; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefix=SSE2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s  --check-prefix=AVX --check-prefix=AVX1
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s  --check-prefix=AVX --check-prefix=AVX2
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f | FileCheck %s  --check-prefix=KNL
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl,+avx512dq,+avx512bw | FileCheck %s  --check-prefix=SKX

define i1 @allones_v16i8_sign(<16 x i8> %arg) {
; SSE2-LABEL: allones_v16i8_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpw $-1, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allones_v16i8_sign:
; AVX:       # %bb.0:
; AVX-NEXT:    vpmovmskb %xmm0, %eax
; AVX-NEXT:    cmpw $-1, %ax
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allones_v16i8_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpmovmskb %xmm0, %eax
; KNL-NEXT:    cmpw $-1, %ax
; KNL-NEXT:    sete %al
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v16i8_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovb2m %xmm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    retq
  %tmp = icmp slt <16 x i8> %arg, zeroinitializer
  %tmp1 = bitcast <16 x i1> %tmp to i16
  %tmp2 = icmp eq i16 %tmp1, -1
  ret i1 %tmp2
}

define i1 @allzeros_v16i8_sign(<16 x i8> %arg) {
; SSE2-LABEL: allzeros_v16i8_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testw %ax, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allzeros_v16i8_sign:
; AVX:       # %bb.0:
; AVX-NEXT:    vpmovmskb %xmm0, %eax
; AVX-NEXT:    testw %ax, %ax
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allzeros_v16i8_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpmovmskb %xmm0, %eax
; KNL-NEXT:    testw %ax, %ax
; KNL-NEXT:    sete %al
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v16i8_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovb2m %xmm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = icmp slt <16 x i8> %arg, zeroinitializer
  %tmp1 = bitcast <16 x i1> %tmp to i16
  %tmp2 = icmp eq i16 %tmp1, 0
  ret i1 %tmp2
}

define i1 @allones_v32i8_sign(<32 x i8> %arg) {
; SSE2-LABEL: allones_v32i8_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    pmovmskb %xmm1, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    cmpl $-1, %ecx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v32i8_sign:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    cmpl $-1, %ecx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v32i8_sign:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpmovmskb %ymm0, %eax
; AVX2-NEXT:    cmpl $-1, %eax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v32i8_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpmovmskb %ymm0, %eax
; KNL-NEXT:    cmpl $-1, %eax
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v32i8_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovb2m %ymm0, %k0
; SKX-NEXT:    kortestd %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <32 x i8> %arg, zeroinitializer
  %tmp1 = bitcast <32 x i1> %tmp to i32
  %tmp2 = icmp eq i32 %tmp1, -1
  ret i1 %tmp2
}

define i1 @allzeros_v32i8_sign(<32 x i8> %arg) {
; SSE2-LABEL: allzeros_v32i8_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    pmovmskb %xmm1, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v32i8_sign:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v32i8_sign:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpmovmskb %ymm0, %eax
; AVX2-NEXT:    testl %eax, %eax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v32i8_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpmovmskb %ymm0, %eax
; KNL-NEXT:    testl %eax, %eax
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v32i8_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovb2m %ymm0, %k0
; SKX-NEXT:    kortestd %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <32 x i8> %arg, zeroinitializer
  %tmp1 = bitcast <32 x i1> %tmp to i32
  %tmp2 = icmp eq i32 %tmp1, 0
  ret i1 %tmp2
}

define i1 @allones_v64i8_sign(<64 x i8> %arg) {
; SSE2-LABEL: allones_v64i8_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    pmovmskb %xmm1, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    pmovmskb %xmm2, %eax
; SSE2-NEXT:    pmovmskb %xmm3, %edx
; SSE2-NEXT:    shll $16, %edx
; SSE2-NEXT:    orl %eax, %edx
; SSE2-NEXT:    shlq $32, %rdx
; SSE2-NEXT:    orq %rcx, %rdx
; SSE2-NEXT:    cmpq $-1, %rdx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v64i8_sign:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    vpmovmskb %xmm1, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %edx
; AVX1-NEXT:    shll $16, %edx
; AVX1-NEXT:    orl %eax, %edx
; AVX1-NEXT:    shlq $32, %rdx
; AVX1-NEXT:    orq %rcx, %rdx
; AVX1-NEXT:    cmpq $-1, %rdx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v64i8_sign:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpmovmskb %ymm1, %eax
; AVX2-NEXT:    shlq $32, %rax
; AVX2-NEXT:    vpmovmskb %ymm0, %ecx
; AVX2-NEXT:    orq %rax, %rcx
; AVX2-NEXT:    cmpq $-1, %rcx
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v64i8_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpmovmskb %ymm1, %eax
; KNL-NEXT:    shlq $32, %rax
; KNL-NEXT:    vpmovmskb %ymm0, %ecx
; KNL-NEXT:    orq %rax, %rcx
; KNL-NEXT:    cmpq $-1, %rcx
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v64i8_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovb2m %zmm0, %k0
; SKX-NEXT:    kortestq %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <64 x i8> %arg, zeroinitializer
  %tmp1 = bitcast <64 x i1> %tmp to i64
  %tmp2 = icmp eq i64 %tmp1, -1
  ret i1 %tmp2
}

define i1 @allzeros_v64i8_sign(<64 x i8> %arg) {
; SSE2-LABEL: allzeros_v64i8_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    pmovmskb %xmm1, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    pmovmskb %xmm2, %eax
; SSE2-NEXT:    pmovmskb %xmm3, %edx
; SSE2-NEXT:    shll $16, %edx
; SSE2-NEXT:    orl %eax, %edx
; SSE2-NEXT:    shlq $32, %rdx
; SSE2-NEXT:    orq %rcx, %rdx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v64i8_sign:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    vpmovmskb %xmm1, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %edx
; AVX1-NEXT:    shll $16, %edx
; AVX1-NEXT:    orl %eax, %edx
; AVX1-NEXT:    shlq $32, %rdx
; AVX1-NEXT:    orq %rcx, %rdx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v64i8_sign:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpmovmskb %ymm1, %eax
; AVX2-NEXT:    shlq $32, %rax
; AVX2-NEXT:    vpmovmskb %ymm0, %ecx
; AVX2-NEXT:    orq %rax, %rcx
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v64i8_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpmovmskb %ymm1, %eax
; KNL-NEXT:    shlq $32, %rax
; KNL-NEXT:    vpmovmskb %ymm0, %ecx
; KNL-NEXT:    orq %rax, %rcx
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v64i8_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovb2m %zmm0, %k0
; SKX-NEXT:    kortestq %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <64 x i8> %arg, zeroinitializer
  %tmp1 = bitcast <64 x i1> %tmp to i64
  %tmp2 = icmp eq i64 %tmp1, 0
  ret i1 %tmp2
}

define i1 @allones_v8i16_sign(<8 x i16> %arg) {
; SSE2-LABEL: allones_v8i16_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpb $-1, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allones_v8i16_sign:
; AVX:       # %bb.0:
; AVX-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
; AVX-NEXT:    vpmovmskb %xmm0, %eax
; AVX-NEXT:    cmpb $-1, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allones_v8i16_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtw %xmm0, %xmm1, %xmm0
; KNL-NEXT:    vpmovsxwq %xmm0, %zmm0
; KNL-NEXT:    vptestmq %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    cmpb $-1, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v8i16_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovw2m %xmm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    retq
  %tmp = icmp slt <8 x i16> %arg, zeroinitializer
  %tmp1 = bitcast <8 x i1> %tmp to i8
  %tmp2 = icmp eq i8 %tmp1, -1
  ret i1 %tmp2
}

define i1 @allzeros_v8i16_sign(<8 x i16> %arg) {
; SSE2-LABEL: allzeros_v8i16_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allzeros_v8i16_sign:
; AVX:       # %bb.0:
; AVX-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
; AVX-NEXT:    vpmovmskb %xmm0, %eax
; AVX-NEXT:    testb %al, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allzeros_v8i16_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtw %xmm0, %xmm1, %xmm0
; KNL-NEXT:    vpmovsxwq %xmm0, %zmm0
; KNL-NEXT:    vptestmq %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb %al, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v8i16_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovw2m %xmm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = icmp slt <8 x i16> %arg, zeroinitializer
  %tmp1 = bitcast <8 x i1> %tmp to i8
  %tmp2 = icmp eq i8 %tmp1, 0
  ret i1 %tmp2
}

define i1 @allones_v16i16_sign(<16 x i16> %arg) {
; SSE2-LABEL: allones_v16i16_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    packsswb %xmm1, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpw $-1, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v16i16_sign:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    cmpw $-1, %ax
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v16i16_sign:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; AVX2-NEXT:    vpcmpgtw %ymm0, %ymm1, %ymm0
; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
; AVX2-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vpmovmskb %xmm0, %eax
; AVX2-NEXT:    cmpw $-1, %ax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v16i16_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtw %ymm0, %ymm1, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kortestw %k0, %k0
; KNL-NEXT:    setb %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v16i16_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovw2m %ymm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <16 x i16> %arg, zeroinitializer
  %tmp1 = bitcast <16 x i1> %tmp to i16
  %tmp2 = icmp eq i16 %tmp1, -1
  ret i1 %tmp2
}

define i1 @allzeros_v16i16_sign(<16 x i16> %arg) {
; SSE2-LABEL: allzeros_v16i16_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    packsswb %xmm1, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testw %ax, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v16i16_sign:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    testw %ax, %ax
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v16i16_sign:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; AVX2-NEXT:    vpcmpgtw %ymm0, %ymm1, %ymm0
; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
; AVX2-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vpmovmskb %xmm0, %eax
; AVX2-NEXT:    testw %ax, %ax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v16i16_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtw %ymm0, %ymm1, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kortestw %k0, %k0
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v16i16_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovw2m %ymm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <16 x i16> %arg, zeroinitializer
  %tmp1 = bitcast <16 x i1> %tmp to i16
  %tmp2 = icmp eq i16 %tmp1, 0
  ret i1 %tmp2
}

define i1 @allones_v32i16_sign(<32 x i16> %arg) {
; SSE2-LABEL: allones_v32i16_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    packsswb %xmm1, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    packsswb %xmm3, %xmm2
; SSE2-NEXT:    pmovmskb %xmm2, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    cmpl $-1, %ecx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v32i16_sign:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpacksswb %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
; AVX1-NEXT:    vpacksswb %xmm0, %xmm1, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    cmpl $-1, %ecx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v32i16_sign:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpacksswb %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vpmovmskb %ymm0, %eax
; AVX2-NEXT:    cmpl $-1, %eax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v32i16_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpxor %xmm2, %xmm2, %xmm2
; KNL-NEXT:    vpcmpgtw %ymm0, %ymm2, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    vpcmpgtw %ymm1, %ymm2, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %ecx
; KNL-NEXT:    shll $16, %ecx
; KNL-NEXT:    orl %eax, %ecx
; KNL-NEXT:    cmpl $-1, %ecx
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v32i16_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovw2m %zmm0, %k0
; SKX-NEXT:    kortestd %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <32 x i16> %arg, zeroinitializer
  %tmp1 = bitcast <32 x i1> %tmp to i32
  %tmp2 = icmp eq i32 %tmp1, -1
  ret i1 %tmp2
}

define i1 @allzeros_v32i16_sign(<32 x i16> %arg) {
; SSE2-LABEL: allzeros_v32i16_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    packsswb %xmm1, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    packsswb %xmm3, %xmm2
; SSE2-NEXT:    pmovmskb %xmm2, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v32i16_sign:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpacksswb %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
; AVX1-NEXT:    vpacksswb %xmm0, %xmm1, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v32i16_sign:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpacksswb %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vpmovmskb %ymm0, %eax
; AVX2-NEXT:    testl %eax, %eax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v32i16_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpxor %xmm2, %xmm2, %xmm2
; KNL-NEXT:    vpcmpgtw %ymm0, %ymm2, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    vpcmpgtw %ymm1, %ymm2, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %ecx
; KNL-NEXT:    shll $16, %ecx
; KNL-NEXT:    orl %eax, %ecx
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v32i16_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovw2m %zmm0, %k0
; SKX-NEXT:    kortestd %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <32 x i16> %arg, zeroinitializer
  %tmp1 = bitcast <32 x i1> %tmp to i32
  %tmp2 = icmp eq i32 %tmp1, 0
  ret i1 %tmp2
}

define i1 @allones_v4i32_sign(<4 x i32> %arg) {
; SSE2-LABEL: allones_v4i32_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    movmskps %xmm0, %eax
; SSE2-NEXT:    cmpb $15, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allones_v4i32_sign:
; AVX:       # %bb.0:
; AVX-NEXT:    vmovmskps %xmm0, %eax
; AVX-NEXT:    cmpb $15, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allones_v4i32_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtd %zmm0, %zmm1, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    andb $15, %al
; KNL-NEXT:    cmpb $15, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v4i32_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovd2m %xmm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    andb $15, %al
; SKX-NEXT:    cmpb $15, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = icmp slt <4 x i32> %arg, zeroinitializer
  %tmp1 = bitcast <4 x i1> %tmp to i4
  %tmp2 = icmp eq i4 %tmp1, -1
  ret i1 %tmp2
}

define i1 @allzeros_v4i32_sign(<4 x i32> %arg) {
; SSE2-LABEL: allzeros_v4i32_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    movmskps %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allzeros_v4i32_sign:
; AVX:       # %bb.0:
; AVX-NEXT:    vmovmskps %xmm0, %eax
; AVX-NEXT:    testb %al, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allzeros_v4i32_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtd %zmm0, %zmm1, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb $15, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v4i32_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovd2m %xmm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    testb $15, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = icmp slt <4 x i32> %arg, zeroinitializer
  %tmp1 = bitcast <4 x i1> %tmp to i4
  %tmp2 = icmp eq i4 %tmp1, 0
  ret i1 %tmp2
}

define i1 @allones_v8i32_sign(<8 x i32> %arg) {
; SSE2-LABEL: allones_v8i32_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpb $-1, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allones_v8i32_sign:
; AVX:       # %bb.0:
; AVX-NEXT:    vmovmskps %ymm0, %eax
; AVX-NEXT:    cmpb $-1, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    vzeroupper
; AVX-NEXT:    retq
;
; KNL-LABEL: allones_v8i32_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtd %zmm0, %zmm1, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    cmpb $-1, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v8i32_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovd2m %ymm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <8 x i32> %arg, zeroinitializer
  %tmp1 = bitcast <8 x i1> %tmp to i8
  %tmp2 = icmp eq i8 %tmp1, -1
  ret i1 %tmp2
}

