm0[8],zero,zmm0[10],zero,zmm0[12],zero,zero,zmm0[15],zero,zero,zmm0[18],zero,zmm0[20],zero,zmm0[22],zero,zmm0[24],zero,zero,zmm0[27],zero,zero,zmm0[30],zero,zmm0[32],zero,zmm0[34],zero,zero,zero,zmm0[38],zero,zmm0[40],zero,zero,zmm0[43,44],zero,zmm0[46],zero,zmm0[48],zero,zmm0[50],zero,zero,zero,zmm0[54],zero,zmm0[56],zero,zero,zmm0[59,60],zero,zmm0[62],zero sched: [8:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %ret = select <64 x i1> <i1 false, i1 false, i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 false, i1 true, i1 false, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 false, i1 true, i1 false, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 true, i1 false, i1 false, i1 true, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 true, i1 false, i1 false, i1 false, i1 true, i1 false, i1 true, i1 false, i1 false, i1 true, i1 true, i1 false, i1 true, i1 false>, <64 x i8> %x, <64 x i8> zeroinitializer
  ret <64 x i8> %ret
}

define void @ktest_1(<8 x double> %in, double * %base) {
; GENERIC-LABEL: ktest_1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovupd (%rdi), %zmm1 # sched: [4:0.50]
; GENERIC-NEXT:    vcmpltpd %zmm0, %zmm1, %k1
; GENERIC-NEXT:    vmovupd 8(%rdi), %zmm1 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    vcmpltpd %zmm1, %zmm0, %k0 {%k1}
; GENERIC-NEXT:    ktestb %k0, %k0
; GENERIC-NEXT:    je .LBB410_2 # sched: [1:1.00]
; GENERIC-NEXT:  # BB#1: # %L1
; GENERIC-NEXT:    vmovapd %zmm0, (%rdi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
; GENERIC-NEXT:  .LBB410_2: # %L2
; GENERIC-NEXT:    vmovapd %zmm0, 8(%rdi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ktest_1:
; SKX:       # BB#0:
; SKX-NEXT:    vmovupd (%rdi), %zmm1 # sched: [8:0.50]
; SKX-NEXT:    vcmpltpd %zmm0, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vmovupd 8(%rdi), %zmm1 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    vcmpltpd %zmm1, %zmm0, %k0 {%k1} # sched: [3:1.00]
; SKX-NEXT:    ktestb %k0, %k0 # sched: [3:1.00]
; SKX-NEXT:    je .LBB410_2 # sched: [1:0.50]
; SKX-NEXT:  # BB#1: # %L1
; SKX-NEXT:    vmovapd %zmm0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
; SKX-NEXT:  .LBB410_2: # %L2
; SKX-NEXT:    vmovapd %zmm0, 8(%rdi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %addr1 = getelementptr double, double * %base, i64 0
  %addr2 = getelementptr double, double * %base, i64 1

  %vaddr1 = bitcast double* %addr1 to <8 x double>*
  %vaddr2 = bitcast double* %addr2 to <8 x double>*

  %val1 = load <8 x double>, <8 x double> *%vaddr1, align 1
  %val2 = load <8 x double>, <8 x double> *%vaddr2, align 1

  %sel1 = fcmp ogt <8 x double>%in, %val1
  %val3 = select <8 x i1> %sel1, <8 x double> %val2, <8 x double> zeroinitializer
  %sel2 = fcmp olt <8 x double> %in, %val3
  %sel3 = and <8 x i1> %sel1, %sel2

  %int_sel3 = bitcast <8 x i1> %sel3 to i8
  %res = icmp eq i8 %int_sel3, zeroinitializer
  br i1 %res, label %L2, label %L1
L1:
  store <8 x double> %in, <8 x double>* %vaddr1
  br label %End
L2:
  store <8 x double> %in, <8 x double>* %vaddr2
  br label %End
End:
  ret void
}

define void @ktest_2(<32 x float> %in, float * %base) {
;
; GENERIC-LABEL: ktest_2:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vmovups (%rdi), %zmm2 # sched: [4:0.50]
; GENERIC-NEXT:    vmovups 64(%rdi), %zmm3 # sched: [4:0.50]
; GENERIC-NEXT:    vcmpltps %zmm0, %zmm2, %k1
; GENERIC-NEXT:    vcmpltps %zmm1, %zmm3, %k2
; GENERIC-NEXT:    kunpckwd %k1, %k2, %k0
; GENERIC-NEXT:    vmovups 68(%rdi), %zmm2 {%k2} {z} # sched: [4:0.50]
; GENERIC-NEXT:    vmovups 4(%rdi), %zmm3 {%k1} {z} # sched: [4:0.50]
; GENERIC-NEXT:    vcmpltps %zmm3, %zmm0, %k1
; GENERIC-NEXT:    vcmpltps %zmm2, %zmm1, %k2
; GENERIC-NEXT:    kunpckwd %k1, %k2, %k1
; GENERIC-NEXT:    kord %k1, %k0, %k0
; GENERIC-NEXT:    ktestd %k0, %k0
; GENERIC-NEXT:    je .LBB411_2 # sched: [1:1.00]
; GENERIC-NEXT:  # BB#1: # %L1
; GENERIC-NEXT:    vmovaps %zmm0, (%rdi)
; GENERIC-NEXT:    vmovaps %zmm1, 64(%rdi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
; GENERIC-NEXT:  .LBB411_2: # %L2
; GENERIC-NEXT:    vmovaps %zmm0, 4(%rdi)
; GENERIC-NEXT:    vmovaps %zmm1, 68(%rdi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: ktest_2:
; SKX:       # BB#0:
; SKX-NEXT:    vmovups (%rdi), %zmm2 # sched: [8:0.50]
; SKX-NEXT:    vmovups 64(%rdi), %zmm3 # sched: [8:0.50]
; SKX-NEXT:    vcmpltps %zmm0, %zmm2, %k1 # sched: [3:1.00]
; SKX-NEXT:    vcmpltps %zmm1, %zmm3, %k2 # sched: [3:1.00]
; SKX-NEXT:    vmovups 68(%rdi), %zmm2 {%k2} {z} # sched: [8:0.50]
; SKX-NEXT:    vmovups 4(%rdi), %zmm3 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    kunpckwd %k1, %k2, %k0 # sched: [3:1.00]
; SKX-NEXT:    vcmpltps %zmm3, %zmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vcmpltps %zmm2, %zmm1, %k2 # sched: [3:1.00]
; SKX-NEXT:    kunpckwd %k1, %k2, %k1 # sched: [3:1.00]
; SKX-NEXT:    kord %k1, %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    ktestd %k0, %k0 # sched: [3:1.00]
; SKX-NEXT:    je .LBB411_2 # sched: [1:0.50]
; SKX-NEXT:  # BB#1: # %L1
; SKX-NEXT:    vmovaps %zmm0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    vmovaps %zmm1, 64(%rdi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
; SKX-NEXT:  .LBB411_2: # %L2
; SKX-NEXT:    vmovaps %zmm0, 4(%rdi) # sched: [1:1.00]
; SKX-NEXT:    vmovaps %zmm1, 68(%rdi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %addr1 = getelementptr float, float * %base, i64 0
  %addr2 = getelementptr float, float * %base, i64 1

