[CodeGen] ExpandReductions - add reduce_and/or(<X x i1> V) -> icmp(iX bitcast(<X x i1> V)) canonicalization

[RKSimon]

Summary

This already exists in InstCombine but was missing from the late stage ExpandReductions pass

Fixes Issue #53419

[craig.topper]

LGTM

llvm/lib/CodeGen/ExpandReductions.cpp

case Intrinsic::vector_reduce_fmul: {
           continue;
 
         Rdx = getShuffleReduction(Builder, Vec, getOpcode(ID), RK);
         Rdx = Builder.CreateBinOp((Instruction::BinaryOps)getOpcode(ID),
                                   Acc, Rdx, "bin.rdx");
       }
       break;
     }
+    case Intrinsic::vector_reduce_and:
+    case Intrinsic::vector_reduce_or: {
+      // Canonicalize logical or/and reductions:
+      // Or reduction for i1 is represented as:
+      // %val = bitcast <ReduxWidth x i1> to iReduxWidth
+      // %res = cmp ne iReduxWidth %val, 0
+      // And reduction for i1 is represented as:
+      // %val = bitcast <ReduxWidth x i1> to iReduxWidth
+      // %res = cmp eq iReduxWidth %val, 11111
+      Value *Vec = II->getArgOperand(0);
+      auto *FTy = cast<FixedVectorType>(Vec->getType());
+      unsigned NumElts = FTy->getNumElements();
+      if (!isPowerOf2_32(NumElts))
+        continue;
+
+      if (FTy->getElementType() == Builder.getInt1Ty()) {
+        Rdx = Builder.CreateBitCast(Vec, Builder.getIntNTy(NumElts));
+        if (ID == Intrinsic::vector_reduce_and) {
+          Rdx = Builder.CreateICmpEQ(
+              Rdx, ConstantInt::getAllOnesValue(Rdx->getType()));
+        } else {
+          assert(ID == Intrinsic::vector_reduce_or && "Expected or reduction.");
+          Rdx = Builder.CreateIsNotNull(Rdx);
+        }
+        break;
+      }
+
+      Rdx = getShuffleReduction(Builder, Vec, getOpcode(ID), RK);
+      break;
+    }
     case Intrinsic::vector_reduce_add:
     case Intrinsic::vector_reduce_mul:
-    case Intrinsic::vector_reduce_and:
-    case Intrinsic::vector_reduce_or:
     case Intrinsic::vector_reduce_xor:
     case Intrinsic::vector_reduce_smax:
     case Intrinsic::vector_reduce_smin:
     case Intrinsic::vector_reduce_umax:
     case Intrinsic::vector_reduce_umin: {
       Value *Vec = II->getArgOperand(0);
       if (!isPowerOf2_32(
               cast<FixedVectorType>(Vec->getType())->getNumElements()))

llvm/test/CodeGen/X86/pr53419.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+sse2     | FileCheck %s --check-prefixes=X64
 ; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+sse4.2   | FileCheck %s --check-prefixes=X64
 ; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx      | FileCheck %s --check-prefixes=X64
 ; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx2     | FileCheck %s --check-prefixes=X64
 ; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx512vl,+avx512bw,+avx512dq | FileCheck %s --check-prefixes=X64
 ; RUN: llc < %s -mtriple=i686-unknown   -mattr=+avx2     | FileCheck %s --check-prefixes=X86
 
 declare i1 @llvm.vector.reduce.and.v2i1(<2 x i1>)
 declare i1 @llvm.vector.reduce.and.v4i1(<4 x i1>)
 declare i1 @llvm.vector.reduce.and.v8i1(<8 x i1>)
 
-; FIXME: All four versions are semantically equivalent and should produce same asm as scalar version.
+; All four versions are semantically equivalent and should produce same asm as scalar version.
 
 define i1 @intrinsic_v2i8(ptr align 1 %arg, ptr align 1 %arg1) {
 ; X64-LABEL: intrinsic_v2i8:
 ; X64:       # %bb.0: # %bb
-; X64-NEXT:    movzwl (%rdi), %eax
-; X64-NEXT:    cmpw %ax, (%rsi)
+; X64-NEXT:    movzwl (%rsi), %eax
+; X64-NEXT:    cmpw (%rdi), %ax
 ; X64-NEXT:    sete %al
 ; X64-NEXT:    retq
 ;
 ; X86-LABEL: intrinsic_v2i8:
 ; X86:       # %bb.0: # %bb
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
-; X86-NEXT:    movzwl (%eax), %eax
-; X86-NEXT:    cmpw %ax, (%ecx)
+; X86-NEXT:    movzwl (%ecx), %ecx
+; X86-NEXT:    cmpw (%eax), %cx
 ; X86-NEXT:    sete %al
 ; X86-NEXT:    retl
 bb:
   %lhs = load <2 x i8>, ptr %arg1, align 1
   %rhs = load <2 x i8>, ptr %arg, align 1
   %cmp = icmp eq <2 x i8> %lhs, %rhs
   %all_eq = call i1 @llvm.vector.reduce.and.v2i1(<2 x i1> %cmp)
   ret i1 %all_eq
 }
 