define i1 @allzeros_v8i32_sign(<8 x i32> %arg) {
; SSE2-LABEL: allzeros_v8i32_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allzeros_v8i32_sign:
; AVX:       # %bb.0:
; AVX-NEXT:    vmovmskps %ymm0, %eax
; AVX-NEXT:    testb %al, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    vzeroupper
; AVX-NEXT:    retq
;
; KNL-LABEL: allzeros_v8i32_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtd %zmm0, %zmm1, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb %al, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v8i32_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovd2m %ymm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <8 x i32> %arg, zeroinitializer
  %tmp1 = bitcast <8 x i1> %tmp to i8
  %tmp2 = icmp eq i8 %tmp1, 0
  ret i1 %tmp2
}

define i1 @allones_v16i32_sign(<16 x i32> %arg) {
; SSE2-LABEL: allones_v16i32_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    packssdw %xmm3, %xmm2
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packsswb %xmm2, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpw $-1, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v16i32_sign:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    cmpw $-1, %ax
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v16i32_sign:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpxor %xmm2, %xmm2, %xmm2
; AVX2-NEXT:    vpcmpgtd %ymm1, %ymm2, %ymm1
; AVX2-NEXT:    vpcmpgtd %ymm0, %ymm2, %ymm0
; AVX2-NEXT:    vpackssdw %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
; AVX2-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vpmovmskb %xmm0, %eax
; AVX2-NEXT:    cmpw $-1, %ax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v16i32_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtd %zmm0, %zmm1, %k0
; KNL-NEXT:    kortestw %k0, %k0
; KNL-NEXT:    setb %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v16i32_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovd2m %zmm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <16 x i32> %arg, zeroinitializer
  %tmp1 = bitcast <16 x i1> %tmp to i16
  %tmp2 = icmp eq i16 %tmp1, -1
  ret i1 %tmp2
}

define i1 @allzeros_v16i32_sign(<16 x i32> %arg) {
; SSE2-LABEL: allzeros_v16i32_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    packssdw %xmm3, %xmm2
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packsswb %xmm2, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testw %ax, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v16i32_sign:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    testw %ax, %ax
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v16i32_sign:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpxor %xmm2, %xmm2, %xmm2
; AVX2-NEXT:    vpcmpgtd %ymm1, %ymm2, %ymm1
; AVX2-NEXT:    vpcmpgtd %ymm0, %ymm2, %ymm0
; AVX2-NEXT:    vpackssdw %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
; AVX2-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vpmovmskb %xmm0, %eax
; AVX2-NEXT:    testw %ax, %ax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v16i32_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtd %zmm0, %zmm1, %k0
; KNL-NEXT:    kortestw %k0, %k0
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v16i32_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovd2m %zmm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <16 x i32> %arg, zeroinitializer
  %tmp1 = bitcast <16 x i1> %tmp to i16
  %tmp2 = icmp eq i16 %tmp1, 0
  ret i1 %tmp2
}

define i1 @allones_v4i64_sign(<4 x i64> %arg) {
; SSE2-LABEL: allones_v4i64_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    movmskps %xmm0, %eax
; SSE2-NEXT:    cmpb $15, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allones_v4i64_sign:
; AVX:       # %bb.0:
; AVX-NEXT:    vmovmskpd %ymm0, %eax
; AVX-NEXT:    cmpb $15, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    vzeroupper
; AVX-NEXT:    retq
;
; KNL-LABEL: allones_v4i64_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtq %zmm0, %zmm1, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    andb $15, %al
; KNL-NEXT:    cmpb $15, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v4i64_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovq2m %ymm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    andb $15, %al
; SKX-NEXT:    cmpb $15, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <4 x i64> %arg, zeroinitializer
  %tmp1 = bitcast <4 x i1> %tmp to i4
  %tmp2 = icmp eq i4 %tmp1, -1
  ret i1 %tmp2
}

define i1 @allzeros_v4i64_sign(<4 x i64> %arg) {
; SSE2-LABEL: allzeros_v4i64_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    movmskps %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allzeros_v4i64_sign:
; AVX:       # %bb.0:
; AVX-NEXT:    vmovmskpd %ymm0, %eax
; AVX-NEXT:    testb %al, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    vzeroupper
; AVX-NEXT:    retq
;
; KNL-LABEL: allzeros_v4i64_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtq %zmm0, %zmm1, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb $15, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v4i64_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovq2m %ymm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    testb $15, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <4 x i64> %arg, zeroinitializer
  %tmp1 = bitcast <4 x i1> %tmp to i4
  %tmp2 = icmp eq i4 %tmp1, 0
  ret i1 %tmp2
}

define i1 @allones_v8i64_sign(<8 x i64> %arg) {
; SSE2-LABEL: allones_v8i64_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [2147483648,2147483648]
; SSE2-NEXT:    pxor %xmm4, %xmm3
; SSE2-NEXT:    movdqa %xmm4, %xmm5
; SSE2-NEXT:    pcmpgtd %xmm3, %xmm5
; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2]
; SSE2-NEXT:    pcmpeqd %xmm4, %xmm3
; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
; SSE2-NEXT:    pand %xmm6, %xmm3
; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3]
; SSE2-NEXT:    por %xmm3, %xmm5
; SSE2-NEXT:    pxor %xmm4, %xmm2
; SSE2-NEXT:    movdqa %xmm4, %xmm3
; SSE2-NEXT:    pcmpgtd %xmm2, %xmm3
; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm3[0,0,2,2]
; SSE2-NEXT:    pcmpeqd %xmm4, %xmm2
; SSE2-NEXT:    pshufd {{.*#+}} xmm7 = xmm2[1,1,3,3]
; SSE2-NEXT:    pand %xmm6, %xmm7
; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
; SSE2-NEXT:    por %xmm7, %xmm2
; SSE2-NEXT:    packssdw %xmm5, %xmm2
; SSE2-NEXT:    pxor %xmm4, %xmm1
; SSE2-NEXT:    movdqa %xmm4, %xmm3
; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm3[0,0,2,2]
; SSE2-NEXT:    pcmpeqd %xmm4, %xmm1
; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
; SSE2-NEXT:    pand %xmm5, %xmm1
; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
; SSE2-NEXT:    por %xmm1, %xmm3
; SSE2-NEXT:    pxor %xmm4, %xmm0
; SSE2-NEXT:    movdqa %xmm4, %xmm1
; SSE2-NEXT:    pcmpgtd %xmm0, %xmm1
; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm1[0,0,2,2]
; SSE2-NEXT:    pcmpeqd %xmm4, %xmm0
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
; SSE2-NEXT:    pand %xmm5, %xmm0
; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
; SSE2-NEXT:    por %xmm0, %xmm1
; SSE2-NEXT:    packssdw %xmm3, %xmm1
; SSE2-NEXT:    packssdw %xmm2, %xmm1
; SSE2-NEXT:    packsswb %xmm0, %xmm1
; SSE2-NEXT:    pmovmskb %xmm1, %eax
; SSE2-NEXT:    cmpb $-1, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v8i64_sign:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
; AVX1-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
; AVX1-NEXT:    vpcmpgtq %xmm0, %xmm3, %xmm0
; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT:    vmovmskps %ymm0, %eax
; AVX1-NEXT:    cmpb $-1, %al
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v8i64_sign:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpackssdw %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vmovmskps %ymm0, %eax
; AVX2-NEXT:    cmpb $-1, %al
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v8i64_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtq %zmm0, %zmm1, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    cmpb $-1, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v8i64_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovq2m %zmm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <8 x i64> %arg, zeroinitializer
  %tmp1 = bitcast <8 x i1> %tmp to i8
  %tmp2 = icmp eq i8 %tmp1, -1
  ret i1 %tmp2
}

define i1 @allzeros_v8i64_sign(<8 x i64> %arg) {
; SSE2-LABEL: allzeros_v8i64_sign:
; SSE2:       # %bb.0:
; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [2147483648,2147483648]
; SSE2-NEXT:    pxor %xmm4, %xmm3
; SSE2-NEXT:    movdqa %xmm4, %xmm5
; SSE2-NEXT:    pcmpgtd %xmm3, %xmm5
; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2]
; SSE2-NEXT:    pcmpeqd %xmm4, %xmm3
; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
; SSE2-NEXT:    pand %xmm6, %xmm3
; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3]
; SSE2-NEXT:    por %xmm3, %xmm5
; SSE2-NEXT:    pxor %xmm4, %xmm2
; SSE2-NEXT:    movdqa %xmm4, %xmm3
; SSE2-NEXT:    pcmpgtd %xmm2, %xmm3
; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm3[0,0,2,2]
; SSE2-NEXT:    pcmpeqd %xmm4, %xmm2
; SSE2-NEXT:    pshufd {{.*#+}} xmm7 = xmm2[1,1,3,3]
; SSE2-NEXT:    pand %xmm6, %xmm7
; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
; SSE2-NEXT:    por %xmm7, %xmm2
; SSE2-NEXT:    packssdw %xmm5, %xmm2
; SSE2-NEXT:    pxor %xmm4, %xmm1
; SSE2-NEXT:    movdqa %xmm4, %xmm3
; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm3[0,0,2,2]
; SSE2-NEXT:    pcmpeqd %xmm4, %xmm1
; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
; SSE2-NEXT:    pand %xmm5, %xmm1
; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
; SSE2-NEXT:    por %xmm1, %xmm3
; SSE2-NEXT:    pxor %xmm4, %xmm0
; SSE2-NEXT:    movdqa %xmm4, %xmm1
; SSE2-NEXT:    pcmpgtd %xmm0, %xmm1
; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm1[0,0,2,2]
; SSE2-NEXT:    pcmpeqd %xmm4, %xmm0
; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
; SSE2-NEXT:    pand %xmm5, %xmm0
; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
; SSE2-NEXT:    por %xmm0, %xmm1
; SSE2-NEXT:    packssdw %xmm3, %xmm1
; SSE2-NEXT:    packssdw %xmm2, %xmm1
; SSE2-NEXT:    packsswb %xmm0, %xmm1
; SSE2-NEXT:    pmovmskb %xmm1, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v8i64_sign:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
; AVX1-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
; AVX1-NEXT:    vpcmpgtq %xmm0, %xmm3, %xmm0
; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT:    vmovmskps %ymm0, %eax
; AVX1-NEXT:    testb %al, %al
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v8i64_sign:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpackssdw %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vmovmskps %ymm0, %eax
; AVX2-NEXT:    testb %al, %al
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v8i64_sign:
; KNL:       # %bb.0:
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtq %zmm0, %zmm1, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb %al, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v8i64_sign:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovq2m %zmm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = icmp slt <8 x i64> %arg, zeroinitializer
  %tmp1 = bitcast <8 x i1> %tmp to i8
  %tmp2 = icmp eq i8 %tmp1, 0
  ret i1 %tmp2
}

