[X86] Add test cases inspired by PR38840.

These are test cases inspired by sequences like below for extracting the same bit from every vector element and checking for all zeros/ones.

define i1 @and256_x8(<8 x i32>) {
    %a = trunc <8 x i32> %0 to <8 x i1>
    %b = bitcast <8 x i1> %a to i8
    %d = icmp eq i8 %b, -1
    ret i1 %d
}

This is what the above looks like after InstCombine.

define i1 @and256_x8_opt(<8 x i32>) {
  %2 = and <8 x i32> %0, <i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1, i32 1>
  %a = icmp ne <8 x i32> %2, zeroinitializer
  %b = bitcast <8 x i1> %a to i8
  %d = icmp eq i8 %b, -1
  ret i1 %d
}

llvm-svn: 341908

1 files changed, 4058 insertions, 0 deletions

diff --git a/llvm/test/CodeGen/X86/movmsk-cmp.ll b/llvm/test/CodeGen/X86/movmsk-cmp.ll
new file mode 100644
index 00000000000..ebfa03a0209
--- /dev/null
+++ b/llvm/test/CodeGen/X86/movmsk-cmp.ll
@@ -0,0 +1,4058 @@
+; 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=+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:    cmpl $65535, %eax # imm = 0xFFFF
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX-LABEL: allones_v16i8_sign:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpmovmskb %xmm0, %eax
+; AVX-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; AVX-NEXT:    sete %al
+; AVX-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
+;
+; 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:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpgtb %xmm0, %xmm1, %xmm2
+; AVX1-NEXT:    vpmovmskb %xmm2, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpcmpgtb %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_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
+;
+; 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:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpgtb %xmm0, %xmm1, %xmm2
+; AVX1-NEXT:    vpmovmskb %xmm2, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpcmpgtb %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_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
+;
+; 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:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpgtb %xmm0, %xmm2, %xmm3
+; AVX1-NEXT:    vpmovmskb %xmm3, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpcmpgtb %xmm0, %xmm2, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
+; AVX1-NEXT:    shll $16, %ecx
+; AVX1-NEXT:    orl %eax, %ecx
+; AVX1-NEXT:    vpcmpgtb %xmm1, %xmm2, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
+; AVX1-NEXT:    vpcmpgtb %xmm0, %xmm2, %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
+;
+; 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:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpgtb %xmm0, %xmm2, %xmm3
+; AVX1-NEXT:    vpmovmskb %xmm3, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpcmpgtb %xmm0, %xmm2, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
+; AVX1-NEXT:    shll $16, %ecx
+; AVX1-NEXT:    orl %eax, %ecx
+; AVX1-NEXT:    vpcmpgtb %xmm1, %xmm2, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
+; AVX1-NEXT:    vpcmpgtb %xmm0, %xmm2, %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
+;
+; 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:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpgtw %xmm0, %xmm1
+; SSE2-NEXT:    packsswb %xmm0, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %eax
+; SSE2-NEXT:    cmpb $-1, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX-LABEL: allones_v8i16_sign:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vpcmpgtw %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
+; AVX-NEXT:    vpmovmskb %xmm0, %eax
+; AVX-NEXT:    cmpb $-1, %al
+; AVX-NEXT:    sete %al
+; AVX-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:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpgtw %xmm0, %xmm1
+; SSE2-NEXT:    packsswb %xmm0, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %eax
+; SSE2-NEXT:    testb %al, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX-LABEL: allzeros_v8i16_sign:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vpcmpgtw %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
+; AVX-NEXT:    vpmovmskb %xmm0, %eax
+; AVX-NEXT:    testb %al, %al
+; AVX-NEXT:    sete %al
+; AVX-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:    pxor %xmm2, %xmm2
+; SSE2-NEXT:    pxor %xmm3, %xmm3
+; SSE2-NEXT:    pcmpgtw %xmm1, %xmm3
+; SSE2-NEXT:    pcmpgtw %xmm0, %xmm2
+; SSE2-NEXT:    packsswb %xmm3, %xmm2
+; SSE2-NEXT:    pmovmskb %xmm2, %eax
+; SSE2-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allones_v16i16_sign:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpgtw %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vpcmpgtw %xmm0, %xmm2, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; 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:    cmpl $65535, %eax # imm = 0xFFFF
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    pxor %xmm2, %xmm2
+; SSE2-NEXT:    pxor %xmm3, %xmm3
+; SSE2-NEXT:    pcmpgtw %xmm1, %xmm3
+; SSE2-NEXT:    pcmpgtw %xmm0, %xmm2
+; SSE2-NEXT:    packsswb %xmm3, %xmm2
+; SSE2-NEXT:    pmovmskb %xmm2, %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:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpgtw %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vpcmpgtw %xmm0, %xmm2, %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_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
+;
+; 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:    pxor %xmm4, %xmm4
+; SSE2-NEXT:    pxor %xmm5, %xmm5
+; SSE2-NEXT:    pcmpgtw %xmm1, %xmm5
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpgtw %xmm0, %xmm1
+; SSE2-NEXT:    packsswb %xmm5, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %eax
+; SSE2-NEXT:    pxor %xmm0, %xmm0
+; SSE2-NEXT:    pcmpgtw %xmm3, %xmm0
+; SSE2-NEXT:    pcmpgtw %xmm2, %xmm4
+; SSE2-NEXT:    packsswb %xmm0, %xmm4
+; SSE2-NEXT:    pmovmskb %xmm4, %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:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpgtw %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vpcmpgtw %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
+; AVX1-NEXT:    vpcmpgtw %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpcmpgtw %xmm1, %xmm3, %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_sign:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX2-NEXT:    vpcmpgtw %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vpcmpgtw %ymm0, %ymm2, %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
+;
+; 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:    pxor %xmm4, %xmm4
+; SSE2-NEXT:    pxor %xmm5, %xmm5
+; SSE2-NEXT:    pcmpgtw %xmm1, %xmm5
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpgtw %xmm0, %xmm1