 define i1 @intrinsic_v4i8(ptr align 1 %arg, ptr align 1 %arg1) {
 ; X64-LABEL: intrinsic_v4i8:
 ; X64:       # %bb.0: # %bb
-; X64-NEXT:    movl (%rdi), %eax
-; X64-NEXT:    cmpl %eax, (%rsi)
+; X64-NEXT:    movl (%rsi), %eax
+; X64-NEXT:    cmpl (%rdi), %eax
 ; X64-NEXT:    sete %al
 ; X64-NEXT:    retq
 ;
 ; X86-LABEL: intrinsic_v4i8:
 ; X86:       # %bb.0: # %bb
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
-; X86-NEXT:    movl (%eax), %eax
-; X86-NEXT:    cmpl %eax, (%ecx)
+; X86-NEXT:    movl (%ecx), %ecx
+; X86-NEXT:    cmpl (%eax), %ecx
 ; X86-NEXT:    sete %al
 ; X86-NEXT:    retl
 bb:
   %lhs = load <4 x i8>, ptr %arg1, align 1
   %rhs = load <4 x i8>, ptr %arg, align 1
   %cmp = icmp eq <4 x i8> %lhs, %rhs
   %all_eq = call i1 @llvm.vector.reduce.and.v4i1(<4 x i1> %cmp)
   ret i1 %all_eq
 }
 
 define i1 @intrinsic_v8i8(ptr align 1 %arg, ptr align 1 %arg1) {
 ; X64-LABEL: intrinsic_v8i8:
 ; X64:       # %bb.0: # %bb
-; X64-NEXT:    movq (%rdi), %rax
-; X64-NEXT:    cmpq %rax, (%rsi)
+; X64-NEXT:    movq (%rsi), %rax
+; X64-NEXT:    cmpq (%rdi), %rax
 ; X64-NEXT:    sete %al
 ; X64-NEXT:    retq
 ;
 ; X86-LABEL: intrinsic_v8i8:
 ; X86:       # %bb.0: # %bb
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X86-NEXT:    movl {{[0-9]+}}(%esp), %ecx
-; X86-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
-; X86-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
-; X86-NEXT:    vpcmpeqb %xmm1, %xmm0, %xmm0
-; X86-NEXT:    vpmovmskb %xmm0, %eax
-; X86-NEXT:    cmpb $-1, %al
+; X86-NEXT:    movl (%ecx), %edx
+; X86-NEXT:    movl 4(%ecx), %ecx
+; X86-NEXT:    xorl 4(%eax), %ecx
+; X86-NEXT:    xorl (%eax), %edx
+; X86-NEXT:    orl %ecx, %edx
 ; X86-NEXT:    sete %al
 ; X86-NEXT:    retl
 bb:
   %lhs = load <8 x i8>, ptr %arg1, align 1
   %rhs = load <8 x i8>, ptr %arg, align 1
   %cmp = icmp eq <8 x i8> %lhs, %rhs
   %all_eq = call i1 @llvm.vector.reduce.and.v8i1(<8 x i1> %cmp)
   ret i1 %all_eq

llvm/test/CodeGen/X86/pr61923.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 2
 ; RUN: llc -mtriple=x86_64-unknown-linux-gnu  -mattr=+avx < %s -o - | FileCheck %s
 
 target datalayout = "e-m:e-i64:64-f80:128-n8:16:32:64-S128"
 target triple = "x86_64-unknown-linux-gnu"
 