define i1 @allones_v16i8_and1(<16 x i8> %arg) {
; SSE2-LABEL: allones_v16i8_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $7, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpw $-1, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allones_v16i8_and1:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsllw $7, %xmm0, %xmm0
; AVX-NEXT:    vpmovmskb %xmm0, %eax
; AVX-NEXT:    cmpw $-1, %ax
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allones_v16i8_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $7, %xmm0, %xmm0
; KNL-NEXT:    vpmovmskb %xmm0, %eax
; KNL-NEXT:    cmpw $-1, %ax
; KNL-NEXT:    sete %al
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v16i8_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmb {{.*}}(%rip), %xmm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    retq
  %tmp = and <16 x i8> %arg, <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>
  %tmp1 = icmp ne <16 x i8> %tmp, zeroinitializer
  %tmp2 = bitcast <16 x i1> %tmp1 to i16
  %tmp3 = icmp eq i16 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v16i8_and1(<16 x i8> %arg) {
; SSE2-LABEL: allzeros_v16i8_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $7, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testw %ax, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allzeros_v16i8_and1:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsllw $7, %xmm0, %xmm0
; AVX-NEXT:    vpmovmskb %xmm0, %eax
; AVX-NEXT:    testw %ax, %ax
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allzeros_v16i8_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $7, %xmm0, %xmm0
; KNL-NEXT:    vpmovmskb %xmm0, %eax
; KNL-NEXT:    testw %ax, %ax
; KNL-NEXT:    sete %al
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v16i8_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmb {{.*}}(%rip), %xmm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = and <16 x i8> %arg, <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>
  %tmp1 = icmp ne <16 x i8> %tmp, zeroinitializer
  %tmp2 = bitcast <16 x i1> %tmp1 to i16
  %tmp3 = icmp eq i16 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v32i8_and1(<32 x i8> %arg) {
; SSE2-LABEL: allones_v32i8_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $7, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    psllw $7, %xmm1
; SSE2-NEXT:    pmovmskb %xmm1, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    cmpl $-1, %ecx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v32i8_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm1
; AVX1-NEXT:    vpmovmskb %xmm1, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    cmpl $-1, %ecx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v32i8_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $7, %ymm0, %ymm0
; AVX2-NEXT:    vpmovmskb %ymm0, %eax
; AVX2-NEXT:    cmpl $-1, %eax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v32i8_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $7, %ymm0, %ymm0
; KNL-NEXT:    vpmovmskb %ymm0, %eax
; KNL-NEXT:    cmpl $-1, %eax
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v32i8_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmb {{.*}}(%rip), %ymm0, %k0
; SKX-NEXT:    kortestd %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <32 x i8> %arg, <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>
  %tmp1 = icmp ne <32 x i8> %tmp, zeroinitializer
  %tmp2 = bitcast <32 x i1> %tmp1 to i32
  %tmp3 = icmp eq i32 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v32i8_and1(<32 x i8> %arg) {
; SSE2-LABEL: allzeros_v32i8_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $7, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    psllw $7, %xmm1
; SSE2-NEXT:    pmovmskb %xmm1, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v32i8_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm1
; AVX1-NEXT:    vpmovmskb %xmm1, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v32i8_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $7, %ymm0, %ymm0
; AVX2-NEXT:    vpmovmskb %ymm0, %eax
; AVX2-NEXT:    testl %eax, %eax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v32i8_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $7, %ymm0, %ymm0
; KNL-NEXT:    vpmovmskb %ymm0, %eax
; KNL-NEXT:    testl %eax, %eax
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v32i8_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmb {{.*}}(%rip), %ymm0, %k0
; SKX-NEXT:    kortestd %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <32 x i8> %arg, <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>
  %tmp1 = icmp ne <32 x i8> %tmp, zeroinitializer
  %tmp2 = bitcast <32 x i1> %tmp1 to i32
  %tmp3 = icmp eq i32 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v64i8_and1(<64 x i8> %arg) {
; SSE2-LABEL: allones_v64i8_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $7, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    psllw $7, %xmm1
; SSE2-NEXT:    pmovmskb %xmm1, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    psllw $7, %xmm2
; SSE2-NEXT:    pmovmskb %xmm2, %eax
; SSE2-NEXT:    psllw $7, %xmm3
; SSE2-NEXT:    pmovmskb %xmm3, %edx
; SSE2-NEXT:    shll $16, %edx
; SSE2-NEXT:    orl %eax, %edx
; SSE2-NEXT:    shlq $32, %rdx
; SSE2-NEXT:    orq %rcx, %rdx
; SSE2-NEXT:    cmpq $-1, %rdx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v64i8_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm2
; AVX1-NEXT:    vpmovmskb %xmm2, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    vpsllw $7, %xmm1, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %edx
; AVX1-NEXT:    shll $16, %edx
; AVX1-NEXT:    orl %eax, %edx
; AVX1-NEXT:    shlq $32, %rdx
; AVX1-NEXT:    orq %rcx, %rdx
; AVX1-NEXT:    cmpq $-1, %rdx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v64i8_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $7, %ymm1, %ymm1
; AVX2-NEXT:    vpmovmskb %ymm1, %eax
; AVX2-NEXT:    shlq $32, %rax
; AVX2-NEXT:    vpsllw $7, %ymm0, %ymm0
; AVX2-NEXT:    vpmovmskb %ymm0, %ecx
; AVX2-NEXT:    orq %rax, %rcx
; AVX2-NEXT:    cmpq $-1, %rcx
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v64i8_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $7, %ymm0, %ymm0
; KNL-NEXT:    vpsllw $7, %ymm1, %ymm1
; KNL-NEXT:    vpmovmskb %ymm1, %eax
; KNL-NEXT:    shlq $32, %rax
; KNL-NEXT:    vpmovmskb %ymm0, %ecx
; KNL-NEXT:    orq %rax, %rcx
; KNL-NEXT:    cmpq $-1, %rcx
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v64i8_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmb {{.*}}(%rip), %zmm0, %k0
; SKX-NEXT:    kortestq %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <64 x i8> %arg, <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>
  %tmp1 = icmp ne <64 x i8> %tmp, zeroinitializer
  %tmp2 = bitcast <64 x i1> %tmp1 to i64
  %tmp3 = icmp eq i64 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v64i8_and1(<64 x i8> %arg) {
; SSE2-LABEL: allzeros_v64i8_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $7, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    psllw $7, %xmm1
; SSE2-NEXT:    pmovmskb %xmm1, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    psllw $7, %xmm2
; SSE2-NEXT:    pmovmskb %xmm2, %eax
; SSE2-NEXT:    psllw $7, %xmm3
; SSE2-NEXT:    pmovmskb %xmm3, %edx
; SSE2-NEXT:    shll $16, %edx
; SSE2-NEXT:    orl %eax, %edx
; SSE2-NEXT:    shlq $32, %rdx
; SSE2-NEXT:    orq %rcx, %rdx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v64i8_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm2
; AVX1-NEXT:    vpmovmskb %xmm2, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    vpsllw $7, %xmm1, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %edx
; AVX1-NEXT:    shll $16, %edx
; AVX1-NEXT:    orl %eax, %edx
; AVX1-NEXT:    shlq $32, %rdx
; AVX1-NEXT:    orq %rcx, %rdx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v64i8_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $7, %ymm1, %ymm1
; AVX2-NEXT:    vpmovmskb %ymm1, %eax
; AVX2-NEXT:    shlq $32, %rax
; AVX2-NEXT:    vpsllw $7, %ymm0, %ymm0
; AVX2-NEXT:    vpmovmskb %ymm0, %ecx
; AVX2-NEXT:    orq %rax, %rcx
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v64i8_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $7, %ymm0, %ymm0
; KNL-NEXT:    vpsllw $7, %ymm1, %ymm1
; KNL-NEXT:    vpmovmskb %ymm1, %eax
; KNL-NEXT:    shlq $32, %rax
; KNL-NEXT:    vpmovmskb %ymm0, %ecx
; KNL-NEXT:    orq %rax, %rcx
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v64i8_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmb {{.*}}(%rip), %zmm0, %k0
; SKX-NEXT:    kortestq %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <64 x i8> %arg, <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>
  %tmp1 = icmp ne <64 x i8> %tmp, zeroinitializer
  %tmp2 = bitcast <64 x i1> %tmp1 to i64
  %tmp3 = icmp eq i64 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v8i16_and1(<8 x i16> %arg) {
; SSE2-LABEL: allones_v8i16_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $15, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpb $-1, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allones_v8i16_and1:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsllw $15, %xmm0, %xmm0
; AVX-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
; AVX-NEXT:    vpmovmskb %xmm0, %eax
; AVX-NEXT:    cmpb $-1, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allones_v8i16_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $15, %xmm0, %xmm0
; KNL-NEXT:    vpsraw $15, %xmm0, %xmm0
; KNL-NEXT:    vpmovsxwq %xmm0, %zmm0
; KNL-NEXT:    vptestmq %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    cmpb $-1, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v8i16_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmw {{.*}}(%rip), %xmm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    retq
  %tmp = and <8 x i16> %arg, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
  %tmp1 = icmp ne <8 x i16> %tmp, zeroinitializer
  %tmp2 = bitcast <8 x i1> %tmp1 to i8
  %tmp3 = icmp eq i8 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v8i16_and1(<8 x i16> %arg) {
; SSE2-LABEL: allzeros_v8i16_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $15, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allzeros_v8i16_and1:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsllw $15, %xmm0, %xmm0
; AVX-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
; AVX-NEXT:    vpmovmskb %xmm0, %eax
; AVX-NEXT:    testb %al, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allzeros_v8i16_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $15, %xmm0, %xmm0
; KNL-NEXT:    vpsraw $15, %xmm0, %xmm0
; KNL-NEXT:    vpmovsxwq %xmm0, %zmm0
; KNL-NEXT:    vptestmq %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb %al, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v8i16_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmw {{.*}}(%rip), %xmm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = and <8 x i16> %arg, <i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1, i16 1>
  %tmp1 = icmp ne <8 x i16> %tmp, zeroinitializer
  %tmp2 = bitcast <8 x i1> %tmp1 to i8
  %tmp3 = icmp eq i8 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v16i16_and1(<16 x i16> %arg) {
; SSE2-LABEL: allones_v16i16_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $15, %xmm1
; SSE2-NEXT:    psllw $15, %xmm0
; SSE2-NEXT:    packsswb %xmm1, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpw $-1, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v16i16_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT:    vpsllw $15, %xmm1, %xmm1
; AVX1-NEXT:    vpsllw $15, %xmm0, %xmm0
; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    cmpw $-1, %ax
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v16i16_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $15, %ymm0, %ymm0
; AVX2-NEXT:    vpsraw $15, %ymm0, %ymm0
; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
; AVX2-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vpmovmskb %xmm0, %eax
; AVX2-NEXT:    cmpw $-1, %ax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v16i16_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $15, %ymm0, %ymm0
; KNL-NEXT:    vpsraw $15, %ymm0, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kortestw %k0, %k0
; KNL-NEXT:    setb %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v16i16_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmw {{.*}}(%rip), %ymm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <16 x i16> %arg, <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>
  %tmp1 = icmp ne <16 x i16> %tmp, zeroinitializer
  %tmp2 = bitcast <16 x i1> %tmp1 to i16
  %tmp3 = icmp eq i16 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allones_v32i16_and1(<32 x i16> %arg) {
; SSE2-LABEL: allones_v32i16_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $15, %xmm1
; SSE2-NEXT:    psllw $15, %xmm0
; SSE2-NEXT:    packsswb %xmm1, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    psllw $15, %xmm3
; SSE2-NEXT:    psllw $15, %xmm2
; SSE2-NEXT:    packsswb %xmm3, %xmm2
; SSE2-NEXT:    pmovmskb %xmm2, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    cmpl $-1, %ecx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v32i16_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpsllw $15, %xmm2, %xmm2
; AVX1-NEXT:    vpsllw $15, %xmm0, %xmm0
; AVX1-NEXT:    vpacksswb %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
; AVX1-NEXT:    vpsllw $15, %xmm0, %xmm0
; AVX1-NEXT:    vpsllw $15, %xmm1, %xmm1
; AVX1-NEXT:    vpacksswb %xmm0, %xmm1, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    cmpl $-1, %ecx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v32i16_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $15, %ymm1, %ymm1
; AVX2-NEXT:    vpsllw $15, %ymm0, %ymm0
; AVX2-NEXT:    vpacksswb %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vpmovmskb %ymm0, %eax
; AVX2-NEXT:    cmpl $-1, %eax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v32i16_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $15, %ymm0, %ymm0
; KNL-NEXT:    vpsraw $15, %ymm0, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    vpsllw $15, %ymm1, %ymm0
; KNL-NEXT:    vpsraw $15, %ymm0, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %ecx
; KNL-NEXT:    shll $16, %ecx
; KNL-NEXT:    orl %eax, %ecx
; KNL-NEXT:    cmpl $-1, %ecx
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v32i16_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmw {{.*}}(%rip), %zmm0, %k0
; SKX-NEXT:    kortestd %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <32 x i16> %arg, <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>
  %tmp1 = icmp ne <32 x i16> %tmp, zeroinitializer
  %tmp2 = bitcast <32 x i1> %tmp1 to i32
  %tmp3 = icmp eq i32 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v32i16_and1(<32 x i16> %arg) {
; SSE2-LABEL: allzeros_v32i16_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $15, %xmm1
; SSE2-NEXT:    psllw $15, %xmm0
; SSE2-NEXT:    packsswb %xmm1, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    psllw $15, %xmm3
; SSE2-NEXT:    psllw $15, %xmm2
; SSE2-NEXT:    packsswb %xmm3, %xmm2
; SSE2-NEXT:    pmovmskb %xmm2, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v32i16_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpsllw $15, %xmm2, %xmm2
; AVX1-NEXT:    vpsllw $15, %xmm0, %xmm0
; AVX1-NEXT:    vpacksswb %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
; AVX1-NEXT:    vpsllw $15, %xmm0, %xmm0
; AVX1-NEXT:    vpsllw $15, %xmm1, %xmm1
; AVX1-NEXT:    vpacksswb %xmm0, %xmm1, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v32i16_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $15, %ymm1, %ymm1
; AVX2-NEXT:    vpsllw $15, %ymm0, %ymm0
; AVX2-NEXT:    vpacksswb %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vpmovmskb %ymm0, %eax
; AVX2-NEXT:    testl %eax, %eax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v32i16_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $15, %ymm0, %ymm0
; KNL-NEXT:    vpsraw $15, %ymm0, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    vpsllw $15, %ymm1, %ymm0
; KNL-NEXT:    vpsraw $15, %ymm0, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %ecx
; KNL-NEXT:    shll $16, %ecx
; KNL-NEXT:    orl %eax, %ecx
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v32i16_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmw {{.*}}(%rip), %zmm0, %k0
; SKX-NEXT:    kortestd %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <32 x i16> %arg, <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>
  %tmp1 = icmp ne <32 x i16> %tmp, zeroinitializer
  %tmp2 = bitcast <32 x i1> %tmp1 to i32
  %tmp3 = icmp eq i32 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allzeros_v16i16_and1(<16 x i16> %arg) {
; SSE2-LABEL: allzeros_v16i16_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $15, %xmm1
; SSE2-NEXT:    psllw $15, %xmm0
; SSE2-NEXT:    packsswb %xmm1, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testw %ax, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v16i16_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT:    vpsllw $15, %xmm1, %xmm1
; AVX1-NEXT:    vpsllw $15, %xmm0, %xmm0
; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    testw %ax, %ax
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v16i16_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $15, %ymm0, %ymm0
; AVX2-NEXT:    vpsraw $15, %ymm0, %ymm0
; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
; AVX2-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vpmovmskb %xmm0, %eax
; AVX2-NEXT:    testw %ax, %ax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v16i16_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $15, %ymm0, %ymm0
; KNL-NEXT:    vpsraw $15, %ymm0, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kortestw %k0, %k0
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v16i16_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmw {{.*}}(%rip), %ymm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <16 x i16> %arg, <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>
  %tmp1 = icmp ne <16 x i16> %tmp, zeroinitializer
  %tmp2 = bitcast <16 x i1> %tmp1 to i16
  %tmp3 = icmp eq i16 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v4i32_and1(<4 x i32> %arg) {
; SSE2-LABEL: allones_v4i32_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pslld $31, %xmm0
; SSE2-NEXT:    movmskps %xmm0, %eax
; SSE2-NEXT:    cmpb $15, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allones_v4i32_and1:
; AVX:       # %bb.0:
; AVX-NEXT:    vpslld $31, %xmm0, %xmm0
; AVX-NEXT:    vmovmskps %xmm0, %eax
; AVX-NEXT:    cmpb $15, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allones_v4i32_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    andb $15, %al
; KNL-NEXT:    cmpb $15, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v4i32_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmd {{.*}}(%rip){1to4}, %xmm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    andb $15, %al
; SKX-NEXT:    cmpb $15, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = and <4 x i32> %arg, <i32 1, i32 1, i32 1, i32 1>
  %tmp1 = icmp ne <4 x i32> %tmp, zeroinitializer
  %tmp2 = bitcast <4 x i1> %tmp1 to i4
  %tmp3 = icmp eq i4 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v4i32_and1(<4 x i32> %arg) {
; SSE2-LABEL: allzeros_v4i32_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pslld $31, %xmm0
; SSE2-NEXT:    movmskps %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allzeros_v4i32_and1:
; AVX:       # %bb.0:
; AVX-NEXT:    vpslld $31, %xmm0, %xmm0
; AVX-NEXT:    vmovmskps %xmm0, %eax
; AVX-NEXT:    testb %al, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allzeros_v4i32_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb $15, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v4i32_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmd {{.*}}(%rip){1to4}, %xmm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    testb $15, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = and <4 x i32> %arg, <i32 1, i32 1, i32 1, i32 1>
  %tmp1 = icmp ne <4 x i32> %tmp, zeroinitializer
  %tmp2 = bitcast <4 x i1> %tmp1 to i4
  %tmp3 = icmp eq i4 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v8i32_and1(<8 x i32> %arg) {
; SSE2-LABEL: allones_v8i32_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pslld $31, %xmm1
; SSE2-NEXT:    pslld $31, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpb $-1, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v8i32_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpslld $31, %xmm0, %xmm1
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpslld $31, %xmm0, %xmm0
; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
; AVX1-NEXT:    vmovmskps %ymm0, %eax
; AVX1-NEXT:    cmpb $-1, %al
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v8i32_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpslld $31, %ymm0, %ymm0
; AVX2-NEXT:    vmovmskps %ymm0, %eax
; AVX2-NEXT:    cmpb $-1, %al
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v8i32_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; KNL-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    cmpb $-1, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v8i32_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmd {{.*}}(%rip){1to8}, %ymm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <8 x i32> %arg, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %tmp1 = icmp ne <8 x i32> %tmp, zeroinitializer
  %tmp2 = bitcast <8 x i1> %tmp1 to i8
  %tmp3 = icmp eq i8 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v8i32_and1(<8 x i32> %arg) {
; SSE2-LABEL: allzeros_v8i32_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pslld $31, %xmm1
; SSE2-NEXT:    pslld $31, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v8i32_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpslld $31, %xmm0, %xmm1
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpslld $31, %xmm0, %xmm0
; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
; AVX1-NEXT:    vmovmskps %ymm0, %eax
; AVX1-NEXT:    testb %al, %al
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v8i32_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpslld $31, %ymm0, %ymm0
; AVX2-NEXT:    vmovmskps %ymm0, %eax
; AVX2-NEXT:    testb %al, %al
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v8i32_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; KNL-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb %al, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v8i32_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmd {{.*}}(%rip){1to8}, %ymm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <8 x i32> %arg, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %tmp1 = icmp ne <8 x i32> %tmp, zeroinitializer
  %tmp2 = bitcast <8 x i1> %tmp1 to i8
  %tmp3 = icmp eq i8 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v16i32_and1(<16 x i32> %arg) {
; SSE2-LABEL: allones_v16i32_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pslld $31, %xmm3
; SSE2-NEXT:    pslld $31, %xmm2
; SSE2-NEXT:    packssdw %xmm3, %xmm2
; SSE2-NEXT:    pslld $31, %xmm1
; SSE2-NEXT:    pslld $31, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packsswb %xmm2, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpw $-1, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v16i32_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
; AVX1-NEXT:    vpslld $31, %xmm2, %xmm2
; AVX1-NEXT:    vpslld $31, %xmm1, %xmm1
; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpslld $31, %xmm2, %xmm2
; AVX1-NEXT:    vpslld $31, %xmm0, %xmm0
; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    cmpw $-1, %ax
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v16i32_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpslld $31, %ymm1, %ymm1
; AVX2-NEXT:    vpsrad $31, %ymm1, %ymm1
; AVX2-NEXT:    vpslld $31, %ymm0, %ymm0
; AVX2-NEXT:    vpsrad $31, %ymm0, %ymm0
; AVX2-NEXT:    vpackssdw %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
; AVX2-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vpmovmskb %xmm0, %eax
; AVX2-NEXT:    cmpw $-1, %ax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v16i32_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; KNL-NEXT:    kortestw %k0, %k0
; KNL-NEXT:    setb %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v16i32_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <16 x i32> %arg, <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>
  %tmp1 = icmp ne <16 x i32> %tmp, zeroinitializer
  %tmp2 = bitcast <16 x i1> %tmp1 to i16
  %tmp3 = icmp eq i16 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v16i32_and1(<16 x i32> %arg) {
; SSE2-LABEL: allzeros_v16i32_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pslld $31, %xmm3
; SSE2-NEXT:    pslld $31, %xmm2
; SSE2-NEXT:    packssdw %xmm3, %xmm2
; SSE2-NEXT:    pslld $31, %xmm1
; SSE2-NEXT:    pslld $31, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packsswb %xmm2, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testw %ax, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v16i32_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
; AVX1-NEXT:    vpslld $31, %xmm2, %xmm2
; AVX1-NEXT:    vpslld $31, %xmm1, %xmm1
; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpslld $31, %xmm2, %xmm2
; AVX1-NEXT:    vpslld $31, %xmm0, %xmm0
; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    testw %ax, %ax
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v16i32_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpslld $31, %ymm1, %ymm1
; AVX2-NEXT:    vpsrad $31, %ymm1, %ymm1
; AVX2-NEXT:    vpslld $31, %ymm0, %ymm0
; AVX2-NEXT:    vpsrad $31, %ymm0, %ymm0
; AVX2-NEXT:    vpackssdw %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
; AVX2-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vpmovmskb %xmm0, %eax
; AVX2-NEXT:    testw %ax, %ax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v16i32_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; KNL-NEXT:    kortestw %k0, %k0
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v16i32_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <16 x i32> %arg, <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>
  %tmp1 = icmp ne <16 x i32> %tmp, zeroinitializer
  %tmp2 = bitcast <16 x i1> %tmp1 to i16
  %tmp3 = icmp eq i16 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v2i64_and1(<2 x i64> %arg) {
; SSE2-LABEL: allones_v2i64_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllq $63, %xmm0
; SSE2-NEXT:    movmskpd %xmm0, %eax
; SSE2-NEXT:    cmpb $3, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allones_v2i64_and1:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsllq $63, %xmm0, %xmm0
; AVX-NEXT:    vmovmskpd %xmm0, %eax
; AVX-NEXT:    cmpb $3, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allones_v2i64_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,1]
; KNL-NEXT:    vptestmq %zmm1, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    andb $3, %al
; KNL-NEXT:    cmpb $3, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v2i64_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmq {{.*}}(%rip), %xmm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    andb $3, %al
; SKX-NEXT:    cmpb $3, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = and <2 x i64> %arg, <i64 1, i64 1>
  %tmp1 = icmp ne <2 x i64> %tmp, zeroinitializer
  %tmp2 = bitcast <2 x i1> %tmp1 to i2
  %tmp3 = icmp eq i2 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v2i64_and1(<2 x i64> %arg) {
; SSE2-LABEL: allzeros_v2i64_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllq $63, %xmm0
; SSE2-NEXT:    movmskpd %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allzeros_v2i64_and1:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsllq $63, %xmm0, %xmm0
; AVX-NEXT:    vmovmskpd %xmm0, %eax
; AVX-NEXT:    testb %al, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allzeros_v2i64_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,1]
; KNL-NEXT:    vptestmq %zmm1, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb $3, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v2i64_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmq {{.*}}(%rip), %xmm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    testb $3, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = and <2 x i64> %arg, <i64 1, i64 1>
  %tmp1 = icmp ne <2 x i64> %tmp, zeroinitializer
  %tmp2 = bitcast <2 x i1> %tmp1 to i2
  %tmp3 = icmp eq i2 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v4i64_and1(<4 x i64> %arg) {
; SSE2-LABEL: allones_v4i64_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllq $63, %xmm1
; SSE2-NEXT:    psllq $63, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    movmskps %xmm0, %eax
; SSE2-NEXT:    cmpb $15, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v4i64_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpsllq $63, %xmm0, %xmm1
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpsllq $63, %xmm0, %xmm0
; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
; AVX1-NEXT:    vmovmskpd %ymm0, %eax
; AVX1-NEXT:    cmpb $15, %al
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v4i64_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllq $63, %ymm0, %ymm0
; AVX2-NEXT:    vmovmskpd %ymm0, %eax
; AVX2-NEXT:    cmpb $15, %al
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v4i64_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; KNL-NEXT:    vptestmq {{.*}}(%rip){1to8}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    andb $15, %al
; KNL-NEXT:    cmpb $15, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v4i64_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmq {{.*}}(%rip){1to4}, %ymm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    andb $15, %al
; SKX-NEXT:    cmpb $15, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <4 x i64> %arg, <i64 1, i64 1, i64 1, i64 1>
  %tmp1 = icmp ne <4 x i64> %tmp, zeroinitializer
  %tmp2 = bitcast <4 x i1> %tmp1 to i4
  %tmp3 = icmp eq i4 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v4i64_and1(<4 x i64> %arg) {
; SSE2-LABEL: allzeros_v4i64_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllq $63, %xmm1
; SSE2-NEXT:    psllq $63, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    movmskps %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v4i64_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpsllq $63, %xmm0, %xmm1
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpsllq $63, %xmm0, %xmm0
; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
; AVX1-NEXT:    vmovmskpd %ymm0, %eax
; AVX1-NEXT:    testb %al, %al
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v4i64_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllq $63, %ymm0, %ymm0
; AVX2-NEXT:    vmovmskpd %ymm0, %eax
; AVX2-NEXT:    testb %al, %al
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v4i64_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; KNL-NEXT:    vptestmq {{.*}}(%rip){1to8}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb $15, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v4i64_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmq {{.*}}(%rip){1to4}, %ymm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    testb $15, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <4 x i64> %arg, <i64 1, i64 1, i64 1, i64 1>
  %tmp1 = icmp ne <4 x i64> %tmp, zeroinitializer
  %tmp2 = bitcast <4 x i1> %tmp1 to i4
  %tmp3 = icmp eq i4 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v8i64_and1(<8 x i64> %arg) {
; SSE2-LABEL: allones_v8i64_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllq $63, %xmm3
; SSE2-NEXT:    psllq $63, %xmm2
; SSE2-NEXT:    packssdw %xmm3, %xmm2
; SSE2-NEXT:    psllq $63, %xmm1
; SSE2-NEXT:    psllq $63, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packssdw %xmm2, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpb $-1, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v8i64_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpsllq $63, %xmm2, %xmm2
; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
; AVX1-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
; AVX1-NEXT:    vpsllq $63, %xmm0, %xmm0
; AVX1-NEXT:    vpcmpgtq %xmm0, %xmm3, %xmm0
; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
; AVX1-NEXT:    vpsllq $63, %xmm2, %xmm2
; AVX1-NEXT:    vpsllq $63, %xmm1, %xmm1
; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT:    vmovmskps %ymm0, %eax
; AVX1-NEXT:    cmpb $-1, %al
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v8i64_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllq $63, %ymm1, %ymm1
; AVX2-NEXT:    vpsllq $63, %ymm0, %ymm0
; AVX2-NEXT:    vpackssdw %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vmovmskps %ymm0, %eax
; AVX2-NEXT:    cmpb $-1, %al
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v8i64_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vptestmq {{.*}}(%rip){1to8}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    cmpb $-1, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v8i64_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmq {{.*}}(%rip){1to8}, %zmm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <8 x i64> %arg, <i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1>
  %tmp1 = icmp ne <8 x i64> %tmp, zeroinitializer
  %tmp2 = bitcast <8 x i1> %tmp1 to i8
  %tmp3 = icmp eq i8 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v8i64_and1(<8 x i64> %arg) {
; SSE2-LABEL: allzeros_v8i64_and1:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllq $63, %xmm3
; SSE2-NEXT:    psllq $63, %xmm2
; SSE2-NEXT:    packssdw %xmm3, %xmm2
; SSE2-NEXT:    psllq $63, %xmm1
; SSE2-NEXT:    psllq $63, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packssdw %xmm2, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v8i64_and1:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpsllq $63, %xmm2, %xmm2
; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
; AVX1-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
; AVX1-NEXT:    vpsllq $63, %xmm0, %xmm0
; AVX1-NEXT:    vpcmpgtq %xmm0, %xmm3, %xmm0
; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
; AVX1-NEXT:    vpsllq $63, %xmm2, %xmm2
; AVX1-NEXT:    vpsllq $63, %xmm1, %xmm1
; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT:    vmovmskps %ymm0, %eax
; AVX1-NEXT:    testb %al, %al
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v8i64_and1:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllq $63, %ymm1, %ymm1
; AVX2-NEXT:    vpsllq $63, %ymm0, %ymm0
; AVX2-NEXT:    vpackssdw %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vmovmskps %ymm0, %eax
; AVX2-NEXT:    testb %al, %al
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v8i64_and1:
; KNL:       # %bb.0:
; KNL-NEXT:    vptestmq {{.*}}(%rip){1to8}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb %al, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v8i64_and1:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmq {{.*}}(%rip){1to8}, %zmm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <8 x i64> %arg, <i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1, i64 1>
  %tmp1 = icmp ne <8 x i64> %tmp, zeroinitializer
  %tmp2 = bitcast <8 x i1> %tmp1 to i8
  %tmp3 = icmp eq i8 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v16i8_and4(<16 x i8> %arg) {
; SSE2-LABEL: allones_v16i8_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $5, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpw $-1, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allones_v16i8_and4:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsllw $5, %xmm0, %xmm0
; AVX-NEXT:    vpmovmskb %xmm0, %eax
; AVX-NEXT:    cmpw $-1, %ax
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allones_v16i8_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $5, %xmm0, %xmm0
; KNL-NEXT:    vpmovmskb %xmm0, %eax
; KNL-NEXT:    cmpw $-1, %ax
; KNL-NEXT:    sete %al
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v16i8_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmb {{.*}}(%rip), %xmm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    retq
  %tmp = and <16 x i8> %arg, <i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4>
  %tmp1 = icmp ne <16 x i8> %tmp, zeroinitializer
  %tmp2 = bitcast <16 x i1> %tmp1 to i16
  %tmp3 = icmp eq i16 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v16i8_and4(<16 x i8> %arg) {
; SSE2-LABEL: allzeros_v16i8_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $5, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testw %ax, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allzeros_v16i8_and4:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsllw $5, %xmm0, %xmm0
; AVX-NEXT:    vpmovmskb %xmm0, %eax
; AVX-NEXT:    testw %ax, %ax
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allzeros_v16i8_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $5, %xmm0, %xmm0
; KNL-NEXT:    vpmovmskb %xmm0, %eax
; KNL-NEXT:    testw %ax, %ax
; KNL-NEXT:    sete %al
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v16i8_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmb {{.*}}(%rip), %xmm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = and <16 x i8> %arg, <i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4>
  %tmp1 = icmp ne <16 x i8> %tmp, zeroinitializer
  %tmp2 = bitcast <16 x i1> %tmp1 to i16
  %tmp3 = icmp eq i16 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v32i8_and4(<32 x i8> %arg) {
; SSE2-LABEL: allones_v32i8_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $5, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    psllw $5, %xmm1
; SSE2-NEXT:    pmovmskb %xmm1, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    cmpl $-1, %ecx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v32i8_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpsllw $5, %xmm0, %xmm1
; AVX1-NEXT:    vpmovmskb %xmm1, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpsllw $5, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    cmpl $-1, %ecx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v32i8_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $5, %ymm0, %ymm0
; AVX2-NEXT:    vpmovmskb %ymm0, %eax
; AVX2-NEXT:    cmpl $-1, %eax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v32i8_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $5, %ymm0, %ymm0
; KNL-NEXT:    vpmovmskb %ymm0, %eax
; KNL-NEXT:    cmpl $-1, %eax
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v32i8_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmb {{.*}}(%rip), %ymm0, %k0
; SKX-NEXT:    kortestd %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <32 x i8> %arg, <i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4>
  %tmp1 = icmp ne <32 x i8> %tmp, zeroinitializer
  %tmp2 = bitcast <32 x i1> %tmp1 to i32
  %tmp3 = icmp eq i32 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v32i8_and4(<32 x i8> %arg) {
; SSE2-LABEL: allzeros_v32i8_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $5, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    psllw $5, %xmm1
; SSE2-NEXT:    pmovmskb %xmm1, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v32i8_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpsllw $5, %xmm0, %xmm1
; AVX1-NEXT:    vpmovmskb %xmm1, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpsllw $5, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v32i8_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $5, %ymm0, %ymm0
; AVX2-NEXT:    vpmovmskb %ymm0, %eax
; AVX2-NEXT:    testl %eax, %eax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v32i8_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $5, %ymm0, %ymm0
; KNL-NEXT:    vpmovmskb %ymm0, %eax
; KNL-NEXT:    testl %eax, %eax
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v32i8_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmb {{.*}}(%rip), %ymm0, %k0
; SKX-NEXT:    kortestd %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <32 x i8> %arg, <i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4>
  %tmp1 = icmp ne <32 x i8> %tmp, zeroinitializer
  %tmp2 = bitcast <32 x i1> %tmp1 to i32
  %tmp3 = icmp eq i32 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v64i8_and4(<64 x i8> %arg) {
; SSE2-LABEL: allones_v64i8_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $5, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    psllw $5, %xmm1
; SSE2-NEXT:    pmovmskb %xmm1, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    psllw $5, %xmm2
; SSE2-NEXT:    pmovmskb %xmm2, %eax
; SSE2-NEXT:    psllw $5, %xmm3
; SSE2-NEXT:    pmovmskb %xmm3, %edx
; SSE2-NEXT:    shll $16, %edx
; SSE2-NEXT:    orl %eax, %edx
; SSE2-NEXT:    shlq $32, %rdx
; SSE2-NEXT:    orq %rcx, %rdx
; SSE2-NEXT:    cmpq $-1, %rdx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v64i8_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpsllw $5, %xmm0, %xmm2
; AVX1-NEXT:    vpmovmskb %xmm2, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpsllw $5, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    vpsllw $5, %xmm1, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
; AVX1-NEXT:    vpsllw $5, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %edx
; AVX1-NEXT:    shll $16, %edx
; AVX1-NEXT:    orl %eax, %edx
; AVX1-NEXT:    shlq $32, %rdx
; AVX1-NEXT:    orq %rcx, %rdx
; AVX1-NEXT:    cmpq $-1, %rdx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v64i8_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $5, %ymm1, %ymm1
; AVX2-NEXT:    vpmovmskb %ymm1, %eax
; AVX2-NEXT:    shlq $32, %rax
; AVX2-NEXT:    vpsllw $5, %ymm0, %ymm0
; AVX2-NEXT:    vpmovmskb %ymm0, %ecx
; AVX2-NEXT:    orq %rax, %rcx
; AVX2-NEXT:    cmpq $-1, %rcx
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v64i8_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $5, %ymm0, %ymm0
; KNL-NEXT:    vpsllw $5, %ymm1, %ymm1
; KNL-NEXT:    vpmovmskb %ymm1, %eax
; KNL-NEXT:    shlq $32, %rax
; KNL-NEXT:    vpmovmskb %ymm0, %ecx
; KNL-NEXT:    orq %rax, %rcx
; KNL-NEXT:    cmpq $-1, %rcx
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v64i8_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmb {{.*}}(%rip), %zmm0, %k0
; SKX-NEXT:    kortestq %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <64 x i8> %arg, <i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4>
  %tmp1 = icmp ne <64 x i8> %tmp, zeroinitializer
  %tmp2 = bitcast <64 x i1> %tmp1 to i64
  %tmp3 = icmp eq i64 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v64i8_and4(<64 x i8> %arg) {
; SSE2-LABEL: allzeros_v64i8_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $5, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    psllw $5, %xmm1
; SSE2-NEXT:    pmovmskb %xmm1, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    psllw $5, %xmm2
; SSE2-NEXT:    pmovmskb %xmm2, %eax
; SSE2-NEXT:    psllw $5, %xmm3
; SSE2-NEXT:    pmovmskb %xmm3, %edx
; SSE2-NEXT:    shll $16, %edx
; SSE2-NEXT:    orl %eax, %edx
; SSE2-NEXT:    shlq $32, %rdx
; SSE2-NEXT:    orq %rcx, %rdx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v64i8_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpsllw $5, %xmm0, %xmm2
; AVX1-NEXT:    vpmovmskb %xmm2, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpsllw $5, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    vpsllw $5, %xmm1, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
; AVX1-NEXT:    vpsllw $5, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %edx
; AVX1-NEXT:    shll $16, %edx
; AVX1-NEXT:    orl %eax, %edx
; AVX1-NEXT:    shlq $32, %rdx
; AVX1-NEXT:    orq %rcx, %rdx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v64i8_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $5, %ymm1, %ymm1
; AVX2-NEXT:    vpmovmskb %ymm1, %eax
; AVX2-NEXT:    shlq $32, %rax
; AVX2-NEXT:    vpsllw $5, %ymm0, %ymm0
; AVX2-NEXT:    vpmovmskb %ymm0, %ecx
; AVX2-NEXT:    orq %rax, %rcx
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v64i8_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $5, %ymm0, %ymm0
; KNL-NEXT:    vpsllw $5, %ymm1, %ymm1
; KNL-NEXT:    vpmovmskb %ymm1, %eax
; KNL-NEXT:    shlq $32, %rax
; KNL-NEXT:    vpmovmskb %ymm0, %ecx
; KNL-NEXT:    orq %rax, %rcx
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v64i8_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmb {{.*}}(%rip), %zmm0, %k0
; SKX-NEXT:    kortestq %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <64 x i8> %arg, <i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4, i8 4>
  %tmp1 = icmp ne <64 x i8> %tmp, zeroinitializer
  %tmp2 = bitcast <64 x i1> %tmp1 to i64
  %tmp3 = icmp eq i64 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v8i16_and4(<8 x i16> %arg) {
; SSE2-LABEL: allones_v8i16_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $13, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpb $-1, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allones_v8i16_and4:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsllw $13, %xmm0, %xmm0
; AVX-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
; AVX-NEXT:    vpmovmskb %xmm0, %eax
; AVX-NEXT:    cmpb $-1, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allones_v8i16_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $13, %xmm0, %xmm0
; KNL-NEXT:    vpsraw $15, %xmm0, %xmm0
; KNL-NEXT:    vpmovsxwq %xmm0, %zmm0
; KNL-NEXT:    vptestmq %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    cmpb $-1, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v8i16_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmw {{.*}}(%rip), %xmm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    retq
  %tmp = and <8 x i16> %arg, <i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4>
  %tmp1 = icmp ne <8 x i16> %tmp, zeroinitializer
  %tmp2 = bitcast <8 x i1> %tmp1 to i8
  %tmp3 = icmp eq i8 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v8i16_and4(<8 x i16> %arg) {
; SSE2-LABEL: allzeros_v8i16_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $13, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allzeros_v8i16_and4:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsllw $13, %xmm0, %xmm0
; AVX-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
; AVX-NEXT:    vpmovmskb %xmm0, %eax
; AVX-NEXT:    testb %al, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allzeros_v8i16_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $13, %xmm0, %xmm0
; KNL-NEXT:    vpsraw $15, %xmm0, %xmm0
; KNL-NEXT:    vpmovsxwq %xmm0, %zmm0
; KNL-NEXT:    vptestmq %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb %al, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v8i16_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmw {{.*}}(%rip), %xmm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = and <8 x i16> %arg, <i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4>
  %tmp1 = icmp ne <8 x i16> %tmp, zeroinitializer
  %tmp2 = bitcast <8 x i1> %tmp1 to i8
  %tmp3 = icmp eq i8 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v16i16_and4(<16 x i16> %arg) {
; SSE2-LABEL: allones_v16i16_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $13, %xmm1
; SSE2-NEXT:    psllw $13, %xmm0
; SSE2-NEXT:    packsswb %xmm1, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpw $-1, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v16i16_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT:    vpsllw $13, %xmm1, %xmm1
; AVX1-NEXT:    vpsllw $13, %xmm0, %xmm0
; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    cmpw $-1, %ax
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v16i16_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $13, %ymm0, %ymm0
; AVX2-NEXT:    vpsraw $15, %ymm0, %ymm0
; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
; AVX2-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vpmovmskb %xmm0, %eax
; AVX2-NEXT:    cmpw $-1, %ax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v16i16_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $13, %ymm0, %ymm0
; KNL-NEXT:    vpsraw $15, %ymm0, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kortestw %k0, %k0
; KNL-NEXT:    setb %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v16i16_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmw {{.*}}(%rip), %ymm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <16 x i16> %arg, <i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4>
  %tmp1 = icmp ne <16 x i16> %tmp, zeroinitializer
  %tmp2 = bitcast <16 x i1> %tmp1 to i16
  %tmp3 = icmp eq i16 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allones_v32i16_and4(<32 x i16> %arg) {
; SSE2-LABEL: allones_v32i16_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $13, %xmm1
; SSE2-NEXT:    psllw $13, %xmm0
; SSE2-NEXT:    packsswb %xmm1, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    psllw $13, %xmm3
; SSE2-NEXT:    psllw $13, %xmm2
; SSE2-NEXT:    packsswb %xmm3, %xmm2
; SSE2-NEXT:    pmovmskb %xmm2, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    cmpl $-1, %ecx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v32i16_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpsllw $13, %xmm2, %xmm2
; AVX1-NEXT:    vpsllw $13, %xmm0, %xmm0
; AVX1-NEXT:    vpacksswb %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
; AVX1-NEXT:    vpsllw $13, %xmm0, %xmm0
; AVX1-NEXT:    vpsllw $13, %xmm1, %xmm1
; AVX1-NEXT:    vpacksswb %xmm0, %xmm1, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    cmpl $-1, %ecx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v32i16_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $13, %ymm1, %ymm1
; AVX2-NEXT:    vpsllw $13, %ymm0, %ymm0
; AVX2-NEXT:    vpacksswb %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vpmovmskb %ymm0, %eax
; AVX2-NEXT:    cmpl $-1, %eax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v32i16_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $13, %ymm0, %ymm0
; KNL-NEXT:    vpsraw $15, %ymm0, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    vpsllw $13, %ymm1, %ymm0
; KNL-NEXT:    vpsraw $15, %ymm0, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %ecx
; KNL-NEXT:    shll $16, %ecx
; KNL-NEXT:    orl %eax, %ecx
; KNL-NEXT:    cmpl $-1, %ecx
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v32i16_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmw {{.*}}(%rip), %zmm0, %k0
; SKX-NEXT:    kortestd %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <32 x i16> %arg, <i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4>
  %tmp1 = icmp ne <32 x i16> %tmp, zeroinitializer
  %tmp2 = bitcast <32 x i1> %tmp1 to i32
  %tmp3 = icmp eq i32 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v32i16_and4(<32 x i16> %arg) {
; SSE2-LABEL: allzeros_v32i16_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $13, %xmm1
; SSE2-NEXT:    psllw $13, %xmm0
; SSE2-NEXT:    packsswb %xmm1, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    psllw $13, %xmm3
; SSE2-NEXT:    psllw $13, %xmm2
; SSE2-NEXT:    packsswb %xmm3, %xmm2
; SSE2-NEXT:    pmovmskb %xmm2, %ecx
; SSE2-NEXT:    shll $16, %ecx
; SSE2-NEXT:    orl %eax, %ecx
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v32i16_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpsllw $13, %xmm2, %xmm2
; AVX1-NEXT:    vpsllw $13, %xmm0, %xmm0
; AVX1-NEXT:    vpacksswb %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
; AVX1-NEXT:    vpsllw $13, %xmm0, %xmm0
; AVX1-NEXT:    vpsllw $13, %xmm1, %xmm1
; AVX1-NEXT:    vpacksswb %xmm0, %xmm1, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    shll $16, %ecx
; AVX1-NEXT:    orl %eax, %ecx
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v32i16_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $13, %ymm1, %ymm1
; AVX2-NEXT:    vpsllw $13, %ymm0, %ymm0
; AVX2-NEXT:    vpacksswb %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vpmovmskb %ymm0, %eax
; AVX2-NEXT:    testl %eax, %eax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v32i16_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $13, %ymm0, %ymm0
; KNL-NEXT:    vpsraw $15, %ymm0, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    vpsllw $13, %ymm1, %ymm0
; KNL-NEXT:    vpsraw $15, %ymm0, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %ecx
; KNL-NEXT:    shll $16, %ecx
; KNL-NEXT:    orl %eax, %ecx
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v32i16_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmw {{.*}}(%rip), %zmm0, %k0
; SKX-NEXT:    kortestd %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <32 x i16> %arg, <i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4>
  %tmp1 = icmp ne <32 x i16> %tmp, zeroinitializer
  %tmp2 = bitcast <32 x i1> %tmp1 to i32
  %tmp3 = icmp eq i32 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allzeros_v16i16_and4(<16 x i16> %arg) {
; SSE2-LABEL: allzeros_v16i16_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllw $13, %xmm1
; SSE2-NEXT:    psllw $13, %xmm0
; SSE2-NEXT:    packsswb %xmm1, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testw %ax, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v16i16_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
; AVX1-NEXT:    vpsllw $13, %xmm1, %xmm1
; AVX1-NEXT:    vpsllw $13, %xmm0, %xmm0
; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    testw %ax, %ax
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v16i16_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllw $13, %ymm0, %ymm0
; AVX2-NEXT:    vpsraw $15, %ymm0, %ymm0
; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
; AVX2-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vpmovmskb %xmm0, %eax
; AVX2-NEXT:    testw %ax, %ax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v16i16_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vpsllw $13, %ymm0, %ymm0
; KNL-NEXT:    vpsraw $15, %ymm0, %ymm0
; KNL-NEXT:    vpmovsxwd %ymm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kortestw %k0, %k0
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v16i16_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmw {{.*}}(%rip), %ymm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <16 x i16> %arg, <i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4, i16 4>
  %tmp1 = icmp ne <16 x i16> %tmp, zeroinitializer
  %tmp2 = bitcast <16 x i1> %tmp1 to i16
  %tmp3 = icmp eq i16 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v4i32_and4(<4 x i32> %arg) {
; SSE2-LABEL: allones_v4i32_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pslld $29, %xmm0
; SSE2-NEXT:    movmskps %xmm0, %eax
; SSE2-NEXT:    cmpb $15, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allones_v4i32_and4:
; AVX:       # %bb.0:
; AVX-NEXT:    vpslld $29, %xmm0, %xmm0
; AVX-NEXT:    vmovmskps %xmm0, %eax
; AVX-NEXT:    cmpb $15, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allones_v4i32_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    andb $15, %al
; KNL-NEXT:    cmpb $15, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v4i32_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmd {{.*}}(%rip){1to4}, %xmm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    andb $15, %al
; SKX-NEXT:    cmpb $15, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = and <4 x i32> %arg, <i32 4, i32 4, i32 4, i32 4>
  %tmp1 = icmp ne <4 x i32> %tmp, zeroinitializer
  %tmp2 = bitcast <4 x i1> %tmp1 to i4
  %tmp3 = icmp eq i4 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v4i32_and4(<4 x i32> %arg) {
; SSE2-LABEL: allzeros_v4i32_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pslld $29, %xmm0
; SSE2-NEXT:    movmskps %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allzeros_v4i32_and4:
; AVX:       # %bb.0:
; AVX-NEXT:    vpslld $29, %xmm0, %xmm0
; AVX-NEXT:    vmovmskps %xmm0, %eax
; AVX-NEXT:    testb %al, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allzeros_v4i32_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb $15, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v4i32_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmd {{.*}}(%rip){1to4}, %xmm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    testb $15, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = and <4 x i32> %arg, <i32 4, i32 4, i32 4, i32 4>
  %tmp1 = icmp ne <4 x i32> %tmp, zeroinitializer
  %tmp2 = bitcast <4 x i1> %tmp1 to i4
  %tmp3 = icmp eq i4 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v8i32_and4(<8 x i32> %arg) {
; SSE2-LABEL: allones_v8i32_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pslld $29, %xmm1
; SSE2-NEXT:    pslld $29, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpb $-1, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v8i32_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpslld $29, %xmm0, %xmm1
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpslld $29, %xmm0, %xmm0
; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
; AVX1-NEXT:    vmovmskps %ymm0, %eax
; AVX1-NEXT:    cmpb $-1, %al
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v8i32_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpslld $29, %ymm0, %ymm0
; AVX2-NEXT:    vmovmskps %ymm0, %eax
; AVX2-NEXT:    cmpb $-1, %al
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v8i32_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; KNL-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    cmpb $-1, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v8i32_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmd {{.*}}(%rip){1to8}, %ymm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <8 x i32> %arg, <i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4>
  %tmp1 = icmp ne <8 x i32> %tmp, zeroinitializer
  %tmp2 = bitcast <8 x i1> %tmp1 to i8
  %tmp3 = icmp eq i8 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v8i32_and4(<8 x i32> %arg) {
; SSE2-LABEL: allzeros_v8i32_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pslld $29, %xmm1
; SSE2-NEXT:    pslld $29, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v8i32_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpslld $29, %xmm0, %xmm1
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpslld $29, %xmm0, %xmm0
; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
; AVX1-NEXT:    vmovmskps %ymm0, %eax
; AVX1-NEXT:    testb %al, %al
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v8i32_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpslld $29, %ymm0, %ymm0
; AVX2-NEXT:    vmovmskps %ymm0, %eax
; AVX2-NEXT:    testb %al, %al
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v8i32_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; KNL-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb %al, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v8i32_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmd {{.*}}(%rip){1to8}, %ymm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <8 x i32> %arg, <i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4>
  %tmp1 = icmp ne <8 x i32> %tmp, zeroinitializer
  %tmp2 = bitcast <8 x i1> %tmp1 to i8
  %tmp3 = icmp eq i8 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v16i32_and4(<16 x i32> %arg) {
; SSE2-LABEL: allones_v16i32_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pslld $29, %xmm3
; SSE2-NEXT:    pslld $29, %xmm2
; SSE2-NEXT:    packssdw %xmm3, %xmm2
; SSE2-NEXT:    pslld $29, %xmm1
; SSE2-NEXT:    pslld $29, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packsswb %xmm2, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpw $-1, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v16i32_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
; AVX1-NEXT:    vpslld $29, %xmm2, %xmm2
; AVX1-NEXT:    vpslld $29, %xmm1, %xmm1
; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpslld $29, %xmm2, %xmm2
; AVX1-NEXT:    vpslld $29, %xmm0, %xmm0
; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    cmpw $-1, %ax
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v16i32_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpslld $29, %ymm1, %ymm1
; AVX2-NEXT:    vpsrad $31, %ymm1, %ymm1
; AVX2-NEXT:    vpslld $29, %ymm0, %ymm0
; AVX2-NEXT:    vpsrad $31, %ymm0, %ymm0
; AVX2-NEXT:    vpackssdw %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
; AVX2-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vpmovmskb %xmm0, %eax
; AVX2-NEXT:    cmpw $-1, %ax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v16i32_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; KNL-NEXT:    kortestw %k0, %k0
; KNL-NEXT:    setb %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v16i32_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <16 x i32> %arg, <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>
  %tmp1 = icmp ne <16 x i32> %tmp, zeroinitializer
  %tmp2 = bitcast <16 x i1> %tmp1 to i16
  %tmp3 = icmp eq i16 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v16i32_and4(<16 x i32> %arg) {
; SSE2-LABEL: allzeros_v16i32_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pslld $29, %xmm3
; SSE2-NEXT:    pslld $29, %xmm2
; SSE2-NEXT:    packssdw %xmm3, %xmm2
; SSE2-NEXT:    pslld $29, %xmm1
; SSE2-NEXT:    pslld $29, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packsswb %xmm2, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testw %ax, %ax
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v16i32_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
; AVX1-NEXT:    vpslld $29, %xmm2, %xmm2
; AVX1-NEXT:    vpslld $29, %xmm1, %xmm1
; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpslld $29, %xmm2, %xmm2
; AVX1-NEXT:    vpslld $29, %xmm0, %xmm0
; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    testw %ax, %ax
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v16i32_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpslld $29, %ymm1, %ymm1
; AVX2-NEXT:    vpsrad $31, %ymm1, %ymm1
; AVX2-NEXT:    vpslld $29, %ymm0, %ymm0
; AVX2-NEXT:    vpsrad $31, %ymm0, %ymm0
; AVX2-NEXT:    vpackssdw %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
; AVX2-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
; AVX2-NEXT:    vpmovmskb %xmm0, %eax
; AVX2-NEXT:    testw %ax, %ax
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v16i32_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; KNL-NEXT:    kortestw %k0, %k0
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v16i32_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmd {{.*}}(%rip){1to16}, %zmm0, %k0
; SKX-NEXT:    kortestw %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <16 x i32> %arg, <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>
  %tmp1 = icmp ne <16 x i32> %tmp, zeroinitializer
  %tmp2 = bitcast <16 x i1> %tmp1 to i16
  %tmp3 = icmp eq i16 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v2i64_and4(<2 x i64> %arg) {
; SSE2-LABEL: allones_v2i64_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllq $61, %xmm0
; SSE2-NEXT:    movmskpd %xmm0, %eax
; SSE2-NEXT:    cmpb $3, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allones_v2i64_and4:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsllq $61, %xmm0, %xmm0
; AVX-NEXT:    vmovmskpd %xmm0, %eax
; AVX-NEXT:    cmpb $3, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allones_v2i64_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vmovdqa {{.*#+}} xmm1 = [4,4]
; KNL-NEXT:    vptestmq %zmm1, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    andb $3, %al
; KNL-NEXT:    cmpb $3, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v2i64_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmq {{.*}}(%rip), %xmm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    andb $3, %al
; SKX-NEXT:    cmpb $3, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = and <2 x i64> %arg, <i64 4, i64 4>
  %tmp1 = icmp ne <2 x i64> %tmp, zeroinitializer
  %tmp2 = bitcast <2 x i1> %tmp1 to i2
  %tmp3 = icmp eq i2 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v2i64_and4(<2 x i64> %arg) {
; SSE2-LABEL: allzeros_v2i64_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllq $61, %xmm0
; SSE2-NEXT:    movmskpd %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: allzeros_v2i64_and4:
; AVX:       # %bb.0:
; AVX-NEXT:    vpsllq $61, %xmm0, %xmm0
; AVX-NEXT:    vmovmskpd %xmm0, %eax
; AVX-NEXT:    testb %al, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: allzeros_v2i64_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vmovdqa {{.*#+}} xmm1 = [4,4]
; KNL-NEXT:    vptestmq %zmm1, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb $3, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v2i64_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmq {{.*}}(%rip), %xmm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    testb $3, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    retq
  %tmp = and <2 x i64> %arg, <i64 4, i64 4>
  %tmp1 = icmp ne <2 x i64> %tmp, zeroinitializer
  %tmp2 = bitcast <2 x i1> %tmp1 to i2
  %tmp3 = icmp eq i2 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v4i64_and4(<4 x i64> %arg) {
; SSE2-LABEL: allones_v4i64_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllq $61, %xmm1
; SSE2-NEXT:    psllq $61, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    movmskps %xmm0, %eax
; SSE2-NEXT:    cmpb $15, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v4i64_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpsllq $61, %xmm0, %xmm1
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpsllq $61, %xmm0, %xmm0
; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
; AVX1-NEXT:    vmovmskpd %ymm0, %eax
; AVX1-NEXT:    cmpb $15, %al
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v4i64_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllq $61, %ymm0, %ymm0
; AVX2-NEXT:    vmovmskpd %ymm0, %eax
; AVX2-NEXT:    cmpb $15, %al
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v4i64_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; KNL-NEXT:    vptestmq {{.*}}(%rip){1to8}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    andb $15, %al
; KNL-NEXT:    cmpb $15, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v4i64_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmq {{.*}}(%rip){1to4}, %ymm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    andb $15, %al
; SKX-NEXT:    cmpb $15, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <4 x i64> %arg, <i64 4, i64 4, i64 4, i64 4>
  %tmp1 = icmp ne <4 x i64> %tmp, zeroinitializer
  %tmp2 = bitcast <4 x i1> %tmp1 to i4
  %tmp3 = icmp eq i4 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v4i64_and4(<4 x i64> %arg) {
; SSE2-LABEL: allzeros_v4i64_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllq $61, %xmm1
; SSE2-NEXT:    psllq $61, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    movmskps %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v4i64_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpsllq $61, %xmm0, %xmm1
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpsllq $61, %xmm0, %xmm0
; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
; AVX1-NEXT:    vmovmskpd %ymm0, %eax
; AVX1-NEXT:    testb %al, %al
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v4i64_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllq $61, %ymm0, %ymm0
; AVX2-NEXT:    vmovmskpd %ymm0, %eax
; AVX2-NEXT:    testb %al, %al
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v4i64_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; KNL-NEXT:    vptestmq {{.*}}(%rip){1to8}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb $15, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v4i64_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmq {{.*}}(%rip){1to4}, %ymm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    testb $15, %al
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <4 x i64> %arg, <i64 4, i64 4, i64 4, i64 4>
  %tmp1 = icmp ne <4 x i64> %tmp, zeroinitializer
  %tmp2 = bitcast <4 x i1> %tmp1 to i4
  %tmp3 = icmp eq i4 %tmp2, 0
  ret i1 %tmp3
}