+; SSE2-NEXT:    packsswb %xmm5, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %eax
+; SSE2-NEXT:    pxor %xmm0, %xmm0
+; SSE2-NEXT:    pcmpgtw %xmm3, %xmm0
+; SSE2-NEXT:    pcmpgtw %xmm2, %xmm4
+; SSE2-NEXT:    packsswb %xmm0, %xmm4
+; SSE2-NEXT:    pmovmskb %xmm4, %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:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpgtw %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vpcmpgtw %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
+; AVX1-NEXT:    vpcmpgtw %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpcmpgtw %xmm1, %xmm3, %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_sign:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX2-NEXT:    vpcmpgtw %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vpcmpgtw %ymm0, %ymm2, %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
+;
+; 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:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    cmpb $15, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX-LABEL: allones_v4i32_sign:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vpcmpgtd %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vmovmskps %xmm0, %eax
+; AVX-NEXT:    cmpb $15, %al
+; AVX-NEXT:    sete %al
+; AVX-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:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    testb %al, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX-LABEL: allzeros_v4i32_sign:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vpcmpgtd %xmm0, %xmm1, %xmm0
+; AVX-NEXT:    vmovmskps %xmm0, %eax
+; AVX-NEXT:    testb %al, %al
+; AVX-NEXT:    sete %al
+; AVX-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:    pxor %xmm2, %xmm2
+; SSE2-NEXT:    pxor %xmm3, %xmm3
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm2
+; SSE2-NEXT:    packssdw %xmm3, %xmm2
+; SSE2-NEXT:    packsswb %xmm0, %xmm2
+; SSE2-NEXT:    pmovmskb %xmm2, %eax
+; SSE2-NEXT:    cmpb $-1, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allones_v8i32_sign:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpgtd %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vpcmpgtd %xmm0, %xmm2, %xmm0
+; 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_v8i32_sign:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX2-NEXT:    vpcmpgtd %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vmovmskps %ymm0, %eax
+; AVX2-NEXT:    cmpb $-1, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    pxor %xmm2, %xmm2
+; SSE2-NEXT:    pxor %xmm3, %xmm3
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm2
+; SSE2-NEXT:    packssdw %xmm3, %xmm2
+; SSE2-NEXT:    packsswb %xmm0, %xmm2
+; SSE2-NEXT:    pmovmskb %xmm2, %eax
+; SSE2-NEXT:    testb %al, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allzeros_v8i32_sign:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpgtd %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vpcmpgtd %xmm0, %xmm2, %xmm0
+; 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_v8i32_sign:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX2-NEXT:    vpcmpgtd %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vmovmskps %ymm0, %eax
+; AVX2-NEXT:    testb %al, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    pxor %xmm4, %xmm4
+; SSE2-NEXT:    pxor %xmm5, %xmm5
+; SSE2-NEXT:    pcmpgtd %xmm3, %xmm5
+; SSE2-NEXT:    pxor %xmm3, %xmm3
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm3
+; SSE2-NEXT:    packssdw %xmm5, %xmm3
+; SSE2-NEXT:    pxor %xmm2, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE2-NEXT:    packssdw %xmm2, %xmm4
+; SSE2-NEXT:    packsswb %xmm3, %xmm4
+; SSE2-NEXT:    pmovmskb %xmm4, %eax
+; SSE2-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allones_v16i32_sign:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpgtd %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vpcmpgtd %xmm1, %xmm3, %xmm1
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpcmpgtd %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vpcmpgtd %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; 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:    cmpl $65535, %eax # imm = 0xFFFF
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    pxor %xmm4, %xmm4
+; SSE2-NEXT:    pxor %xmm5, %xmm5
+; SSE2-NEXT:    pcmpgtd %xmm3, %xmm5
+; SSE2-NEXT:    pxor %xmm3, %xmm3
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm3
+; SSE2-NEXT:    packssdw %xmm5, %xmm3
+; SSE2-NEXT:    pxor %xmm2, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE2-NEXT:    packssdw %xmm2, %xmm4
+; SSE2-NEXT:    packsswb %xmm3, %xmm4
+; SSE2-NEXT:    pmovmskb %xmm4, %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:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpgtd %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vpcmpgtd %xmm1, %xmm3, %xmm1
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpcmpgtd %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vpcmpgtd %xmm0, %xmm3, %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_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
+;
+; 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:    movdqa {{.*#+}} xmm2 = [2147483648,0,2147483648,0]
+; SSE2-NEXT:    pxor %xmm2, %xmm1
+; SSE2-NEXT:    movdqa %xmm2, %xmm3
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm3
+; SSE2-NEXT:    pxor %xmm2, %xmm0
+; SSE2-NEXT:    movdqa %xmm2, %xmm1
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSE2-NEXT:    por %xmm0, %xmm1
+; SSE2-NEXT:    packssdw %xmm3, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    cmpb $15, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allones_v4i64_sign:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpgtq %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vpcmpgtq %xmm0, %xmm2, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %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_sign:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX2-NEXT:    vpcmpgtq %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vmovmskpd %ymm0, %eax
+; AVX2-NEXT:    cmpb $15, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    movdqa {{.*#+}} xmm2 = [2147483648,0,2147483648,0]
+; SSE2-NEXT:    pxor %xmm2, %xmm1
+; SSE2-NEXT:    movdqa %xmm2, %xmm3
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm3
+; SSE2-NEXT:    pxor %xmm2, %xmm0
+; SSE2-NEXT:    movdqa %xmm2, %xmm1
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSE2-NEXT:    por %xmm0, %xmm1
+; SSE2-NEXT:    packssdw %xmm3, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    testb %al, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allzeros_v4i64_sign:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpgtq %xmm1, %xmm2, %xmm1