-; FIXME: Regressed test from https://github.com/llvm/llvm-project/issues/61923
+; Regressed test from https://github.com/llvm/llvm-project/issues/61923
 define void @test_loop(ptr align 1 %src, ptr align 1 %dest, i32 %len) {
 ; CHECK-LABEL: test_loop:
 ; CHECK:       # %bb.0: # %entry
 ; CHECK-NEXT:    cmpl $32, %edx
 ; CHECK-NEXT:    jb .LBB0_4
 ; CHECK-NEXT:  # %bb.1: # %memcmp.loop.preheader
 ; CHECK-NEXT:    movl %edx, %eax
 ; CHECK-NEXT:    andl $-32, %eax
 ; CHECK-NEXT:    xorl %ecx, %ecx
 ; CHECK-NEXT:    .p2align 4, 0x90
 ; CHECK-NEXT:  .LBB0_2: # %memcmp.loop
 ; CHECK-NEXT:    # =>This Inner Loop Header: Depth=1
-; CHECK-NEXT:    vmovdqu (%rsi,%rcx), %xmm0
-; CHECK-NEXT:    vmovdqu 16(%rsi,%rcx), %xmm1
-; CHECK-NEXT:    vpcmpeqb (%rdi,%rcx), %xmm0, %xmm0
-; CHECK-NEXT:    vpmovmskb %xmm0, %edx
-; CHECK-NEXT:    vpcmpeqb 16(%rdi,%rcx), %xmm1, %xmm0
-; CHECK-NEXT:    vpmovmskb %xmm0, %r8d
-; CHECK-NEXT:    shll $16, %r8d
-; CHECK-NEXT:    orl %edx, %r8d
-; CHECK-NEXT:    cmpl $-1, %r8d
+; CHECK-NEXT:    vmovups (%rsi,%rcx), %ymm0
+; CHECK-NEXT:    vxorps (%rdi,%rcx), %ymm0, %ymm0
+; CHECK-NEXT:    vptest %ymm0, %ymm0
 ; CHECK-NEXT:    jne .LBB0_4
 ; CHECK-NEXT:  # %bb.3: # %memcmp.loop.latch
 ; CHECK-NEXT:    # in Loop: Header=BB0_2 Depth=1
 ; CHECK-NEXT:    addq $32, %rcx
 ; CHECK-NEXT:    cmpq %rax, %rcx
 ; CHECK-NEXT:    jb .LBB0_2
 ; CHECK-NEXT:  .LBB0_4: # %done
+; CHECK-NEXT:    vzeroupper
 ; CHECK-NEXT:    retq
 entry:
   %len.wide = zext i32 %len to i64
   %elements.len = lshr i64 %len.wide, 0
   %loop.entry.cond = icmp uge i64 %elements.len, 32
   br i1 %loop.entry.cond, label %memcmp.loop.preheader, label %done
 
 memcmp.loop.preheader:                            ; preds = %entry

llvm/test/CodeGen/X86/vector-reduce-and-bool.ll

 ; 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-prefixes=SSE,SSE2
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.1 | FileCheck %s --check-prefixes=SSE,SSE41
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefixes=AVX,AVX1OR2,AVX1
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX1OR2,AVX2
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefixes=AVX,AVX512,AVX512F
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX512,AVX512BW
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw,+avx512vl | FileCheck %s --check-prefixes=AVX,AVX512,AVX512VL
 
 ;
 ; Truncate
 ;
 