define i1 @allones_v8i64_and4(<8 x i64> %arg) {
; SSE2-LABEL: allones_v8i64_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllq $61, %xmm3
; SSE2-NEXT:    psllq $61, %xmm2
; SSE2-NEXT:    packssdw %xmm3, %xmm2
; SSE2-NEXT:    psllq $61, %xmm1
; SSE2-NEXT:    psllq $61, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packssdw %xmm2, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    cmpb $-1, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allones_v8i64_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpsllq $61, %xmm2, %xmm2
; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
; AVX1-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
; AVX1-NEXT:    vpsllq $61, %xmm0, %xmm0
; AVX1-NEXT:    vpcmpgtq %xmm0, %xmm3, %xmm0
; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
; AVX1-NEXT:    vpsllq $61, %xmm2, %xmm2
; AVX1-NEXT:    vpsllq $61, %xmm1, %xmm1
; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT:    vmovmskps %ymm0, %eax
; AVX1-NEXT:    cmpb $-1, %al
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allones_v8i64_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllq $61, %ymm1, %ymm1
; AVX2-NEXT:    vpsllq $61, %ymm0, %ymm0
; AVX2-NEXT:    vpackssdw %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vmovmskps %ymm0, %eax
; AVX2-NEXT:    cmpb $-1, %al
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allones_v8i64_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vptestmq {{.*}}(%rip){1to8}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    cmpb $-1, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allones_v8i64_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmq {{.*}}(%rip){1to8}, %zmm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    setb %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <8 x i64> %arg, <i64 4, i64 4, i64 4, i64 4, i64 4, i64 4, i64 4, i64 4>
  %tmp1 = icmp ne <8 x i64> %tmp, zeroinitializer
  %tmp2 = bitcast <8 x i1> %tmp1 to i8
  %tmp3 = icmp eq i8 %tmp2, -1
  ret i1 %tmp3
}