+; AVX1-NEXT:    vpcmpgtq %xmm0, %xmm2, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %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_sign:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX2-NEXT:    vpcmpgtq %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vmovmskpd %ymm0, %eax
+; AVX2-NEXT:    testb %al, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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 = [2147483648,0,2147483648,0]
+; 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, %ymm1, %xmm2
+; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vpcmpgtq %xmm1, %xmm3, %xmm1
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vpcmpgtq %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
+; 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:    vpxor %xmm2, %xmm2, %xmm2
+; AVX2-NEXT:    vpcmpgtq %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vpcmpgtq %ymm0, %ymm2, %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
+;
+; 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 = [2147483648,0,2147483648,0]
+; 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, %ymm1, %xmm2
+; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vpcmpgtq %xmm1, %xmm3, %xmm1
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpcmpgtq %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vpcmpgtq %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
+; 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:    vpxor %xmm2, %xmm2, %xmm2
+; AVX2-NEXT:    vpcmpgtq %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vpcmpgtq %ymm0, %ymm2, %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
+;
+; 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:    movdqa {{.*#+}} xmm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pcmpeqb %xmm1, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX-LABEL: allones_v16i8_and1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpcmpeqb %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpmovmskb %xmm0, %eax
+; AVX-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; AVX-NEXT:    sete %al
+; AVX-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:    movdqa {{.*#+}} xmm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pcmpeqb %xmm1, %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:    vmovdqa {{.*#+}} xmm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpcmpeqb %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpmovmskb %xmm0, %eax
+; AVX-NEXT:    testw %ax, %ax
+; AVX-NEXT:    sete %al
+; AVX-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:    movdqa {{.*#+}} xmm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pcmpeqb %xmm2, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    pcmpeqb %xmm2, %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:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqb %xmm1, %xmm0, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpmovmskb %xmm2, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpcmpeqb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %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:    vmovdqa {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqb %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpmovmskb %ymm0, %eax
+; AVX2-NEXT:    cmpl $-1, %eax
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    movdqa {{.*#+}} xmm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pcmpeqb %xmm2, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    pcmpeqb %xmm2, %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:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqb %xmm1, %xmm0, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpmovmskb %xmm2, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpcmpeqb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %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:    vmovdqa {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqb %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpmovmskb %ymm0, %eax
+; AVX2-NEXT:    testl %eax, %eax
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    movdqa {{.*#+}} xmm4 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pcmpeqb %xmm4, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    pcmpeqb %xmm4, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %ecx
+; SSE2-NEXT:    shll $16, %ecx
+; SSE2-NEXT:    orl %eax, %ecx
+; SSE2-NEXT:    pcmpeqb %xmm4, %xmm2
+; SSE2-NEXT:    pmovmskb %xmm2, %eax
+; SSE2-NEXT:    pcmpeqb %xmm4, %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:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vxorps %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm3
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm3, %xmm3
+; AVX1-NEXT:    vpmovmskb %xmm3, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
+; AVX1-NEXT:    shll $16, %ecx
+; AVX1-NEXT:    orl %eax, %ecx
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm1, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %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:    vmovdqa {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqb %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpmovmskb %ymm0, %eax
+; AVX2-NEXT:    vpcmpeqb %ymm2, %ymm1, %ymm0
+; AVX2-NEXT:    vpmovmskb %ymm0, %ecx
+; AVX2-NEXT:    shlq $32, %rcx
+; AVX2-NEXT:    orq %rax, %rcx
+; AVX2-NEXT:    cmpq $-1, %rcx
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    movdqa {{.*#+}} xmm4 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pcmpeqb %xmm4, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    pcmpeqb %xmm4, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %ecx
+; SSE2-NEXT:    shll $16, %ecx
+; SSE2-NEXT:    orl %eax, %ecx
+; SSE2-NEXT:    pcmpeqb %xmm4, %xmm2
+; SSE2-NEXT:    pmovmskb %xmm2, %eax
+; SSE2-NEXT:    pcmpeqb %xmm4, %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:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vxorps %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm3
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm3, %xmm3
+; AVX1-NEXT:    vpmovmskb %xmm3, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
+; AVX1-NEXT:    shll $16, %ecx
+; AVX1-NEXT:    orl %eax, %ecx
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm1, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %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:    vmovdqa {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqb %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpmovmskb %ymm0, %eax
+; AVX2-NEXT:    vpcmpeqb %ymm2, %ymm1, %ymm0
+; AVX2-NEXT:    vpmovmskb %ymm0, %ecx
+; AVX2-NEXT:    shlq $32, %rcx
+; AVX2-NEXT:    orq %rax, %rcx