  %vaddr1 = bitcast float* %addr1 to <32 x float>*
  %vaddr2 = bitcast float* %addr2 to <32 x float>*

  %val1 = load <32 x float>, <32 x float> *%vaddr1, align 1
  %val2 = load <32 x float>, <32 x float> *%vaddr2, align 1

  %sel1 = fcmp ogt <32 x float>%in, %val1
  %val3 = select <32 x i1> %sel1, <32 x float> %val2, <32 x float> zeroinitializer
  %sel2 = fcmp olt <32 x float> %in, %val3
  %sel3 = or <32 x i1> %sel1, %sel2

  %int_sel3 = bitcast <32 x i1> %sel3 to i32
  %res = icmp eq i32 %int_sel3, zeroinitializer
  br i1 %res, label %L2, label %L1
L1:
  store <32 x float> %in, <32 x float>* %vaddr1
  br label %End
L2:
  store <32 x float> %in, <32 x float>* %vaddr2
  br label %End
End:
  ret void
}

define <8 x i64> @load_8i1(<8 x i1>* %a) {
; GENERIC-LABEL: load_8i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovb (%rdi), %k0
; GENERIC-NEXT:    vpmovm2q %k0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: load_8i1:
; SKX:       # BB#0:
; SKX-NEXT:    kmovb (%rdi), %k0 # sched: [7:1.00]
; SKX-NEXT:    vpmovm2q %k0, %zmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = load <8 x i1>, <8 x i1>* %a
  %c = sext <8 x i1> %b to <8 x i64>
  ret <8 x i64> %c
}

define <16 x i32> @load_16i1(<16 x i1>* %a) {
; GENERIC-LABEL: load_16i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovw (%rdi), %k0
; GENERIC-NEXT:    vpmovm2d %k0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: load_16i1:
; SKX:       # BB#0:
; SKX-NEXT:    kmovw (%rdi), %k0 # sched: [7:1.00]
; SKX-NEXT:    vpmovm2d %k0, %zmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = load <16 x i1>, <16 x i1>* %a
  %c = sext <16 x i1> %b to <16 x i32>
  ret <16 x i32> %c
}

define <2 x i16> @load_2i1(<2 x i1>* %a) {
; GENERIC-LABEL: load_2i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovb (%rdi), %k0
; GENERIC-NEXT:    vpmovm2q %k0, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: load_2i1:
; SKX:       # BB#0:
; SKX-NEXT:    kmovb (%rdi), %k0 # sched: [7:1.00]
; SKX-NEXT:    vpmovm2q %k0, %xmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = load <2 x i1>, <2 x i1>* %a
  %c = sext <2 x i1> %b to <2 x i16>
  ret <2 x i16> %c
}

define <4 x i16> @load_4i1(<4 x i1>* %a) {
; GENERIC-LABEL: load_4i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovb (%rdi), %k0
; GENERIC-NEXT:    vpmovm2d %k0, %xmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: load_4i1:
; SKX:       # BB#0:
; SKX-NEXT:    kmovb (%rdi), %k0 # sched: [7:1.00]
; SKX-NEXT:    vpmovm2d %k0, %xmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = load <4 x i1>, <4 x i1>* %a
  %c = sext <4 x i1> %b to <4 x i16>
  ret <4 x i16> %c
}

define <32 x i16> @load_32i1(<32 x i1>* %a) {
; GENERIC-LABEL: load_32i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd (%rdi), %k0
; GENERIC-NEXT:    vpmovm2w %k0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: load_32i1:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd (%rdi), %k0 # sched: [7:1.00]
; SKX-NEXT:    vpmovm2w %k0, %zmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = load <32 x i1>, <32 x i1>* %a
  %c = sext <32 x i1> %b to <32 x i16>
  ret <32 x i16> %c
}

define <64 x i8> @load_64i1(<64 x i1>* %a) {
; GENERIC-LABEL: load_64i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovq (%rdi), %k0
; GENERIC-NEXT:    vpmovm2b %k0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: load_64i1:
; SKX:       # BB#0:
; SKX-NEXT:    kmovq (%rdi), %k0 # sched: [7:1.00]
; SKX-NEXT:    vpmovm2b %k0, %zmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = load <64 x i1>, <64 x i1>* %a
  %c = sext <64 x i1> %b to <64 x i8>
  ret <64 x i8> %c
}

define void @store_8i1(<8 x i1>* %a, <8 x i1> %v) {
; GENERIC-LABEL: store_8i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k0
; GENERIC-NEXT:    kmovb %k0, (%rdi)
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: store_8i1:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovb %k0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  store <8 x i1> %v, <8 x i1>* %a
  ret void
}