 define i1 @trunc_v2i64_v2i1(<2 x i64>) {
-; SSE-LABEL: trunc_v2i64_v2i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    psllq $63, %xmm0
-; SSE-NEXT:    movmskpd %xmm0, %eax
-; SSE-NEXT:    cmpb $3, %al
-; SSE-NEXT:    sete %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: trunc_v2i64_v2i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    psllq $63, %xmm0
+; SSE2-NEXT:    movmskpd %xmm0, %eax
+; SSE2-NEXT:    cmpb $3, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; SSE41-LABEL: trunc_v2i64_v2i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setb %al
+; SSE41-NEXT:    retq
 ;
 ; AVX1OR2-LABEL: trunc_v2i64_v2i1:
 ; AVX1OR2:       # %bb.0:
-; AVX1OR2-NEXT:    vpsllq $63, %xmm0, %xmm0
-; AVX1OR2-NEXT:    vmovmskpd %xmm0, %eax
-; AVX1OR2-NEXT:    cmpb $3, %al
-; AVX1OR2-NEXT:    sete %al
+; AVX1OR2-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; AVX1OR2-NEXT:    setb %al
 ; AVX1OR2-NEXT:    retq
 ;
 ; AVX512F-LABEL: trunc_v2i64_v2i1:
 ; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpsllq $63, %xmm0, %xmm0
-; AVX512F-NEXT:    vptestnmq %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    testb $3, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
+; AVX512F-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; AVX512F-NEXT:    setb %al
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: trunc_v2i64_v2i1:
 ; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpsllq $63, %xmm0, %xmm0
-; AVX512BW-NEXT:    vptestnmq %zmm0, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb $3, %al
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
+; AVX512BW-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; AVX512BW-NEXT:    setb %al
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512VL-LABEL: trunc_v2i64_v2i1:
 ; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpsllq $63, %xmm0, %xmm0
-; AVX512VL-NEXT:    vptestnmq %xmm0, %xmm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    testb $3, %al
-; AVX512VL-NEXT:    sete %al
+; AVX512VL-NEXT:    vpbroadcastq {{.*#+}} xmm1 = [1,1]
+; AVX512VL-NEXT:    vptest %xmm1, %xmm0
+; AVX512VL-NEXT:    setb %al
 ; AVX512VL-NEXT:    retq
   %a = trunc <2 x i64> %0 to <2 x i1>
   %b = call i1 @llvm.vector.reduce.and.v2i1(<2 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v4i32_v4i1(<4 x i32>) {
-; SSE-LABEL: trunc_v4i32_v4i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    pslld $31, %xmm0
-; SSE-NEXT:    movmskps %xmm0, %eax
-; SSE-NEXT:    cmpb $15, %al
-; SSE-NEXT:    sete %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: trunc_v4i32_v4i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    pslld $31, %xmm0
+; SSE2-NEXT:    movmskps %xmm0, %eax
+; SSE2-NEXT:    xorl $15, %eax
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
 ;
-; AVX1OR2-LABEL: trunc_v4i32_v4i1:
-; AVX1OR2:       # %bb.0:
-; AVX1OR2-NEXT:    vpslld $31, %xmm0, %xmm0
-; AVX1OR2-NEXT:    vmovmskps %xmm0, %eax
-; AVX1OR2-NEXT:    cmpb $15, %al
-; AVX1OR2-NEXT:    sete %al
-; AVX1OR2-NEXT:    retq
+; SSE41-LABEL: trunc_v4i32_v4i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setb %al
+; SSE41-NEXT:    retq
 ;
-; AVX512F-LABEL: trunc_v4i32_v4i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpslld $31, %xmm0, %xmm0
-; AVX512F-NEXT:    vptestnmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    testb $15, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
+; AVX1-LABEL: trunc_v4i32_v4i1:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; AVX1-NEXT:    setb %al
+; AVX1-NEXT:    retq
 ;
-; AVX512BW-LABEL: trunc_v4i32_v4i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpslld $31, %xmm0, %xmm0
-; AVX512BW-NEXT:    vptestnmd %zmm0, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb $15, %al
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
+; AVX2-LABEL: trunc_v4i32_v4i1:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [1,1,1,1]
+; AVX2-NEXT:    vptest %xmm1, %xmm0
+; AVX2-NEXT:    setb %al
+; AVX2-NEXT:    retq
 ;
-; AVX512VL-LABEL: trunc_v4i32_v4i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpslld $31, %xmm0, %xmm0
-; AVX512VL-NEXT:    vptestnmd %xmm0, %xmm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    testb $15, %al
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: trunc_v4i32_v4i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [1,1,1,1]
+; AVX512-NEXT:    vptest %xmm1, %xmm0
+; AVX512-NEXT:    setb %al
+; AVX512-NEXT:    retq