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    movdqa {{.*#+}} xmm1 = [1,1,1,1,1,1,1,1]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pcmpeqw %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_v8i16_and1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,1,1,1,1,1,1,1]
+; AVX-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpcmpeqw %xmm1, %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
+;
+; 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:    movdqa {{.*#+}} xmm1 = [1,1,1,1,1,1,1,1]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pcmpeqw %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_v8i16_and1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,1,1,1,1,1,1,1]
+; AVX-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpcmpeqw %xmm1, %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
+;
+; 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:    movdqa {{.*#+}} xmm2 = [1,1,1,1,1,1,1,1]
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm2, %xmm0
+; SSE2-NEXT:    packsswb %xmm1, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allones_v16i16_and1:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqw %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: allones_v16i16_and1:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqw %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpmovmskb %xmm0, %eax
+; AVX2-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    movdqa {{.*#+}} xmm4 = [1,1,1,1,1,1,1,1]
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm4, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm4, %xmm0
+; SSE2-NEXT:    packsswb %xmm1, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    pcmpeqw %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqw %xmm4, %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:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpxor %xmm4, %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:    vmovdqa {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqw %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqw %ymm2, %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
+;
+; 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:    movdqa {{.*#+}} xmm4 = [1,1,1,1,1,1,1,1]
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm4, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm4, %xmm0
+; SSE2-NEXT:    packsswb %xmm1, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    pcmpeqw %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqw %xmm4, %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:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpxor %xmm4, %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:    vmovdqa {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqw %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqw %ymm2, %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
+;
+; 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:    movdqa {{.*#+}} xmm2 = [1,1,1,1,1,1,1,1]
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm2, %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:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqw %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %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:    vmovdqa {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqw %ymm1, %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
+;
+; 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:    movdqa {{.*#+}} xmm1 = [1,1,1,1]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSE2-NEXT:    movmskps %xmm0, %eax
+; SSE2-NEXT:    cmpb $15, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allones_v4i32_and1:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,1,1,1]
+; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpcmpeqd %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vmovmskps %xmm0, %eax
+; AVX1-NEXT:    cmpb $15, %al
+; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: allones_v4i32_and1:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [1,1,1,1]
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpcmpeqd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vmovmskps %xmm0, %eax
+; AVX2-NEXT:    cmpb $15, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    retq
+;
+; SKX-LABEL: allones_v4i32_and1:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [1,1,1,1]
+; SKX-NEXT:    vptestmd %xmm1, %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:    movdqa {{.*#+}} xmm1 = [1,1,1,1]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSE2-NEXT:    movmskps %xmm0, %eax
+; SSE2-NEXT:    testb %al, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allzeros_v4i32_and1:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,1,1,1]
+; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpcmpeqd %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vmovmskps %xmm0, %eax
+; AVX1-NEXT:    testb %al, %al
+; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: allzeros_v4i32_and1:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [1,1,1,1]
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpcmpeqd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vmovmskps %xmm0, %eax
+; AVX2-NEXT:    testb %al, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    retq
+;
+; SKX-LABEL: allzeros_v4i32_and1:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [1,1,1,1]
+; SKX-NEXT:    vptestmd %xmm1, %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:    movdqa {{.*#+}} xmm2 = [1,1,1,1]
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm2, %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:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm0
+; 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_v8i32_and1:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqd %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vmovmskps %ymm0, %eax
+; AVX2-NEXT:    cmpb $-1, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
+;
+; SKX-LABEL: allones_v8i32_and1:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
+; SKX-NEXT:    vptestmd %ymm1, %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:    movdqa {{.*#+}} xmm2 = [1,1,1,1]
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm2, %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:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm0
+; 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_v8i32_and1:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqd %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vmovmskps %ymm0, %eax
+; AVX2-NEXT:    testb %al, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
+;
+; SKX-LABEL: allzeros_v8i32_and1:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