define void @store_8i1_1(<8 x i1>* %a, <8 x i16> %v) {
; GENERIC-LABEL: store_8i1_1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovw2m %xmm0, %k0
; GENERIC-NEXT:    kmovb %k0, (%rdi)
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: store_8i1_1:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %xmm0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovb %k0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %v1 = trunc <8 x i16> %v to <8 x i1>
  store <8 x i1> %v1, <8 x i1>* %a
  ret void
}

define void @store_16i1(<16 x i1>* %a, <16 x i1> %v) {
; GENERIC-LABEL: store_16i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %xmm0, %k0
; GENERIC-NEXT:    kmovw %k0, (%rdi)
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: store_16i1:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %xmm0, %xmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %xmm0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovw %k0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  store <16 x i1> %v, <16 x i1>* %a
  ret void
}

define void @store_32i1(<32 x i1>* %a, <32 x i1> %v) {
; GENERIC-LABEL: store_32i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %ymm0, %ymm0 # sched: [1:1.00]
; GENERIC-NEXT:    vpmovb2m %ymm0, %k0
; GENERIC-NEXT:    kmovd %k0, (%rdi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: store_32i1:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %ymm0, %ymm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %ymm0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %k0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  store <32 x i1> %v, <32 x i1>* %a
  ret void
}

define void @store_32i1_1(<32 x i1>* %a, <32 x i16> %v) {
; GENERIC-LABEL: store_32i1_1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $15, %zmm0, %zmm0
; GENERIC-NEXT:    vpmovw2m %zmm0, %k0
; GENERIC-NEXT:    kmovd %k0, (%rdi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: store_32i1_1:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $15, %zmm0, %zmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovw2m %zmm0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %k0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %v1 = trunc <32 x i16> %v to <32 x i1>
  store <32 x i1> %v1, <32 x i1>* %a
  ret void
}


define void @store_64i1(<64 x i1>* %a, <64 x i1> %v) {
;
; GENERIC-LABEL: store_64i1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpsllw $7, %zmm0, %zmm0
; GENERIC-NEXT:    vpmovb2m %zmm0, %k0
; GENERIC-NEXT:    kmovq %k0, (%rdi)
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: store_64i1:
; SKX:       # BB#0:
; SKX-NEXT:    vpsllw $7, %zmm0, %zmm0 # sched: [1:0.50]
; SKX-NEXT:    vpmovb2m %zmm0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovq %k0, (%rdi) # sched: [1:1.00]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  store <64 x i1> %v, <64 x i1>* %a
  ret void
}

define i32 @test_bitcast_v8i1_zext(<16 x i32> %a) {
; GENERIC-LABEL: test_bitcast_v8i1_zext:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpeqd %zmm1, %zmm0, %k0
; GENERIC-NEXT:    kmovb %k0, %eax
; GENERIC-NEXT:    addl %eax, %eax # sched: [1:0.33]
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_bitcast_v8i1_zext:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpcmpeqd %zmm1, %zmm0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kmovb %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    addl %eax, %eax # sched: [1:0.25]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
   %v1 = icmp eq <16 x i32> %a, zeroinitializer
   %mask = shufflevector <16 x i1> %v1, <16 x i1> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   %mask1 = bitcast <8 x i1> %mask to i8
   %val = zext i8 %mask1 to i32
   %val1 = add i32 %val, %val
   ret i32 %val1
}

define i32 @test_bitcast_v16i1_zext(<16 x i32> %a) {
; GENERIC-LABEL: test_bitcast_v16i1_zext:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpeqd %zmm1, %zmm0, %k0
; GENERIC-NEXT:    kmovw %k0, %eax
; GENERIC-NEXT:    addl %eax, %eax # sched: [1:0.33]
; GENERIC-NEXT:    vzeroupper
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_bitcast_v16i1_zext:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpcmpeqd %zmm1, %zmm0, %k0 # sched: [3:1.00]
; SKX-NEXT:    kmovw %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    addl %eax, %eax # sched: [1:0.25]
; SKX-NEXT:    vzeroupper # sched: [4:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
   %v1 = icmp eq <16 x i32> %a, zeroinitializer
   %mask1 = bitcast <16 x i1> %v1 to i16
   %val = zext i16 %mask1 to i32
   %val1 = add i32 %val, %val
   ret i32 %val1
}

define i16 @test_v16i1_add(i16 %x, i16 %y) {
; GENERIC-LABEL: test_v16i1_add:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k0
; GENERIC-NEXT:    kmovd %esi, %k1
; GENERIC-NEXT:    kxorw %k1, %k0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_v16i1_add:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %esi, %k1 # sched: [1:1.00]
; SKX-NEXT:    kxorw %k1, %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; SKX-NEXT:    retq # sched: [7:1.00]
  %m0 = bitcast i16 %x to <16 x i1>
  %m1 = bitcast i16 %y to <16 x i1>
  %m2 = add <16 x i1> %m0,  %m1
  %ret = bitcast <16 x i1> %m2 to i16
  ret i16 %ret
}

define i16 @test_v16i1_sub(i16 %x, i16 %y) {
; GENERIC-LABEL: test_v16i1_sub:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k0
; GENERIC-NEXT:    kmovd %esi, %k1
; GENERIC-NEXT:    kxorw %k1, %k0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_v16i1_sub:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %esi, %k1 # sched: [1:1.00]
; SKX-NEXT:    kxorw %k1, %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; SKX-NEXT:    retq # sched: [7:1.00]
  %m0 = bitcast i16 %x to <16 x i1>
  %m1 = bitcast i16 %y to <16 x i1>
  %m2 = sub <16 x i1> %m0,  %m1
  %ret = bitcast <16 x i1> %m2 to i16
  ret i16 %ret
}