define i1 @allzeros_v8i64_and4(<8 x i64> %arg) {
; SSE2-LABEL: allzeros_v8i64_and4:
; SSE2:       # %bb.0:
; SSE2-NEXT:    psllq $61, %xmm3
; SSE2-NEXT:    psllq $61, %xmm2
; SSE2-NEXT:    packssdw %xmm3, %xmm2
; SSE2-NEXT:    psllq $61, %xmm1
; SSE2-NEXT:    psllq $61, %xmm0
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packssdw %xmm2, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    testb %al, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX1-LABEL: allzeros_v8i64_and4:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
; AVX1-NEXT:    vpsllq $61, %xmm2, %xmm2
; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
; AVX1-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
; AVX1-NEXT:    vpsllq $61, %xmm0, %xmm0
; AVX1-NEXT:    vpcmpgtq %xmm0, %xmm3, %xmm0
; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
; AVX1-NEXT:    vpsllq $61, %xmm2, %xmm2
; AVX1-NEXT:    vpsllq $61, %xmm1, %xmm1
; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; AVX1-NEXT:    vmovmskps %ymm0, %eax
; AVX1-NEXT:    testb %al, %al
; AVX1-NEXT:    sete %al
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: allzeros_v8i64_and4:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpsllq $61, %ymm1, %ymm1
; AVX2-NEXT:    vpsllq $61, %ymm0, %ymm0
; AVX2-NEXT:    vpackssdw %ymm1, %ymm0, %ymm0
; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
; AVX2-NEXT:    vmovmskps %ymm0, %eax
; AVX2-NEXT:    testb %al, %al
; AVX2-NEXT:    sete %al
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: allzeros_v8i64_and4:
; KNL:       # %bb.0:
; KNL-NEXT:    vptestmq {{.*}}(%rip){1to8}, %zmm0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    testb %al, %al
; KNL-NEXT:    sete %al
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: allzeros_v8i64_and4:
; SKX:       # %bb.0:
; SKX-NEXT:    vptestmq {{.*}}(%rip){1to8}, %zmm0, %k0
; SKX-NEXT:    kortestb %k0, %k0
; SKX-NEXT:    sete %al
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %tmp = and <8 x i64> %arg, <i64 4, i64 4, i64 4, i64 4, i64 4, i64 4, i64 4, i64 4>
  %tmp1 = icmp ne <8 x i64> %tmp, zeroinitializer
  %tmp2 = bitcast <8 x i1> %tmp1 to i8
  %tmp3 = icmp eq i8 %tmp2, 0
  ret i1 %tmp3
}