+; SKX-NEXT:    vptestmd %ymm1, %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:    movdqa {{.*#+}} xmm4 = [1,1,1,1]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm2
+; SSE2-NEXT:    packssdw %xmm3, %xmm2
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm0
+; SSE2-NEXT:    packssdw %xmm1, %xmm0
+; SSE2-NEXT:    packsswb %xmm2, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allones_v16i32_and1:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpxor %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: allones_v16i32_and1:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqd %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqd %ymm2, %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:    cmpl $65535, %eax # imm = 0xFFFF
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
+;
+; SKX-LABEL: allones_v16i32_and1:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpbroadcastd {{.*#+}} zmm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; SKX-NEXT:    vptestmd %zmm1, %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:    movdqa {{.*#+}} xmm4 = [1,1,1,1]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm2
+; SSE2-NEXT:    packssdw %xmm3, %xmm2
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm4, %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:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpxor %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %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:    vpbroadcastd {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqd %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqd %ymm2, %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
+;
+; SKX-LABEL: allzeros_v16i32_and1:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpbroadcastd {{.*#+}} zmm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; SKX-NEXT:    vptestmd %zmm1, %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:    movdqa {{.*#+}} xmm1 = [1,1]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; 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:    cmpb $3, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX-LABEL: allones_v2i64_and1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,1]
+; AVX-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpcmpeqq %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vmovmskpd %xmm0, %eax
+; AVX-NEXT:    cmpb $3, %al
+; AVX-NEXT:    sete %al
+; AVX-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:    movdqa {{.*#+}} xmm1 = [1,1]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; 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:    testb %al, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX-LABEL: allzeros_v2i64_and1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [1,1]
+; AVX-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpcmpeqq %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vmovmskpd %xmm0, %eax
+; AVX-NEXT:    testb %al, %al
+; AVX-NEXT:    sete %al
+; AVX-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:    movdqa {{.*#+}} xmm2 = [1,1]
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[1,0,3,2]
+; SSE2-NEXT:    pand %xmm1, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,0,3,2]
+; SSE2-NEXT:    pand %xmm0, %xmm1
+; SSE2-NEXT:    packssdw %xmm3, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    cmpb $15, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allones_v4i64_and1:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqq %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %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:    vpbroadcastq {{.*#+}} ymm1 = [1,1,1,1]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqq %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vmovmskpd %ymm0, %eax
+; AVX2-NEXT:    cmpb $15, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    movdqa {{.*#+}} xmm2 = [1,1]
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[1,0,3,2]
+; SSE2-NEXT:    pand %xmm1, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,0,3,2]
+; SSE2-NEXT:    pand %xmm0, %xmm1
+; SSE2-NEXT:    packssdw %xmm3, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    testb %al, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allzeros_v4i64_and1:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqq %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %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:    vpbroadcastq {{.*#+}} ymm1 = [1,1,1,1]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqq %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vmovmskpd %ymm0, %eax
+; AVX2-NEXT:    testb %al, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    movdqa {{.*#+}} xmm4 = [1,1]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm3[1,0,3,2]
+; SSE2-NEXT:    pand %xmm3, %xmm5
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,0,3,2]
+; SSE2-NEXT:    pand %xmm2, %xmm3
+; SSE2-NEXT:    packssdw %xmm5, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,0,3,2]
+; SSE2-NEXT:    pand %xmm1, %xmm2
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,0,3,2]
+; SSE2-NEXT:    pand %xmm0, %xmm1
+; SSE2-NEXT:    packssdw %xmm2, %xmm1
+; SSE2-NEXT:    packssdw %xmm3, %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_and1:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1]
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpxor %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
+; 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:    vpbroadcastq {{.*#+}} ymm2 = [1,1,1,1]
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqq %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqq %ymm2, %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
+;
+; 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:    movdqa {{.*#+}} xmm4 = [1,1]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm3[1,0,3,2]
+; SSE2-NEXT:    pand %xmm3, %xmm5
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,0,3,2]
+; SSE2-NEXT:    pand %xmm2, %xmm3
+; SSE2-NEXT:    packssdw %xmm5, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,0,3,2]
+; SSE2-NEXT:    pand %xmm1, %xmm2
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,0,3,2]
+; SSE2-NEXT:    pand %xmm0, %xmm1
+; SSE2-NEXT:    packssdw %xmm2, %xmm1
+; SSE2-NEXT:    packssdw %xmm3, %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_and1:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1]
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpxor %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
+; 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:    vpbroadcastq {{.*#+}} ymm2 = [1,1,1,1]
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqq %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqq %ymm2, %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
+;