define i16 @test_v16i1_mul(i16 %x, i16 %y) {
; GENERIC-LABEL: test_v16i1_mul:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k0
; GENERIC-NEXT:    kmovd %esi, %k1
; GENERIC-NEXT:    kandw %k1, %k0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_v16i1_mul:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %esi, %k1 # sched: [1:1.00]
; SKX-NEXT:    kandw %k1, %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    # kill: %AX<def> %AX<kill> %EAX<kill>
; SKX-NEXT:    retq # sched: [7:1.00]
  %m0 = bitcast i16 %x to <16 x i1>
  %m1 = bitcast i16 %y to <16 x i1>
  %m2 = mul <16 x i1> %m0,  %m1
  %ret = bitcast <16 x i1> %m2 to i16
  ret i16 %ret
}

define i8 @test_v8i1_add(i8 %x, i8 %y) {
; GENERIC-LABEL: test_v8i1_add:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k0
; GENERIC-NEXT:    kmovd %esi, %k1
; GENERIC-NEXT:    kxorb %k1, %k0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    # kill: %AL<def> %AL<kill> %EAX<kill>
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_v8i1_add:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %esi, %k1 # sched: [1:1.00]
; SKX-NEXT:    kxorb %k1, %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    # kill: %AL<def> %AL<kill> %EAX<kill>
; SKX-NEXT:    retq # sched: [7:1.00]
  %m0 = bitcast i8 %x to <8 x i1>
  %m1 = bitcast i8 %y to <8 x i1>
  %m2 = add <8 x i1> %m0,  %m1
  %ret = bitcast <8 x i1> %m2 to i8
  ret i8 %ret
}

define i8 @test_v8i1_sub(i8 %x, i8 %y) {
; GENERIC-LABEL: test_v8i1_sub:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k0
; GENERIC-NEXT:    kmovd %esi, %k1
; GENERIC-NEXT:    kxorb %k1, %k0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    # kill: %AL<def> %AL<kill> %EAX<kill>
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_v8i1_sub:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %esi, %k1 # sched: [1:1.00]
; SKX-NEXT:    kxorb %k1, %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    # kill: %AL<def> %AL<kill> %EAX<kill>
; SKX-NEXT:    retq # sched: [7:1.00]
  %m0 = bitcast i8 %x to <8 x i1>
  %m1 = bitcast i8 %y to <8 x i1>
  %m2 = sub <8 x i1> %m0,  %m1
  %ret = bitcast <8 x i1> %m2 to i8
  ret i8 %ret
}

define i8 @test_v8i1_mul(i8 %x, i8 %y) {
; GENERIC-LABEL: test_v8i1_mul:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    kmovd %edi, %k0
; GENERIC-NEXT:    kmovd %esi, %k1
; GENERIC-NEXT:    kandb %k1, %k0, %k0
; GENERIC-NEXT:    kmovd %k0, %eax
; GENERIC-NEXT:    # kill: %AL<def> %AL<kill> %EAX<kill>
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_v8i1_mul:
; SKX:       # BB#0:
; SKX-NEXT:    kmovd %edi, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %esi, %k1 # sched: [1:1.00]
; SKX-NEXT:    kandb %k1, %k0, %k0 # sched: [1:1.00]
; SKX-NEXT:    kmovd %k0, %eax # sched: [3:1.00]
; SKX-NEXT:    # kill: %AL<def> %AL<kill> %EAX<kill>
; SKX-NEXT:    retq # sched: [7:1.00]
  %m0 = bitcast i8 %x to <8 x i1>
  %m1 = bitcast i8 %y to <8 x i1>
  %m2 = mul <8 x i1> %m0,  %m1
  %ret = bitcast <8 x i1> %m2 to i8
  ret i8 %ret
}

define   <16 x i32> @_inreg16xi32(i32 %a) {
; GENERIC-LABEL: _inreg16xi32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpbroadcastd %edi, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _inreg16xi32:
; SKX:       # BB#0:
; SKX-NEXT:    vpbroadcastd %edi, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = insertelement <16 x i32> undef, i32 %a, i32 0
  %c = shufflevector <16 x i32> %b, <16 x i32> undef, <16 x i32> zeroinitializer
  ret <16 x i32> %c
}

define   <8 x i64> @_inreg8xi64(i64 %a) {
; GENERIC-LABEL: _inreg8xi64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpbroadcastq %rdi, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _inreg8xi64:
; SKX:       # BB#0:
; SKX-NEXT:    vpbroadcastq %rdi, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = insertelement <8 x i64> undef, i64 %a, i32 0
  %c = shufflevector <8 x i64> %b, <8 x i64> undef, <8 x i32> zeroinitializer
  ret <8 x i64> %c
}

define   <16 x float> @_ss16xfloat_v4(<4 x float> %a) {
; GENERIC-LABEL: _ss16xfloat_v4:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vbroadcastss %xmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _ss16xfloat_v4:
; SKX:       # BB#0:
; SKX-NEXT:    vbroadcastss %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = shufflevector <4 x float> %a, <4 x float> undef, <16 x i32> zeroinitializer
  ret <16 x float> %b
}

define   <16 x float> @_inreg16xfloat(float %a) {
; GENERIC-LABEL: _inreg16xfloat:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vbroadcastss %xmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _inreg16xfloat:
; SKX:       # BB#0:
; SKX-NEXT:    vbroadcastss %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = insertelement <16 x float> undef, float %a, i32 0
  %c = shufflevector <16 x float> %b, <16 x float> undef, <16 x i32> zeroinitializer
  ret <16 x float> %c
}

define   <16 x float> @_ss16xfloat_mask(float %a, <16 x float> %i, <16 x i32> %mask1) {
; GENERIC-LABEL: _ss16xfloat_mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm3, %xmm3, %xmm3 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm3, %zmm2, %k1
; GENERIC-NEXT:    vbroadcastss %xmm0, %zmm1 {%k1}
; GENERIC-NEXT:    vmovaps %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _ss16xfloat_mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm3, %xmm3, %xmm3 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm3, %zmm2, %k1 # sched: [3:1.00]
; SKX-NEXT:    vbroadcastss %xmm0, %zmm1 {%k1} # sched: [3:1.00]
; SKX-NEXT:    vmovaps %zmm1, %zmm0 # sched: [1:0.33]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %b = insertelement <16 x float> undef, float %a, i32 0
  %c = shufflevector <16 x float> %b, <16 x float> undef, <16 x i32> zeroinitializer
  %r = select <16 x i1> %mask, <16 x float> %c, <16 x float> %i
  ret <16 x float> %r
}