; The below are IR patterns that should directly represent the behavior of a
; MOVMSK instruction.

define i32 @movmskpd(<2 x double> %x) {
; SSE2-LABEL: movmskpd:
; SSE2:       # %bb.0:
; SSE2-NEXT:    movmskpd %xmm0, %eax
; SSE2-NEXT:    retq
;
; AVX-LABEL: movmskpd:
; AVX:       # %bb.0:
; AVX-NEXT:    vmovmskpd %xmm0, %eax
; AVX-NEXT:    retq
;
; KNL-LABEL: movmskpd:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtq %zmm0, %zmm1, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    andl $3, %eax
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: movmskpd:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovq2m %xmm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    andl $3, %eax
; SKX-NEXT:    retq
  %a = bitcast <2 x double> %x to <2 x i64>
  %b = icmp slt <2 x i64> %a, zeroinitializer
  %c = bitcast <2 x i1> %b to i2
  %d = zext i2 %c to i32
  ret i32 %d
}

define i32 @movmskps(<4 x float> %x) {
; SSE2-LABEL: movmskps:
; SSE2:       # %bb.0:
; SSE2-NEXT:    movmskps %xmm0, %eax
; SSE2-NEXT:    retq
;
; AVX-LABEL: movmskps:
; AVX:       # %bb.0:
; AVX-NEXT:    vmovmskps %xmm0, %eax
; AVX-NEXT:    retq
;
; KNL-LABEL: movmskps:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtd %zmm0, %zmm1, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    andl $15, %eax
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: movmskps:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovd2m %xmm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    andl $15, %eax
; SKX-NEXT:    retq
  %a = bitcast <4 x float> %x to <4 x i32>
  %b = icmp slt <4 x i32> %a, zeroinitializer
  %c = bitcast <4 x i1> %b to i4
  %d = zext i4 %c to i32
  ret i32 %d
}

define i32 @movmskpd256(<4 x double> %x) {
; SSE2-LABEL: movmskpd256:
; SSE2:       # %bb.0:
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    movmskps %xmm0, %eax
; SSE2-NEXT:    retq
;
; AVX-LABEL: movmskpd256:
; AVX:       # %bb.0:
; AVX-NEXT:    vmovmskpd %ymm0, %eax
; AVX-NEXT:    vzeroupper
; AVX-NEXT:    retq
;
; KNL-LABEL: movmskpd256:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtq %zmm0, %zmm1, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    andl $15, %eax
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: movmskpd256:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovq2m %ymm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    andl $15, %eax
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %a = bitcast <4 x double> %x to <4 x i64>
  %b = icmp slt <4 x i64> %a, zeroinitializer
  %c = bitcast <4 x i1> %b to i4
  %d = zext i4 %c to i32
  ret i32 %d
}

define i32 @movmskps256(<8 x float> %x) {
; SSE2-LABEL: movmskps256:
; SSE2:       # %bb.0:
; SSE2-NEXT:    packssdw %xmm1, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    movzbl %al, %eax
; SSE2-NEXT:    retq
;
; AVX-LABEL: movmskps256:
; AVX:       # %bb.0:
; AVX-NEXT:    vmovmskps %ymm0, %eax
; AVX-NEXT:    vzeroupper
; AVX-NEXT:    retq
;
; KNL-LABEL: movmskps256:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
; KNL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; KNL-NEXT:    vpcmpgtd %zmm0, %zmm1, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    movzbl %al, %eax
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: movmskps256:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovd2m %ymm0, %k0
; SKX-NEXT:    kmovb %k0, %eax
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %a = bitcast <8 x float> %x to <8 x i32>
  %b = icmp slt <8 x i32> %a, zeroinitializer
  %c = bitcast <8 x i1> %b to i8
  %d = zext i8 %c to i32
  ret i32 %d
}

define i32 @movmskb(<16 x i8> %x) {
; SSE2-LABEL: movmskb:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    retq
;
; AVX-LABEL: movmskb:
; AVX:       # %bb.0:
; AVX-NEXT:    vpmovmskb %xmm0, %eax
; AVX-NEXT:    retq
;
; KNL-LABEL: movmskb:
; KNL:       # %bb.0:
; KNL-NEXT:    vpmovmskb %xmm0, %eax
; KNL-NEXT:    retq
;
; SKX-LABEL: movmskb:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovb2m %xmm0, %k0
; SKX-NEXT:    kmovw %k0, %eax
; SKX-NEXT:    retq
  %a = icmp slt <16 x i8> %x, zeroinitializer
  %b = bitcast <16 x i1> %a to i16
  %c = zext i16 %b to i32
  ret i32 %c
}

define i32 @movmskb256(<32 x i8> %x) {
; SSE2-LABEL: movmskb256:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pmovmskb %xmm0, %ecx
; SSE2-NEXT:    pmovmskb %xmm1, %eax
; SSE2-NEXT:    shll $16, %eax
; SSE2-NEXT:    orl %ecx, %eax
; SSE2-NEXT:    retq
;
; AVX1-LABEL: movmskb256:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
; AVX1-NEXT:    vpmovmskb %xmm0, %eax
; AVX1-NEXT:    shll $16, %eax
; AVX1-NEXT:    orl %ecx, %eax
; AVX1-NEXT:    vzeroupper
; AVX1-NEXT:    retq
;
; AVX2-LABEL: movmskb256:
; AVX2:       # %bb.0:
; AVX2-NEXT:    vpmovmskb %ymm0, %eax
; AVX2-NEXT:    vzeroupper
; AVX2-NEXT:    retq
;
; KNL-LABEL: movmskb256:
; KNL:       # %bb.0:
; KNL-NEXT:    vpmovmskb %ymm0, %eax
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: movmskb256:
; SKX:       # %bb.0:
; SKX-NEXT:    vpmovb2m %ymm0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    vzeroupper
; SKX-NEXT:    retq
  %a = icmp slt <32 x i8> %x, zeroinitializer
  %b = bitcast <32 x i1> %a to i32
  ret i32 %b
}

; Multiple extract elements from a vector compare.

define i1 @movmsk_v16i8(<16 x i8> %x, <16 x i8> %y) {
; SSE2-LABEL: movmsk_v16i8:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pcmpeqb %xmm1, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %eax
; SSE2-NEXT:    movl %eax, %ecx
; SSE2-NEXT:    shrl $15, %ecx
; SSE2-NEXT:    movl %eax, %edx
; SSE2-NEXT:    shrl $8, %edx
; SSE2-NEXT:    andl $1, %edx
; SSE2-NEXT:    andl $8, %eax
; SSE2-NEXT:    shrl $3, %eax
; SSE2-NEXT:    xorl %edx, %eax
; SSE2-NEXT:    andl %ecx, %eax
; SSE2-NEXT:    # kill: def $al killed $al killed $eax
; SSE2-NEXT:    retq
;
; AVX-LABEL: movmsk_v16i8:
; AVX:       # %bb.0:
; AVX-NEXT:    vpcmpeqb %xmm1, %xmm0, %xmm0
; AVX-NEXT:    vpmovmskb %xmm0, %eax
; AVX-NEXT:    movl %eax, %ecx
; AVX-NEXT:    shrl $15, %ecx
; AVX-NEXT:    movl %eax, %edx
; AVX-NEXT:    shrl $8, %edx
; AVX-NEXT:    andl $1, %edx
; AVX-NEXT:    andl $8, %eax
; AVX-NEXT:    shrl $3, %eax
; AVX-NEXT:    xorl %edx, %eax
; AVX-NEXT:    andl %ecx, %eax
; AVX-NEXT:    # kill: def $al killed $al killed $eax
; AVX-NEXT:    retq
;
; KNL-LABEL: movmsk_v16i8:
; KNL:       # %bb.0:
; KNL-NEXT:    vpcmpeqb %xmm1, %xmm0, %xmm0
; KNL-NEXT:    vpmovsxbd %xmm0, %zmm0
; KNL-NEXT:    vptestmd %zmm0, %zmm0, %k0
; KNL-NEXT:    kshiftrw $15, %k0, %k1
; KNL-NEXT:    kmovw %k1, %ecx
; KNL-NEXT:    kshiftrw $8, %k0, %k1
; KNL-NEXT:    kmovw %k1, %edx
; KNL-NEXT:    kshiftrw $3, %k0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    xorb %dl, %al
; KNL-NEXT:    andb %cl, %al
; KNL-NEXT:    # kill: def $al killed $al killed $eax
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: movmsk_v16i8:
; SKX:       # %bb.0:
; SKX-NEXT:    vpcmpeqb %xmm1, %xmm0, %k0
; SKX-NEXT:    kshiftrw $15, %k0, %k1
; SKX-NEXT:    kmovd %k1, %ecx
; SKX-NEXT:    kshiftrw $8, %k0, %k1
; SKX-NEXT:    kmovd %k1, %edx
; SKX-NEXT:    kshiftrw $3, %k0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    xorb %dl, %al
; SKX-NEXT:    andb %cl, %al
; SKX-NEXT:    # kill: def $al killed $al killed $eax
; SKX-NEXT:    retq
  %cmp = icmp eq <16 x i8> %x, %y
  %e1 = extractelement <16 x i1> %cmp, i32 3
  %e2 = extractelement <16 x i1> %cmp, i32 8
  %e3 = extractelement <16 x i1> %cmp, i32 15
  %u1 = xor i1 %e1, %e2
  %u2 = and i1 %e3, %u1
  ret i1 %u2
}