+; 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:    movdqa {{.*#+}} xmm1 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pcmpeqb %xmm1, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX-LABEL: allones_v16i8_and4:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; AVX-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpcmpeqb %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpmovmskb %xmm0, %eax
+; AVX-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; AVX-NEXT:    sete %al
+; AVX-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:    movdqa {{.*#+}} xmm1 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pcmpeqb %xmm1, %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:    vmovdqa {{.*#+}} xmm1 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; AVX-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpcmpeqb %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpmovmskb %xmm0, %eax
+; AVX-NEXT:    testw %ax, %ax
+; AVX-NEXT:    sete %al
+; AVX-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:    movdqa {{.*#+}} xmm2 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pcmpeqb %xmm2, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    pcmpeqb %xmm2, %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:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqb %xmm1, %xmm0, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpmovmskb %xmm2, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpcmpeqb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %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:    vmovdqa {{.*#+}} ymm1 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqb %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpmovmskb %ymm0, %eax
+; AVX2-NEXT:    cmpl $-1, %eax
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    movdqa {{.*#+}} xmm2 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pcmpeqb %xmm2, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    pcmpeqb %xmm2, %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:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqb %xmm1, %xmm0, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpmovmskb %xmm2, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpcmpeqb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %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:    vmovdqa {{.*#+}} ymm1 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqb %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpmovmskb %ymm0, %eax
+; AVX2-NEXT:    testl %eax, %eax
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    movdqa {{.*#+}} xmm4 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pcmpeqb %xmm4, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    pcmpeqb %xmm4, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %ecx
+; SSE2-NEXT:    shll $16, %ecx
+; SSE2-NEXT:    orl %eax, %ecx
+; SSE2-NEXT:    pcmpeqb %xmm4, %xmm2
+; SSE2-NEXT:    pmovmskb %xmm2, %eax
+; SSE2-NEXT:    pcmpeqb %xmm4, %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:    vmovaps {{.*#+}} ymm2 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vxorps %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm3
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm3, %xmm3
+; AVX1-NEXT:    vpmovmskb %xmm3, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
+; AVX1-NEXT:    shll $16, %ecx
+; AVX1-NEXT:    orl %eax, %ecx
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm1, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %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:    vmovdqa {{.*#+}} ymm2 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqb %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpmovmskb %ymm0, %eax
+; AVX2-NEXT:    vpcmpeqb %ymm2, %ymm1, %ymm0
+; AVX2-NEXT:    vpmovmskb %ymm0, %ecx
+; AVX2-NEXT:    shlq $32, %rcx
+; AVX2-NEXT:    orq %rax, %rcx
+; AVX2-NEXT:    cmpq $-1, %rcx
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    movdqa {{.*#+}} xmm4 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pcmpeqb %xmm4, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    pcmpeqb %xmm4, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %ecx
+; SSE2-NEXT:    shll $16, %ecx
+; SSE2-NEXT:    orl %eax, %ecx
+; SSE2-NEXT:    pcmpeqb %xmm4, %xmm2
+; SSE2-NEXT:    pmovmskb %xmm2, %eax
+; SSE2-NEXT:    pcmpeqb %xmm4, %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:    vmovaps {{.*#+}} ymm2 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vxorps %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm3
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm3, %xmm3
+; AVX1-NEXT:    vpmovmskb %xmm3, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
+; AVX1-NEXT:    shll $16, %ecx
+; AVX1-NEXT:    orl %eax, %ecx
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm1, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
+; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %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:    vmovdqa {{.*#+}} ymm2 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqb %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpmovmskb %ymm0, %eax
+; AVX2-NEXT:    vpcmpeqb %ymm2, %ymm1, %ymm0
+; AVX2-NEXT:    vpmovmskb %ymm0, %ecx
+; AVX2-NEXT:    shlq $32, %rcx
+; AVX2-NEXT:    orq %rax, %rcx
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    movdqa {{.*#+}} xmm1 = [4,4,4,4,4,4,4,4]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pcmpeqw %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_v8i16_and4:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [4,4,4,4,4,4,4,4]
+; AVX-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpcmpeqw %xmm1, %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
+;
+; 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:    movdqa {{.*#+}} xmm1 = [4,4,4,4,4,4,4,4]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pcmpeqw %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_v8i16_and4:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [4,4,4,4,4,4,4,4]
+; AVX-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpcmpeqw %xmm1, %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
+;
+; 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:    movdqa {{.*#+}} xmm2 = [4,4,4,4,4,4,4,4]
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm2, %xmm0
+; SSE2-NEXT:    packsswb %xmm1, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allones_v16i16_and4:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqw %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: allones_v16i16_and4:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqw %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpmovmskb %xmm0, %eax