   %a = trunc <4 x i32> %0 to <4 x i1>
   %b = call i1 @llvm.vector.reduce.and.v4i1(<4 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v8i16_v8i1(<8 x i16>) {
-; SSE-LABEL: trunc_v8i16_v8i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    psllw $15, %xmm0
-; SSE-NEXT:    packsswb %xmm0, %xmm0
-; SSE-NEXT:    pmovmskb %xmm0, %eax
-; SSE-NEXT:    cmpb $-1, %al
-; SSE-NEXT:    sete %al
-; SSE-NEXT:    retq
-;
-; AVX1OR2-LABEL: trunc_v8i16_v8i1:
-; AVX1OR2:       # %bb.0:
-; AVX1OR2-NEXT:    vpsllw $15, %xmm0, %xmm0
-; AVX1OR2-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
-; AVX1OR2-NEXT:    vpmovmskb %xmm0, %eax
-; AVX1OR2-NEXT:    cmpb $-1, %al
-; AVX1OR2-NEXT:    sete %al
-; AVX1OR2-NEXT:    retq
-;
-; AVX512F-LABEL: trunc_v8i16_v8i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpmovsxwq %xmm0, %zmm0
-; AVX512F-NEXT:    vpsllq $63, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmq %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    cmpb $-1, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
+; SSE2-LABEL: trunc_v8i16_v8i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    psllw $7, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    notl %eax
+; SSE2-NEXT:    testl $21845, %eax # imm = 0x5555
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
 ;
-; AVX512BW-LABEL: trunc_v8i16_v8i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpsllw $15, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpmovw2m %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    cmpb $-1, %al
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
+; SSE41-LABEL: trunc_v8i16_v8i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setb %al
+; SSE41-NEXT:    retq
 ;
-; AVX512VL-LABEL: trunc_v8i16_v8i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpsllw $15, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpmovw2m %xmm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    cmpb $-1, %al
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    retq
+; AVX-LABEL: trunc_v8i16_v8i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; AVX-NEXT:    setb %al
+; AVX-NEXT:    retq
   %a = trunc <8 x i16> %0 to <8 x i1>
   %b = call i1 @llvm.vector.reduce.and.v8i1(<8 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v16i8_v16i1(<16 x i8>) {
-; SSE-LABEL: trunc_v16i8_v16i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    psllw $7, %xmm0
-; SSE-NEXT:    pmovmskb %xmm0, %eax
-; SSE-NEXT:    cmpw $-1, %ax
-; SSE-NEXT:    sete %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: trunc_v16i8_v16i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    psllw $7, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; SSE41-LABEL: trunc_v16i8_v16i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setb %al
+; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: trunc_v16i8_v16i1:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX-NEXT:    vpmovmskb %xmm0, %eax
-; AVX-NEXT:    cmpw $-1, %ax
-; AVX-NEXT:    sete %al
+; AVX-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; AVX-NEXT:    setb %al
 ; AVX-NEXT:    retq
   %a = trunc <16 x i8> %0 to <16 x i1>
   %b = call i1 @llvm.vector.reduce.and.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v4i64_v4i1(<4 x i64>) {
-; SSE-LABEL: trunc_v4i64_v4i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
-; SSE-NEXT:    pslld $31, %xmm0
-; SSE-NEXT:    movmskps %xmm0, %eax
-; SSE-NEXT:    cmpb $15, %al
-; SSE-NEXT:    sete %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: trunc_v4i64_v4i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
+; SSE2-NEXT:    pslld $31, %xmm0
+; SSE2-NEXT:    movmskps %xmm0, %eax
+; SSE2-NEXT:    cmpb $15, %al
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; SSE41-LABEL: trunc_v4i64_v4i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    pand %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setb %al
+; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_v4i64_v4i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
-; AVX1-NEXT:    vpslld $31, %xmm0, %xmm0
-; AVX1-NEXT:    vmovmskps %xmm0, %eax
-; AVX1-NEXT:    cmpb $15, %al
-; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
+; AVX1-NEXT:    setb %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: trunc_v4i64_v4i1:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpsllq $63, %ymm0, %ymm0
-; AVX2-NEXT:    vmovmskpd %ymm0, %eax
-; AVX2-NEXT:    cmpb $15, %al
-; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vpbroadcastq {{.*#+}} ymm1 = [1,1,1,1]
+; AVX2-NEXT:    vptest %ymm1, %ymm0
+; AVX2-NEXT:    setb %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: trunc_v4i64_v4i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpsllq $63, %ymm0, %ymm0
-; AVX512F-NEXT:    vptestnmq %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    testb $15, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: trunc_v4i64_v4i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpsllq $63, %ymm0, %ymm0
-; AVX512BW-NEXT:    vptestnmq %zmm0, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb $15, %al
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: trunc_v4i64_v4i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpsllq $63, %ymm0, %ymm0
-; AVX512VL-NEXT:    vptestnmq %ymm0, %ymm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    testb $15, %al
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: trunc_v4i64_v4i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpbroadcastq {{.*#+}} ymm1 = [1,1,1,1]
+; AVX512-NEXT:    vptest %ymm1, %ymm0
+; AVX512-NEXT:    setb %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = trunc <4 x i64> %0 to <4 x i1>
   %b = call i1 @llvm.vector.reduce.and.v4i1(<4 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v8i32_v8i1(<8 x i32>) {
 ; SSE2-LABEL: trunc_v8i32_v8i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pslld $16, %xmm1
-; SSE2-NEXT:    psrad $16, %xmm1
-; SSE2-NEXT:    pslld $16, %xmm0
-; SSE2-NEXT:    psrad $16, %xmm0
-; SSE2-NEXT:    packssdw %xmm1, %xmm0
-; SSE2-NEXT:    psllw $15, %xmm0
-; SSE2-NEXT:    packsswb %xmm0, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpb $-1, %al
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pslld $31, %xmm0
+; SSE2-NEXT:    movmskps %xmm0, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: trunc_v8i32_v8i1:
 ; SSE41:       # %bb.0:
-; SSE41-NEXT:    pxor %xmm2, %xmm2
-; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0],xmm2[1],xmm1[2],xmm2[3],xmm1[4],xmm2[5],xmm1[6],xmm2[7]
-; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0],xmm2[1],xmm0[2],xmm2[3],xmm0[4],xmm2[5],xmm0[6],xmm2[7]
-; SSE41-NEXT:    packusdw %xmm1, %xmm0
-; SSE41-NEXT:    psllw $15, %xmm0
-; SSE41-NEXT:    packsswb %xmm0, %xmm0
-; SSE41-NEXT:    pmovmskb %xmm0, %eax
-; SSE41-NEXT:    cmpb $-1, %al
-; SSE41-NEXT:    sete %al
+; SSE41-NEXT:    pand %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setb %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_v8i32_v8i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpslld $31, %xmm0, %xmm0
-; AVX1-NEXT:    vmovmskps %xmm0, %eax
-; AVX1-NEXT:    cmpl $15, %eax
-; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
+; AVX1-NEXT:    setb %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: trunc_v8i32_v8i1:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpslld $31, %ymm0, %ymm0
-; AVX2-NEXT:    vmovmskps %ymm0, %eax
-; AVX2-NEXT:    cmpb $-1, %al
-; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vptest %ymm1, %ymm0
+; AVX2-NEXT:    setb %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: trunc_v8i32_v8i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpslld $31, %ymm0, %ymm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    cmpb $-1, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: trunc_v8i32_v8i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpslld $31, %ymm0, %ymm0
-; AVX512BW-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    cmpb $-1, %al
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: trunc_v8i32_v8i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpslld $31, %ymm0, %ymm0
-; AVX512VL-NEXT:    vptestmd %ymm0, %ymm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    cmpb $-1, %al
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: trunc_v8i32_v8i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
+; AVX512-NEXT:    vptest %ymm1, %ymm0
+; AVX512-NEXT:    setb %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = trunc <8 x i32> %0 to <8 x i1>
   %b = call i1 @llvm.vector.reduce.and.v8i1(<8 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v16i16_v16i1(<16 x i16>) {
-; SSE-LABEL: trunc_v16i16_v16i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
-; SSE-NEXT:    pand %xmm2, %xmm1
-; SSE-NEXT:    pand %xmm2, %xmm0
-; SSE-NEXT:    packuswb %xmm1, %xmm0
-; SSE-NEXT:    psllw $7, %xmm0
-; SSE-NEXT:    pmovmskb %xmm0, %eax
-; SSE-NEXT:    cmpw $-1, %ax