define   <16 x float> @_ss16xfloat_maskz(float %a, <16 x i32> %mask1) {
; GENERIC-LABEL: _ss16xfloat_maskz:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vbroadcastss %xmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _ss16xfloat_maskz:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vbroadcastss %xmm0, %zmm0 {%k1} {z} # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %b = insertelement <16 x float> undef, float %a, i32 0
  %c = shufflevector <16 x float> %b, <16 x float> undef, <16 x i32> zeroinitializer
  %r = select <16 x i1> %mask, <16 x float> %c, <16 x float> zeroinitializer
  ret <16 x float> %r
}

define   <16 x float> @_ss16xfloat_load(float* %a.ptr) {
; GENERIC-LABEL: _ss16xfloat_load:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vbroadcastss (%rdi), %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _ss16xfloat_load:
; SKX:       # BB#0:
; SKX-NEXT:    vbroadcastss (%rdi), %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = load float, float* %a.ptr
  %b = insertelement <16 x float> undef, float %a, i32 0
  %c = shufflevector <16 x float> %b, <16 x float> undef, <16 x i32> zeroinitializer
  ret <16 x float> %c
}

define   <16 x float> @_ss16xfloat_mask_load(float* %a.ptr, <16 x float> %i, <16 x i32> %mask1) {
; GENERIC-LABEL: _ss16xfloat_mask_load:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1
; GENERIC-NEXT:    vbroadcastss (%rdi), %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _ss16xfloat_mask_load:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm2, %zmm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vbroadcastss (%rdi), %zmm0 {%k1} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = load float, float* %a.ptr
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %b = insertelement <16 x float> undef, float %a, i32 0
  %c = shufflevector <16 x float> %b, <16 x float> undef, <16 x i32> zeroinitializer
  %r = select <16 x i1> %mask, <16 x float> %c, <16 x float> %i
  ret <16 x float> %r
}

define   <16 x float> @_ss16xfloat_maskz_load(float* %a.ptr, <16 x i32> %mask1) {
; GENERIC-LABEL: _ss16xfloat_maskz_load:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %zmm1, %zmm0, %k1
; GENERIC-NEXT:    vbroadcastss (%rdi), %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _ss16xfloat_maskz_load:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %zmm1, %zmm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vbroadcastss (%rdi), %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = load float, float* %a.ptr
  %mask = icmp ne <16 x i32> %mask1, zeroinitializer
  %b = insertelement <16 x float> undef, float %a, i32 0
  %c = shufflevector <16 x float> %b, <16 x float> undef, <16 x i32> zeroinitializer
  %r = select <16 x i1> %mask, <16 x float> %c, <16 x float> zeroinitializer
  ret <16 x float> %r
}

define   <8 x double> @_inreg8xdouble(double %a) {
; GENERIC-LABEL: _inreg8xdouble:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vbroadcastsd %xmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _inreg8xdouble:
; SKX:       # BB#0:
; SKX-NEXT:    vbroadcastsd %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = insertelement <8 x double> undef, double %a, i32 0
  %c = shufflevector <8 x double> %b, <8 x double> undef, <8 x i32> zeroinitializer
  ret <8 x double> %c
}

define   <8 x double> @_sd8xdouble_mask(double %a, <8 x double> %i, <8 x i32> %mask1) {
; GENERIC-LABEL: _sd8xdouble_mask:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm3, %xmm3, %xmm3 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %ymm3, %ymm2, %k1
; GENERIC-NEXT:    vbroadcastsd %xmm0, %zmm1 {%k1}
; GENERIC-NEXT:    vmovapd %zmm1, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _sd8xdouble_mask:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm3, %xmm3, %xmm3 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %ymm3, %ymm2, %k1 # sched: [3:1.00]
; SKX-NEXT:    vbroadcastsd %xmm0, %zmm1 {%k1} # sched: [3:1.00]
; SKX-NEXT:    vmovapd %zmm1, %zmm0
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %b = insertelement <8 x double> undef, double %a, i32 0
  %c = shufflevector <8 x double> %b, <8 x double> undef, <8 x i32> zeroinitializer
  %r = select <8 x i1> %mask, <8 x double> %c, <8 x double> %i
  ret <8 x double> %r
}

define   <8 x double> @_sd8xdouble_maskz(double %a, <8 x i32> %mask1) {
; GENERIC-LABEL: _sd8xdouble_maskz:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %ymm2, %ymm1, %k1
; GENERIC-NEXT:    vbroadcastsd %xmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _sd8xdouble_maskz:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %ymm2, %ymm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vbroadcastsd %xmm0, %zmm0 {%k1} {z} # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %b = insertelement <8 x double> undef, double %a, i32 0
  %c = shufflevector <8 x double> %b, <8 x double> undef, <8 x i32> zeroinitializer
  %r = select <8 x i1> %mask, <8 x double> %c, <8 x double> zeroinitializer
  ret <8 x double> %r
}

define   <8 x double> @_sd8xdouble_load(double* %a.ptr) {
; GENERIC-LABEL: _sd8xdouble_load:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vbroadcastsd (%rdi), %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _sd8xdouble_load:
; SKX:       # BB#0:
; SKX-NEXT:    vbroadcastsd (%rdi), %zmm0 # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = load double, double* %a.ptr
  %b = insertelement <8 x double> undef, double %a, i32 0
  %c = shufflevector <8 x double> %b, <8 x double> undef, <8 x i32> zeroinitializer
  ret <8 x double> %c
}