+; AVX2-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    movdqa {{.*#+}} xmm4 = [4,4,4,4,4,4,4,4]
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm4, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm4, %xmm0
+; SSE2-NEXT:    packsswb %xmm1, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    pcmpeqw %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqw %xmm4, %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:    vmovaps {{.*#+}} ymm2 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpxor %xmm4, %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:    vmovdqa {{.*#+}} ymm2 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqw %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqw %ymm2, %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
+;
+; 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:    movdqa {{.*#+}} xmm4 = [4,4,4,4,4,4,4,4]
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm4, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm4, %xmm0
+; SSE2-NEXT:    packsswb %xmm1, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    pcmpeqw %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqw %xmm4, %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:    vmovaps {{.*#+}} ymm2 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpcmpeqw %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpxor %xmm4, %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:    vmovdqa {{.*#+}} ymm2 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqw %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqw %ymm2, %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
+;
+; 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:    movdqa {{.*#+}} xmm2 = [4,4,4,4,4,4,4,4]
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqw %xmm2, %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:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqw %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %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:    vmovdqa {{.*#+}} ymm1 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqw %ymm1, %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
+;
+; 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:    movdqa {{.*#+}} xmm1 = [4,4,4,4]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSE2-NEXT:    movmskps %xmm0, %eax
+; SSE2-NEXT:    cmpb $15, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allones_v4i32_and4:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [4,4,4,4]
+; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpcmpeqd %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vmovmskps %xmm0, %eax
+; AVX1-NEXT:    cmpb $15, %al
+; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: allones_v4i32_and4:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [4,4,4,4]
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpcmpeqd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vmovmskps %xmm0, %eax
+; AVX2-NEXT:    cmpb $15, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    retq
+;
+; SKX-LABEL: allones_v4i32_and4:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [4,4,4,4]
+; SKX-NEXT:    vptestmd %xmm1, %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:    movdqa {{.*#+}} xmm1 = [4,4,4,4]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSE2-NEXT:    movmskps %xmm0, %eax
+; SSE2-NEXT:    testb %al, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allzeros_v4i32_and4:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [4,4,4,4]
+; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpcmpeqd %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vmovmskps %xmm0, %eax
+; AVX1-NEXT:    testb %al, %al
+; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: allzeros_v4i32_and4:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [4,4,4,4]
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpcmpeqd %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vmovmskps %xmm0, %eax
+; AVX2-NEXT:    testb %al, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    retq
+;
+; SKX-LABEL: allzeros_v4i32_and4:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [4,4,4,4]
+; SKX-NEXT:    vptestmd %xmm1, %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:    movdqa {{.*#+}} xmm2 = [4,4,4,4]
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm2, %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:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm0
+; 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_v8i32_and4:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [4,4,4,4,4,4,4,4]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqd %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vmovmskps %ymm0, %eax
+; AVX2-NEXT:    cmpb $-1, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
+;
+; SKX-LABEL: allones_v8i32_and4:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [4,4,4,4,4,4,4,4]
+; SKX-NEXT:    vptestmd %ymm1, %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:    movdqa {{.*#+}} xmm2 = [4,4,4,4]
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm2, %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:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm0
+; 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_v8i32_and4:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [4,4,4,4,4,4,4,4]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqd %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vmovmskps %ymm0, %eax
+; AVX2-NEXT:    testb %al, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
+;
+; SKX-LABEL: allzeros_v8i32_and4:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [4,4,4,4,4,4,4,4]
+; SKX-NEXT:    vptestmd %ymm1, %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:    movdqa {{.*#+}} xmm4 = [4,4,4,4]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm2
+; SSE2-NEXT:    packssdw %xmm3, %xmm2
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm0
+; SSE2-NEXT:    packssdw %xmm1, %xmm0
+; SSE2-NEXT:    packsswb %xmm2, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allones_v16i32_and4:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovaps {{.*#+}} ymm2 = [4,4,4,4,4,4,4,4]
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpxor %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: allones_v16i32_and4:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm2 = [4,4,4,4,4,4,4,4]
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqd %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqd %ymm2, %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:    cmpl $65535, %eax # imm = 0xFFFF
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
+;