; TODO: Replace shift+mask chain with NOT+TEST+SETE
define i1 @movmsk_v8i16(<8 x i16> %x, <8 x i16> %y) {
; SSE2-LABEL: movmsk_v8i16:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pcmpgtw %xmm1, %xmm0
; SSE2-NEXT:    packsswb %xmm0, %xmm0
; SSE2-NEXT:    pmovmskb %xmm0, %ecx
; SSE2-NEXT:    movl %ecx, %eax
; SSE2-NEXT:    shrb $7, %al
; SSE2-NEXT:    movl %ecx, %edx
; SSE2-NEXT:    andb $16, %dl
; SSE2-NEXT:    shrb $4, %dl
; SSE2-NEXT:    andb %al, %dl
; SSE2-NEXT:    movl %ecx, %eax
; SSE2-NEXT:    shrb %al
; SSE2-NEXT:    andb %dl, %al
; SSE2-NEXT:    andb %cl, %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: movmsk_v8i16:
; AVX:       # %bb.0:
; AVX-NEXT:    vpcmpgtw %xmm1, %xmm0, %xmm0
; AVX-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
; AVX-NEXT:    vpmovmskb %xmm0, %ecx
; AVX-NEXT:    movl %ecx, %eax
; AVX-NEXT:    shrb $7, %al
; AVX-NEXT:    movl %ecx, %edx
; AVX-NEXT:    andb $16, %dl
; AVX-NEXT:    shrb $4, %dl
; AVX-NEXT:    andb %al, %dl
; AVX-NEXT:    movl %ecx, %eax
; AVX-NEXT:    shrb %al
; AVX-NEXT:    andb %dl, %al
; AVX-NEXT:    andb %cl, %al
; AVX-NEXT:    retq
;
; KNL-LABEL: movmsk_v8i16:
; KNL:       # %bb.0:
; KNL-NEXT:    vpcmpgtw %xmm1, %xmm0, %xmm0
; KNL-NEXT:    vpmovsxwq %xmm0, %zmm0
; KNL-NEXT:    vptestmq %zmm0, %zmm0, %k0
; KNL-NEXT:    kshiftrw $4, %k0, %k1
; KNL-NEXT:    kmovw %k1, %ecx
; KNL-NEXT:    kshiftrw $7, %k0, %k1
; KNL-NEXT:    kmovw %k1, %eax
; KNL-NEXT:    kshiftrw $1, %k0, %k1
; KNL-NEXT:    kmovw %k1, %edx
; KNL-NEXT:    kmovw %k0, %esi
; KNL-NEXT:    andb %cl, %al
; KNL-NEXT:    andb %dl, %al
; KNL-NEXT:    andb %sil, %al
; KNL-NEXT:    # kill: def $al killed $al killed $eax
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: movmsk_v8i16:
; SKX:       # %bb.0:
; SKX-NEXT:    vpcmpgtw %xmm1, %xmm0, %k0
; SKX-NEXT:    kshiftrb $4, %k0, %k1
; SKX-NEXT:    kmovd %k1, %ecx
; SKX-NEXT:    kshiftrb $7, %k0, %k1
; SKX-NEXT:    kmovd %k1, %eax
; SKX-NEXT:    kshiftrb $1, %k0, %k1
; SKX-NEXT:    kmovd %k1, %edx
; SKX-NEXT:    kmovd %k0, %esi
; SKX-NEXT:    andb %cl, %al
; SKX-NEXT:    andb %dl, %al
; SKX-NEXT:    andb %sil, %al
; SKX-NEXT:    # kill: def $al killed $al killed $eax
; SKX-NEXT:    retq
  %cmp = icmp sgt <8 x i16> %x, %y
  %e1 = extractelement <8 x i1> %cmp, i32 0
  %e2 = extractelement <8 x i1> %cmp, i32 1
  %e3 = extractelement <8 x i1> %cmp, i32 7
  %e4 = extractelement <8 x i1> %cmp, i32 4
  %u1 = and i1 %e1, %e2
  %u2 = and i1 %e3, %e4
  %u3 = and i1 %u1, %u2
  ret i1 %u3
}

; TODO: Replace shift+mask chain with AND+CMP.
define i1 @movmsk_v4i32(<4 x i32> %x, <4 x i32> %y) {
; SSE2-LABEL: movmsk_v4i32:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pcmpgtd %xmm0, %xmm1
; SSE2-NEXT:    movmskps %xmm1, %eax
; SSE2-NEXT:    movl %eax, %ecx
; SSE2-NEXT:    shrb $3, %cl
; SSE2-NEXT:    andb $4, %al
; SSE2-NEXT:    shrb $2, %al
; SSE2-NEXT:    xorb %cl, %al
; SSE2-NEXT:    # kill: def $al killed $al killed $eax
; SSE2-NEXT:    retq
;
; AVX-LABEL: movmsk_v4i32:
; AVX:       # %bb.0:
; AVX-NEXT:    vpcmpgtd %xmm0, %xmm1, %xmm0
; AVX-NEXT:    vmovmskps %xmm0, %eax
; AVX-NEXT:    movl %eax, %ecx
; AVX-NEXT:    shrb $3, %cl
; AVX-NEXT:    andb $4, %al
; AVX-NEXT:    shrb $2, %al
; AVX-NEXT:    xorb %cl, %al
; AVX-NEXT:    # kill: def $al killed $al killed $eax
; AVX-NEXT:    retq
;
; KNL-LABEL: movmsk_v4i32:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm1 killed $xmm1 def $zmm1
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vpcmpgtd %zmm0, %zmm1, %k0
; KNL-NEXT:    kshiftrw $3, %k0, %k1
; KNL-NEXT:    kmovw %k1, %ecx
; KNL-NEXT:    kshiftrw $2, %k0, %k0
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    xorb %cl, %al
; KNL-NEXT:    # kill: def $al killed $al killed $eax
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: movmsk_v4i32:
; SKX:       # %bb.0:
; SKX-NEXT:    vpcmpgtd %xmm0, %xmm1, %k0
; SKX-NEXT:    kshiftrb $3, %k0, %k1
; SKX-NEXT:    kmovd %k1, %ecx
; SKX-NEXT:    kshiftrb $2, %k0, %k0
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    xorb %cl, %al
; SKX-NEXT:    # kill: def $al killed $al killed $eax
; SKX-NEXT:    retq
  %cmp = icmp slt <4 x i32> %x, %y
  %e1 = extractelement <4 x i1> %cmp, i32 2
  %e2 = extractelement <4 x i1> %cmp, i32 3
  %u1 = xor i1 %e1, %e2
  ret i1 %u1
}

define i1 @movmsk_v2i64(<2 x i64> %x, <2 x i64> %y) {
; SSE2-LABEL: movmsk_v2i64:
; SSE2:       # %bb.0:
; SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,0,3,2]
; SSE2-NEXT:    pand %xmm0, %xmm1
; SSE2-NEXT:    movmskpd %xmm1, %eax
; SSE2-NEXT:    xorl $3, %eax
; SSE2-NEXT:    cmpb $3, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: movmsk_v2i64:
; AVX:       # %bb.0:
; AVX-NEXT:    vpcmpeqq %xmm1, %xmm0, %xmm0
; AVX-NEXT:    vmovmskpd %xmm0, %eax
; AVX-NEXT:    xorl $3, %eax
; AVX-NEXT:    cmpb $3, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: movmsk_v2i64:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm1 killed $xmm1 def $zmm1
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vpcmpneqq %zmm1, %zmm0, %k0
; KNL-NEXT:    kshiftrw $1, %k0, %k1
; KNL-NEXT:    kmovw %k1, %ecx
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    andb %cl, %al
; KNL-NEXT:    # kill: def $al killed $al killed $eax
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: movmsk_v2i64:
; SKX:       # %bb.0:
; SKX-NEXT:    vpcmpneqq %xmm1, %xmm0, %k0
; SKX-NEXT:    kshiftrb $1, %k0, %k1
; SKX-NEXT:    kmovd %k1, %ecx
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    andb %cl, %al
; SKX-NEXT:    # kill: def $al killed $al killed $eax
; SKX-NEXT:    retq
  %cmp = icmp ne <2 x i64> %x, %y
  %e1 = extractelement <2 x i1> %cmp, i32 0
  %e2 = extractelement <2 x i1> %cmp, i32 1
  %u1 = and i1 %e1, %e2
  ret i1 %u1
}

define i1 @movmsk_v4f32(<4 x float> %x, <4 x float> %y) {
; SSE2-LABEL: movmsk_v4f32:
; SSE2:       # %bb.0:
; SSE2-NEXT:    movaps %xmm0, %xmm2
; SSE2-NEXT:    cmpeqps %xmm1, %xmm2
; SSE2-NEXT:    cmpunordps %xmm1, %xmm0
; SSE2-NEXT:    orps %xmm2, %xmm0
; SSE2-NEXT:    movmskps %xmm0, %eax
; SSE2-NEXT:    testb $14, %al
; SSE2-NEXT:    setne %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: movmsk_v4f32:
; AVX:       # %bb.0:
; AVX-NEXT:    vcmpeq_uqps %xmm1, %xmm0, %xmm0
; AVX-NEXT:    vmovmskps %xmm0, %eax
; AVX-NEXT:    testb $14, %al
; AVX-NEXT:    setne %al
; AVX-NEXT:    retq
;
; KNL-LABEL: movmsk_v4f32:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm1 killed $xmm1 def $zmm1
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vcmpeq_uqps %zmm1, %zmm0, %k0
; KNL-NEXT:    kshiftrw $3, %k0, %k1
; KNL-NEXT:    kmovw %k1, %ecx
; KNL-NEXT:    kshiftrw $2, %k0, %k1
; KNL-NEXT:    kmovw %k1, %eax
; KNL-NEXT:    kshiftrw $1, %k0, %k0
; KNL-NEXT:    kmovw %k0, %edx
; KNL-NEXT:    orb %cl, %al
; KNL-NEXT:    orb %dl, %al
; KNL-NEXT:    # kill: def $al killed $al killed $eax
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: movmsk_v4f32:
; SKX:       # %bb.0:
; SKX-NEXT:    vcmpeq_uqps %xmm1, %xmm0, %k0
; SKX-NEXT:    kshiftrb $3, %k0, %k1
; SKX-NEXT:    kmovd %k1, %ecx
; SKX-NEXT:    kshiftrb $2, %k0, %k1
; SKX-NEXT:    kmovd %k1, %eax
; SKX-NEXT:    kshiftrb $1, %k0, %k0
; SKX-NEXT:    kmovd %k0, %edx
; SKX-NEXT:    orb %cl, %al
; SKX-NEXT:    orb %dl, %al
; SKX-NEXT:    # kill: def $al killed $al killed $eax
; SKX-NEXT:    retq
  %cmp = fcmp ueq <4 x float> %x, %y
  %e1 = extractelement <4 x i1> %cmp, i32 1
  %e2 = extractelement <4 x i1> %cmp, i32 2
  %e3 = extractelement <4 x i1> %cmp, i32 3
  %u1 = or i1 %e1, %e2
  %u2 = or i1 %u1, %e3
  ret i1 %u2
}

define i1 @movmsk_v2f64(<2 x double> %x, <2 x double> %y) {
; SSE2-LABEL: movmsk_v2f64:
; SSE2:       # %bb.0:
; SSE2-NEXT:    cmplepd %xmm0, %xmm1
; SSE2-NEXT:    movmskpd %xmm1, %eax
; SSE2-NEXT:    cmpb $3, %al
; SSE2-NEXT:    sete %al
; SSE2-NEXT:    retq
;
; AVX-LABEL: movmsk_v2f64:
; AVX:       # %bb.0:
; AVX-NEXT:    vcmplepd %xmm0, %xmm1, %xmm0
; AVX-NEXT:    vmovmskpd %xmm0, %eax
; AVX-NEXT:    cmpb $3, %al
; AVX-NEXT:    sete %al
; AVX-NEXT:    retq
;
; KNL-LABEL: movmsk_v2f64:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm1 killed $xmm1 def $zmm1
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vcmplepd %zmm0, %zmm1, %k0
; KNL-NEXT:    kshiftrw $1, %k0, %k1
; KNL-NEXT:    kmovw %k1, %ecx
; KNL-NEXT:    kmovw %k0, %eax
; KNL-NEXT:    andb %cl, %al
; KNL-NEXT:    # kill: def $al killed $al killed $eax
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: movmsk_v2f64:
; SKX:       # %bb.0:
; SKX-NEXT:    vcmplepd %xmm0, %xmm1, %k0
; SKX-NEXT:    kshiftrb $1, %k0, %k1
; SKX-NEXT:    kmovd %k1, %ecx
; SKX-NEXT:    kmovd %k0, %eax
; SKX-NEXT:    andb %cl, %al
; SKX-NEXT:    # kill: def $al killed $al killed $eax
; SKX-NEXT:    retq
  %cmp = fcmp oge <2 x double> %x, %y
  %e1 = extractelement <2 x i1> %cmp, i32 0
  %e2 = extractelement <2 x i1> %cmp, i32 1
  %u1 = and i1 %e1, %e2
  ret i1 %u1
}

define i32 @PR39665_c_ray(<2 x double> %x, <2 x double> %y) {
; SSE2-LABEL: PR39665_c_ray:
; SSE2:       # %bb.0:
; SSE2-NEXT:    cmpltpd %xmm0, %xmm1
; SSE2-NEXT:    movmskpd %xmm1, %eax
; SSE2-NEXT:    cmpb $3, %al
; SSE2-NEXT:    movl $42, %ecx
; SSE2-NEXT:    movl $99, %eax
; SSE2-NEXT:    cmovel %ecx, %eax
; SSE2-NEXT:    retq
;
; AVX-LABEL: PR39665_c_ray:
; AVX:       # %bb.0:
; AVX-NEXT:    vcmpltpd %xmm0, %xmm1, %xmm0
; AVX-NEXT:    vmovmskpd %xmm0, %eax
; AVX-NEXT:    cmpb $3, %al
; AVX-NEXT:    movl $42, %ecx
; AVX-NEXT:    movl $99, %eax
; AVX-NEXT:    cmovel %ecx, %eax
; AVX-NEXT:    retq
;
; KNL-LABEL: PR39665_c_ray:
; KNL:       # %bb.0:
; KNL-NEXT:    # kill: def $xmm1 killed $xmm1 def $zmm1
; KNL-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; KNL-NEXT:    vcmpltpd %zmm0, %zmm1, %k0
; KNL-NEXT:    kshiftrw $1, %k0, %k1
; KNL-NEXT:    kmovw %k1, %eax
; KNL-NEXT:    kmovw %k0, %ecx
; KNL-NEXT:    testb $1, %al
; KNL-NEXT:    movl $42, %eax
; KNL-NEXT:    movl $99, %edx
; KNL-NEXT:    cmovel %edx, %eax
; KNL-NEXT:    testb $1, %cl
; KNL-NEXT:    cmovel %edx, %eax
; KNL-NEXT:    vzeroupper
; KNL-NEXT:    retq
;
; SKX-LABEL: PR39665_c_ray:
; SKX:       # %bb.0:
; SKX-NEXT:    vcmpltpd %xmm0, %xmm1, %k0
; SKX-NEXT:    kshiftrb $1, %k0, %k1
; SKX-NEXT:    kmovd %k1, %eax
; SKX-NEXT:    kmovd %k0, %ecx
; SKX-NEXT:    testb $1, %al
; SKX-NEXT:    movl $42, %eax
; SKX-NEXT:    movl $99, %edx
; SKX-NEXT:    cmovel %edx, %eax
; SKX-NEXT:    testb $1, %cl
; SKX-NEXT:    cmovel %edx, %eax
; SKX-NEXT:    retq
  %cmp = fcmp ogt <2 x double> %x, %y
  %e1 = extractelement <2 x i1> %cmp, i32 0
  %e2 = extractelement <2 x i1> %cmp, i32 1
  %u = and i1 %e1, %e2
  %r = select i1 %u, i32 42, i32 99
  ret i32 %r
}