define   <8 x double> @_sd8xdouble_mask_load(double* %a.ptr, <8 x double> %i, <8 x i32> %mask1) {
; GENERIC-LABEL: _sd8xdouble_mask_load:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %ymm2, %ymm1, %k1
; GENERIC-NEXT:    vbroadcastsd (%rdi), %zmm0 {%k1}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _sd8xdouble_mask_load:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %ymm2, %ymm1, %k1 # sched: [3:1.00]
; SKX-NEXT:    vbroadcastsd (%rdi), %zmm0 {%k1} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = load double, double* %a.ptr
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %b = insertelement <8 x double> undef, double %a, i32 0
  %c = shufflevector <8 x double> %b, <8 x double> undef, <8 x i32> zeroinitializer
  %r = select <8 x i1> %mask, <8 x double> %c, <8 x double> %i
  ret <8 x double> %r
}

define   <8 x double> @_sd8xdouble_maskz_load(double* %a.ptr, <8 x i32> %mask1) {
; GENERIC-LABEL: _sd8xdouble_maskz_load:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; GENERIC-NEXT:    vpcmpneqd %ymm1, %ymm0, %k1
; GENERIC-NEXT:    vbroadcastsd (%rdi), %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _sd8xdouble_maskz_load:
; SKX:       # BB#0:
; SKX-NEXT:    vpxor %xmm1, %xmm1, %xmm1 # sched: [1:0.33]
; SKX-NEXT:    vpcmpneqd %ymm1, %ymm0, %k1 # sched: [3:1.00]
; SKX-NEXT:    vbroadcastsd (%rdi), %zmm0 {%k1} {z} # sched: [8:0.50]
; SKX-NEXT:    retq # sched: [7:1.00]
  %a = load double, double* %a.ptr
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %b = insertelement <8 x double> undef, double %a, i32 0
  %c = shufflevector <8 x double> %b, <8 x double> undef, <8 x i32> zeroinitializer
  %r = select <8 x i1> %mask, <8 x double> %c, <8 x double> zeroinitializer
  ret <8 x double> %r
}

define   <16 x i32> @_xmm16xi32(<16 x i32> %a) {
; GENERIC-LABEL: _xmm16xi32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vbroadcastss %xmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _xmm16xi32:
; SKX:       # BB#0:
; SKX-NEXT:    vbroadcastss %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = shufflevector <16 x i32> %a, <16 x i32> undef, <16 x i32> zeroinitializer
  ret <16 x i32> %b
}

define   <16 x float> @_xmm16xfloat(<16 x float> %a) {
; GENERIC-LABEL: _xmm16xfloat:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vbroadcastss %xmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _xmm16xfloat:
; SKX:       # BB#0:
; SKX-NEXT:    vbroadcastss %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %b = shufflevector <16 x float> %a, <16 x float> undef, <16 x i32> zeroinitializer
  ret <16 x float> %b
}

define <16 x i32> @test_vbroadcast() {
; GENERIC-LABEL: test_vbroadcast:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vxorps %xmm0, %xmm0, %xmm0 # sched: [1:1.00]
; GENERIC-NEXT:    vcmpunordps %zmm0, %zmm0, %k0
; GENERIC-NEXT:    vpmovm2d %k0, %zmm0
; GENERIC-NEXT:    knotw %k0, %k1
; GENERIC-NEXT:    vmovdqa32 %zmm0, %zmm0 {%k1} {z}
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_vbroadcast:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vxorps %xmm0, %xmm0, %xmm0 # sched: [1:0.33]
; SKX-NEXT:    vcmpunordps %zmm0, %zmm0, %k0 # sched: [3:1.00]
; SKX-NEXT:    vpmovm2d %k0, %zmm0
; SKX-NEXT:    knotw %k0, %k1 # sched: [1:1.00]
; SKX-NEXT:    vmovdqa32 %zmm0, %zmm0 {%k1} {z}
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = sext <16 x i1> zeroinitializer to <16 x i32>
  %1 = fcmp uno <16 x float> undef, zeroinitializer
  %2 = sext <16 x i1> %1 to <16 x i32>
  %3 = select <16 x i1> %1, <16 x i32> %0, <16 x i32> %2
  ret <16 x i32> %3
}

; We implement the set1 intrinsics with vector initializers.  Verify that the
; IR generated will produce broadcasts at the end.
define <8 x double> @test_set1_pd(double %d) #2 {
; GENERIC-LABEL: test_set1_pd:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vbroadcastsd %xmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_set1_pd:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vbroadcastsd %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %vecinit.i = insertelement <8 x double> undef, double %d, i32 0
  %vecinit1.i = insertelement <8 x double> %vecinit.i, double %d, i32 1
  %vecinit2.i = insertelement <8 x double> %vecinit1.i, double %d, i32 2
  %vecinit3.i = insertelement <8 x double> %vecinit2.i, double %d, i32 3
  %vecinit4.i = insertelement <8 x double> %vecinit3.i, double %d, i32 4
  %vecinit5.i = insertelement <8 x double> %vecinit4.i, double %d, i32 5
  %vecinit6.i = insertelement <8 x double> %vecinit5.i, double %d, i32 6
  %vecinit7.i = insertelement <8 x double> %vecinit6.i, double %d, i32 7
  ret <8 x double> %vecinit7.i
}

define <8 x i64> @test_set1_epi64(i64 %d) #2 {
; GENERIC-LABEL: test_set1_epi64:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vpbroadcastq %rdi, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_set1_epi64:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vpbroadcastq %rdi, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %vecinit.i = insertelement <8 x i64> undef, i64 %d, i32 0
  %vecinit1.i = insertelement <8 x i64> %vecinit.i, i64 %d, i32 1
  %vecinit2.i = insertelement <8 x i64> %vecinit1.i, i64 %d, i32 2
  %vecinit3.i = insertelement <8 x i64> %vecinit2.i, i64 %d, i32 3
  %vecinit4.i = insertelement <8 x i64> %vecinit3.i, i64 %d, i32 4
  %vecinit5.i = insertelement <8 x i64> %vecinit4.i, i64 %d, i32 5
  %vecinit6.i = insertelement <8 x i64> %vecinit5.i, i64 %d, i32 6
  %vecinit7.i = insertelement <8 x i64> %vecinit6.i, i64 %d, i32 7
  ret <8 x i64> %vecinit7.i
}