+; SKX-LABEL: allones_v16i32_and4:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpbroadcastd {{.*#+}} zmm1 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; SKX-NEXT:    vptestmd %zmm1, %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:    movdqa {{.*#+}} xmm4 = [4,4,4,4]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm2
+; SSE2-NEXT:    packssdw %xmm3, %xmm2
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm4, %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:    vmovaps {{.*#+}} ymm2 = [4,4,4,4,4,4,4,4]
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpxor %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %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:    vpbroadcastd {{.*#+}} ymm2 = [4,4,4,4,4,4,4,4]
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqd %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqd %ymm2, %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
+;
+; SKX-LABEL: allzeros_v16i32_and4:
+; SKX:       # %bb.0:
+; SKX-NEXT:    vpbroadcastd {{.*#+}} zmm1 = [4,4,4,4,4,4,4,4,4,4,4,4,4,4,4,4]
+; SKX-NEXT:    vptestmd %zmm1, %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:    movdqa {{.*#+}} xmm1 = [4,4]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; 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:    cmpb $3, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX-LABEL: allones_v2i64_and4:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [4,4]
+; AVX-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpcmpeqq %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vmovmskpd %xmm0, %eax
+; AVX-NEXT:    cmpb $3, %al
+; AVX-NEXT:    sete %al
+; AVX-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:    movdqa {{.*#+}} xmm1 = [4,4]
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; 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:    testb %al, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX-LABEL: allzeros_v2i64_and4:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [4,4]
+; AVX-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpcmpeqq %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vmovmskpd %xmm0, %eax
+; AVX-NEXT:    testb %al, %al
+; AVX-NEXT:    sete %al
+; AVX-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:    movdqa {{.*#+}} xmm2 = [4,4]
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[1,0,3,2]
+; SSE2-NEXT:    pand %xmm1, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,0,3,2]
+; SSE2-NEXT:    pand %xmm0, %xmm1
+; SSE2-NEXT:    packssdw %xmm3, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    cmpb $15, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allones_v4i64_and4:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqq %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %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:    vpbroadcastq {{.*#+}} ymm1 = [4,4,4,4]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqq %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vmovmskpd %ymm0, %eax
+; AVX2-NEXT:    cmpb $15, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    movdqa {{.*#+}} xmm2 = [4,4]
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm2, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[1,0,3,2]
+; SSE2-NEXT:    pand %xmm1, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,0,3,2]
+; SSE2-NEXT:    pand %xmm0, %xmm1
+; SSE2-NEXT:    packssdw %xmm3, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    testb %al, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; AVX1-LABEL: allzeros_v4i64_and4:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vandps {{.*}}(%rip), %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpcmpeqq %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %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:    vpbroadcastq {{.*#+}} ymm1 = [4,4,4,4]
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpcmpeqq %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vmovmskpd %ymm0, %eax
+; AVX2-NEXT:    testb %al, %al
+; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-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:    movdqa {{.*#+}} xmm4 = [4,4]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm3[1,0,3,2]
+; SSE2-NEXT:    pand %xmm3, %xmm5
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,0,3,2]
+; SSE2-NEXT:    pand %xmm2, %xmm3
+; SSE2-NEXT:    packssdw %xmm5, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,0,3,2]
+; SSE2-NEXT:    pand %xmm1, %xmm2
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,0,3,2]
+; SSE2-NEXT:    pand %xmm0, %xmm1
+; SSE2-NEXT:    packssdw %xmm2, %xmm1
+; SSE2-NEXT:    packssdw %xmm3, %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_and4:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovaps {{.*#+}} ymm2 = [4,4,4,4]
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpxor %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
+; 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:    vpbroadcastq {{.*#+}} ymm2 = [4,4,4,4]
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqq %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqq %ymm2, %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
+;
+; 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:    movdqa {{.*#+}} xmm4 = [4,4]
+; SSE2-NEXT:    pand %xmm4, %xmm0
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pand %xmm4, %xmm2
+; SSE2-NEXT:    pand %xmm4, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm3[1,0,3,2]
+; SSE2-NEXT:    pand %xmm3, %xmm5
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,0,3,2]
+; SSE2-NEXT:    pand %xmm2, %xmm3
+; SSE2-NEXT:    packssdw %xmm5, %xmm3
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,0,3,2]
+; SSE2-NEXT:    pand %xmm1, %xmm2
+; SSE2-NEXT:    pcmpeqd %xmm4, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,0,3,2]
+; SSE2-NEXT:    pand %xmm0, %xmm1
+; SSE2-NEXT:    packssdw %xmm2, %xmm1
+; SSE2-NEXT:    packssdw %xmm3, %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_and4:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovaps {{.*#+}} ymm2 = [4,4,4,4]
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm4, %xmm4
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpxor %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpxor %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqq %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpxor %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm0, %xmm0
+; 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:    vpbroadcastq {{.*#+}} ymm2 = [4,4,4,4]
+; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
+; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqq %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpcmpeqq %ymm2, %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
+;
+; 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
+}