define <16 x float> @test_set1_ps(float %f) #2 {
; GENERIC-LABEL: test_set1_ps:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vbroadcastss %xmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_set1_ps:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vbroadcastss %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %vecinit.i = insertelement <16 x float> undef, float %f, i32 0
  %vecinit1.i = insertelement <16 x float> %vecinit.i, float %f, i32 1
  %vecinit2.i = insertelement <16 x float> %vecinit1.i, float %f, i32 2
  %vecinit3.i = insertelement <16 x float> %vecinit2.i, float %f, i32 3
  %vecinit4.i = insertelement <16 x float> %vecinit3.i, float %f, i32 4
  %vecinit5.i = insertelement <16 x float> %vecinit4.i, float %f, i32 5
  %vecinit6.i = insertelement <16 x float> %vecinit5.i, float %f, i32 6
  %vecinit7.i = insertelement <16 x float> %vecinit6.i, float %f, i32 7
  %vecinit8.i = insertelement <16 x float> %vecinit7.i, float %f, i32 8
  %vecinit9.i = insertelement <16 x float> %vecinit8.i, float %f, i32 9
  %vecinit10.i = insertelement <16 x float> %vecinit9.i, float %f, i32 10
  %vecinit11.i = insertelement <16 x float> %vecinit10.i, float %f, i32 11
  %vecinit12.i = insertelement <16 x float> %vecinit11.i, float %f, i32 12
  %vecinit13.i = insertelement <16 x float> %vecinit12.i, float %f, i32 13
  %vecinit14.i = insertelement <16 x float> %vecinit13.i, float %f, i32 14
  %vecinit15.i = insertelement <16 x float> %vecinit14.i, float %f, i32 15
  ret <16 x float> %vecinit15.i
}

define <16 x i32> @test_set1_epi32(i32 %f) #2 {
; GENERIC-LABEL: test_set1_epi32:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vpbroadcastd %edi, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_set1_epi32:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vpbroadcastd %edi, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %vecinit.i = insertelement <16 x i32> undef, i32 %f, i32 0
  %vecinit1.i = insertelement <16 x i32> %vecinit.i, i32 %f, i32 1
  %vecinit2.i = insertelement <16 x i32> %vecinit1.i, i32 %f, i32 2
  %vecinit3.i = insertelement <16 x i32> %vecinit2.i, i32 %f, i32 3
  %vecinit4.i = insertelement <16 x i32> %vecinit3.i, i32 %f, i32 4
  %vecinit5.i = insertelement <16 x i32> %vecinit4.i, i32 %f, i32 5
  %vecinit6.i = insertelement <16 x i32> %vecinit5.i, i32 %f, i32 6
  %vecinit7.i = insertelement <16 x i32> %vecinit6.i, i32 %f, i32 7
  %vecinit8.i = insertelement <16 x i32> %vecinit7.i, i32 %f, i32 8
  %vecinit9.i = insertelement <16 x i32> %vecinit8.i, i32 %f, i32 9
  %vecinit10.i = insertelement <16 x i32> %vecinit9.i, i32 %f, i32 10
  %vecinit11.i = insertelement <16 x i32> %vecinit10.i, i32 %f, i32 11
  %vecinit12.i = insertelement <16 x i32> %vecinit11.i, i32 %f, i32 12
  %vecinit13.i = insertelement <16 x i32> %vecinit12.i, i32 %f, i32 13
  %vecinit14.i = insertelement <16 x i32> %vecinit13.i, i32 %f, i32 14
  %vecinit15.i = insertelement <16 x i32> %vecinit14.i, i32 %f, i32 15
  ret <16 x i32> %vecinit15.i
}

; We implement the scalar broadcast intrinsics with vector initializers.
; Verify that the IR generated will produce the broadcast at the end.
define <8 x double> @test_mm512_broadcastsd_pd(<2 x double> %a) {
; GENERIC-LABEL: test_mm512_broadcastsd_pd:
; GENERIC:       # BB#0: # %entry
; GENERIC-NEXT:    vbroadcastsd %xmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: test_mm512_broadcastsd_pd:
; SKX:       # BB#0: # %entry
; SKX-NEXT:    vbroadcastsd %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
entry:
  %0 = extractelement <2 x double> %a, i32 0
  %vecinit.i = insertelement <8 x double> undef, double %0, i32 0
  %vecinit1.i = insertelement <8 x double> %vecinit.i, double %0, i32 1
  %vecinit2.i = insertelement <8 x double> %vecinit1.i, double %0, i32 2
  %vecinit3.i = insertelement <8 x double> %vecinit2.i, double %0, i32 3
  %vecinit4.i = insertelement <8 x double> %vecinit3.i, double %0, i32 4
  %vecinit5.i = insertelement <8 x double> %vecinit4.i, double %0, i32 5
  %vecinit6.i = insertelement <8 x double> %vecinit5.i, double %0, i32 6
  %vecinit7.i = insertelement <8 x double> %vecinit6.i, double %0, i32 7
  ret <8 x double> %vecinit7.i
}

define <16 x float> @suff_test1(<8 x float>%a)  {
; GENERIC-LABEL: suff_test1:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vbroadcastss %xmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: suff_test1:
; SKX:       # BB#0:
; SKX-NEXT:    vbroadcastss %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = shufflevector <8 x float> %a, <8 x float> undef, <16 x i32> zeroinitializer
  ret <16 x float>%res
}

define <8 x double> @suff_test2(<4 x double>%a)  {
; GENERIC-LABEL: suff_test2:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vbroadcastsd %xmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: suff_test2:
; SKX:       # BB#0:
; SKX-NEXT:    vbroadcastsd %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = shufflevector <4 x double> %a, <4 x double> undef, <8 x i32> zeroinitializer
  ret <8 x double>%res
}

define <64 x i8> @_invec32xi8(<32 x i8>%a)  {
; GENERIC-LABEL: _invec32xi8:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpbroadcastb %xmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _invec32xi8:
; SKX:       # BB#0:
; SKX-NEXT:    vpbroadcastb %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = shufflevector <32 x i8> %a, <32 x i8> undef, <64 x i32> zeroinitializer
  ret <64 x i8>%res
}

define <32 x i16> @_invec16xi16(<16 x i16>%a)  {
; GENERIC-LABEL: _invec16xi16:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vpbroadcastw %xmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _invec16xi16:
; SKX:       # BB#0:
; SKX-NEXT:    vpbroadcastw %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = shufflevector <16 x i16> %a, <16 x i16> undef, <32 x i32> zeroinitializer
  ret <32 x i16>%res
}

define <16 x i32> @_invec8xi32(<8 x i32>%a)  {
; GENERIC-LABEL: _invec8xi32:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vbroadcastss %xmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _invec8xi32:
; SKX:       # BB#0:
; SKX-NEXT:    vbroadcastss %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = shufflevector <8 x i32> %a, <8 x i32> undef, <16 x i32> zeroinitializer
  ret <16 x i32>%res
}

define <8 x i64> @_invec4xi64(<4 x i64>%a)  {
; GENERIC-LABEL: _invec4xi64:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    vbroadcastsd %xmm0, %zmm0
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: _invec4xi64:
; SKX:       # BB#0:
; SKX-NEXT:    vbroadcastsd %xmm0, %zmm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
  %res = shufflevector <4 x i64> %a, <4 x i64> undef, <8 x i32> zeroinitializer
  ret <8 x i64>%res
}

declare void @func_f32(float)
define <16 x float> @broadcast_ss_spill(float %x) {
; GENERIC-LABEL: broadcast_ss_spill:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    subq $24, %rsp # sched: [1:0.33]
; GENERIC-NEXT:    .cfi_def_cfa_offset 32
; GENERIC-NEXT:    vaddss %xmm0, %xmm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill sched: [5:1.00]
; GENERIC-NEXT:    # sched: [5:1.00]
; GENERIC-NEXT:    callq func_f32
; GENERIC-NEXT:    vbroadcastss (%rsp), %zmm0 # 16-byte Folded Reload
; GENERIC-NEXT:    addq $24, %rsp # sched: [1:0.33]
; GENERIC-NEXT:    .cfi_def_cfa_offset 8
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: broadcast_ss_spill:
; SKX:       # BB#0:
; SKX-NEXT:    subq $24, %rsp # sched: [1:0.25]
; SKX-NEXT:    .cfi_def_cfa_offset 32
; SKX-NEXT:    vaddss %xmm0, %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill sched: [1:1.00]
; SKX-NEXT:    # sched: [1:1.00]
; SKX-NEXT:    callq func_f32
; SKX-NEXT:    vbroadcastss (%rsp), %zmm0 # 16-byte Folded Reload sched: [8:0.50]
; SKX-NEXT:    # sched: [8:0.50]
; SKX-NEXT:    addq $24, %rsp # sched: [1:0.25]
; SKX-NEXT:    .cfi_def_cfa_offset 8
; SKX-NEXT:    retq # sched: [7:1.00]
  %a  = fadd float %x, %x
  call void @func_f32(float %a)
  %b = insertelement <16 x float> undef, float %a, i32 0
  %c = shufflevector <16 x float> %b, <16 x float> undef, <16 x i32> zeroinitializer
  ret <16 x float> %c
}

declare void @func_f64(double)
define <8 x double> @broadcast_sd_spill(double %x) {
; GENERIC-LABEL: broadcast_sd_spill:
; GENERIC:       # BB#0:
; GENERIC-NEXT:    subq $24, %rsp # sched: [1:0.33]
; GENERIC-NEXT:    .cfi_def_cfa_offset 32
; GENERIC-NEXT:    vaddsd %xmm0, %xmm0, %xmm0 # sched: [3:1.00]
; GENERIC-NEXT:    vmovapd %xmm0, (%rsp) # 16-byte Spill sched: [5:1.00]
; GENERIC-NEXT:    # sched: [5:1.00]
; GENERIC-NEXT:    callq func_f64
; GENERIC-NEXT:    vbroadcastsd (%rsp), %zmm0 # 16-byte Folded Reload
; GENERIC-NEXT:    addq $24, %rsp # sched: [1:0.33]
; GENERIC-NEXT:    .cfi_def_cfa_offset 8
; GENERIC-NEXT:    retq # sched: [1:1.00]
;
; SKX-LABEL: broadcast_sd_spill:
; SKX:       # BB#0:
; SKX-NEXT:    subq $24, %rsp # sched: [1:0.25]
; SKX-NEXT:    .cfi_def_cfa_offset 32
; SKX-NEXT:    vaddsd %xmm0, %xmm0, %xmm0 # sched: [4:0.33]
; SKX-NEXT:    vmovapd %xmm0, (%rsp) # 16-byte Spill sched: [1:1.00]
; SKX-NEXT:    # sched: [1:1.00]
; SKX-NEXT:    callq func_f64
; SKX-NEXT:    vbroadcastsd (%rsp), %zmm0 # 16-byte Folded Reload sched: [8:0.50]
; SKX-NEXT:    # sched: [8:0.50]
; SKX-NEXT:    addq $24, %rsp # sched: [1:0.25]
; SKX-NEXT:    .cfi_def_cfa_offset 8
; SKX-NEXT:    retq # sched: [7:1.00]
  %a  = fadd double %x, %x
  call void @func_f64(double %a)
  %b = insertelement <8 x double> undef, double %a, i32 0
  %c = shufflevector <8 x double> %b, <8 x double> undef, <8 x i32> zeroinitializer
  ret <8 x double> %c
}