-; SSE-NEXT:    sete %al
-; SSE-NEXT:    retq
-;
-; AVX1-LABEL: trunc_v16i16_v16i1:
-; AVX1:       # %bb.0:
-; AVX1-NEXT:    vandps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovmskb %xmm0, %eax
-; AVX1-NEXT:    cmpw $-1, %ax
-; AVX1-NEXT:    sete %al
-; AVX1-NEXT:    vzeroupper
-; AVX1-NEXT:    retq
-;
-; AVX2-LABEL: trunc_v16i16_v16i1:
-; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX2-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX2-NEXT:    vpmovmskb %xmm0, %eax
-; AVX2-NEXT:    cmpw $-1, %ax
-; AVX2-NEXT:    sete %al
-; AVX2-NEXT:    vzeroupper
-; AVX2-NEXT:    retq
-;
-; AVX512F-LABEL: trunc_v16i16_v16i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpmovsxwd %ymm0, %zmm0
-; AVX512F-NEXT:    vpslld $31, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kortestw %k0, %k0
-; AVX512F-NEXT:    setb %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
+; SSE2-LABEL: trunc_v16i16_v16i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    psllw $7, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    notl %eax
+; SSE2-NEXT:    testl $21845, %eax # imm = 0x5555
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
 ;
-; AVX512BW-LABEL: trunc_v16i16_v16i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpsllw $15, %ymm0, %ymm0
-; AVX512BW-NEXT:    vpmovw2m %zmm0, %k0
-; AVX512BW-NEXT:    kortestw %k0, %k0
-; AVX512BW-NEXT:    setb %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
+; SSE41-LABEL: trunc_v16i16_v16i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    pand %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setb %al
+; SSE41-NEXT:    retq
 ;
-; AVX512VL-LABEL: trunc_v16i16_v16i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpsllw $15, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpmovw2m %ymm0, %k0
-; AVX512VL-NEXT:    kortestw %k0, %k0
-; AVX512VL-NEXT:    setb %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX-LABEL: trunc_v16i16_v16i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
+; AVX-NEXT:    setb %al
+; AVX-NEXT:    vzeroupper
+; AVX-NEXT:    retq
   %a = trunc <16 x i16> %0 to <16 x i1>
   %b = call i1 @llvm.vector.reduce.and.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v32i8_v32i1(<32 x i8>) {
-; SSE-LABEL: trunc_v32i8_v32i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    pand %xmm1, %xmm0
-; SSE-NEXT:    psllw $7, %xmm0
-; SSE-NEXT:    pmovmskb %xmm0, %eax
-; SSE-NEXT:    cmpw $-1, %ax
-; SSE-NEXT:    sete %al
-; SSE-NEXT:    retq
-;
-; AVX1-LABEL: trunc_v32i8_v32i1:
-; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovmskb %xmm0, %eax
-; AVX1-NEXT:    cmpw $-1, %ax
-; AVX1-NEXT:    sete %al
-; AVX1-NEXT:    vzeroupper
-; AVX1-NEXT:    retq
-;
-; AVX2-LABEL: trunc_v32i8_v32i1:
-; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpsllw $7, %ymm0, %ymm0
-; AVX2-NEXT:    vpmovmskb %ymm0, %eax
-; AVX2-NEXT:    cmpl $-1, %eax
-; AVX2-NEXT:    sete %al
-; AVX2-NEXT:    vzeroupper
-; AVX2-NEXT:    retq
-;
-; AVX512F-LABEL: trunc_v32i8_v32i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512F-NEXT:    vpand %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxbd %xmm0, %zmm0
-; AVX512F-NEXT:    vpslld $31, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
+; SSE2-LABEL: trunc_v32i8_v32i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    psllw $7, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
 ;
-; AVX512BW-LABEL: trunc_v32i8_v32i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpsllw $7, %ymm0, %ymm0
-; AVX512BW-NEXT:    vpmovmskb %ymm0, %eax
-; AVX512BW-NEXT:    cmpl $-1, %eax
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
+; SSE41-LABEL: trunc_v32i8_v32i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    pand %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setb %al
+; SSE41-NEXT:    retq
 ;
-; AVX512VL-LABEL: trunc_v32i8_v32i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpsllw $7, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpmovmskb %ymm0, %eax
-; AVX512VL-NEXT:    cmpl $-1, %eax
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX-LABEL: trunc_v32i8_v32i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
+; AVX-NEXT:    setb %al
+; AVX-NEXT:    vzeroupper
+; AVX-NEXT:    retq
   %a = trunc <32 x i8> %0 to <32 x i1>
   %b = call i1 @llvm.vector.reduce.and.v32i1(<32 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v8i64_v8i1(<8 x i64>) {
 ; SSE2-LABEL: trunc_v8i64_v8i1:
 ; SSE2:       # %bb.0:

 ; SSE2-NEXT:    packsswb %xmm2, %xmm2
 ; SSE2-NEXT:    pmovmskb %xmm2, %eax
 ; SSE2-NEXT:    cmpb $-1, %al
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: trunc_v8i64_v8i1:
 ; SSE41:       # %bb.0:
-; SSE41-NEXT:    pxor %xmm4, %xmm4
-; SSE41-NEXT:    pblendw {{.*#+}} xmm3 = xmm3[0],xmm4[1,2,3],xmm3[4],xmm4[5,6,7]
-; SSE41-NEXT:    pblendw {{.*#+}} xmm2 = xmm2[0],xmm4[1,2,3],xmm2[4],xmm4[5,6,7]
-; SSE41-NEXT:    packusdw %xmm3, %xmm2
-; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0],xmm4[1,2,3],xmm1[4],xmm4[5,6,7]
-; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0],xmm4[1,2,3],xmm0[4],xmm4[5,6,7]
-; SSE41-NEXT:    packusdw %xmm1, %xmm0
-; SSE41-NEXT:    packusdw %xmm2, %xmm0
-; SSE41-NEXT:    psllw $15, %xmm0
-; SSE41-NEXT:    packsswb %xmm0, %xmm0
-; SSE41-NEXT:    pmovmskb %xmm0, %eax
-; SSE41-NEXT:    cmpb $-1, %al
-; SSE41-NEXT:    sete %al
+; SSE41-NEXT:    pand %xmm3, %xmm1
+; SSE41-NEXT:    pand %xmm2, %xmm0
+; SSE41-NEXT:    pand %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setb %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_v8i64_v8i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
-; AVX1-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,2],xmm2[0,2]
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
-; AVX1-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[0,2]
-; AVX1-NEXT:    vandps %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpslld $31, %xmm0, %xmm0
-; AVX1-NEXT:    vmovmskps %xmm0, %eax
-; AVX1-NEXT:    cmpl $15, %eax
-; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vandps %ymm1, %ymm0, %ymm0
+; AVX1-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
+; AVX1-NEXT:    setb %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: trunc_v8i64_v8i1:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vperm2f128 {{.*#+}} ymm2 = ymm0[2,3],ymm1[2,3]
-; AVX2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
-; AVX2-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[0,2],ymm2[0,2],ymm0[4,6],ymm2[4,6]
-; AVX2-NEXT:    vpslld $31, %ymm0, %ymm0
-; AVX2-NEXT:    vmovmskps %ymm0, %eax
-; AVX2-NEXT:    cmpb $-1, %al
-; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpbroadcastq {{.*#+}} ymm1 = [1,1,1,1]
+; AVX2-NEXT:    vptest %ymm1, %ymm0
+; AVX2-NEXT:    setb %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: trunc_v8i64_v8i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpsllq $63, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmq %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    cmpb $-1, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: trunc_v8i64_v8i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpsllq $63, %zmm0, %zmm0
-; AVX512BW-NEXT:    vptestmq %zmm0, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    cmpb $-1, %al
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: trunc_v8i64_v8i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpsllq $63, %zmm0, %zmm0
-; AVX512VL-NEXT:    vptestmq %zmm0, %zmm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    cmpb $-1, %al
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: trunc_v8i64_v8i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpbroadcastq {{.*#+}} zmm1 = [1,1,1,1,1,1,1,1]
+; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
+; AVX512-NEXT:    kortestw %k0, %k0
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = trunc <8 x i64> %0 to <8 x i1>
   %b = call i1 @llvm.vector.reduce.and.v8i1(<8 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v16i32_v16i1(<16 x i32>) {
 ; SSE2-LABEL: trunc_v16i32_v16i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; SSE2-NEXT:    pand %xmm4, %xmm3
-; SSE2-NEXT:    pand %xmm4, %xmm2
-; SSE2-NEXT:    packuswb %xmm3, %xmm2
-; SSE2-NEXT:    pand %xmm4, %xmm1
-; SSE2-NEXT:    pand %xmm4, %xmm0
-; SSE2-NEXT:    packuswb %xmm1, %xmm0
-; SSE2-NEXT:    packuswb %xmm2, %xmm0
-; SSE2-NEXT:    psllw $7, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pand %xmm3, %xmm1
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    pslld $31, %xmm0
+; SSE2-NEXT:    movmskps %xmm0, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: trunc_v16i32_v16i1:
 ; SSE41:       # %bb.0:
-; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; SSE41-NEXT:    pand %xmm4, %xmm3
-; SSE41-NEXT:    pand %xmm4, %xmm2
-; SSE41-NEXT:    packusdw %xmm3, %xmm2
-; SSE41-NEXT:    pand %xmm4, %xmm1
-; SSE41-NEXT:    pand %xmm4, %xmm0
-; SSE41-NEXT:    packusdw %xmm1, %xmm0
-; SSE41-NEXT:    packuswb %xmm2, %xmm0
-; SSE41-NEXT:    psllw $7, %xmm0
-; SSE41-NEXT:    pmovmskb %xmm0, %eax
-; SSE41-NEXT:    cmpw $-1, %ax
-; SSE41-NEXT:    sete %al
+; SSE41-NEXT:    pand %xmm3, %xmm1
+; SSE41-NEXT:    pand %xmm2, %xmm0
+; SSE41-NEXT:    pand %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setb %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_v16i32_v16i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vmovaps {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm3
-; AVX1-NEXT:    vpackusdw %xmm3, %xmm1, %xmm1
-; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
-; AVX1-NEXT:    vpackusdw %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovmskb %xmm0, %eax
-; AVX1-NEXT:    cmpw $-1, %ax
-; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vandps %ymm1, %ymm0, %ymm0
+; AVX1-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
+; AVX1-NEXT:    setb %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: trunc_v16i32_v16i1:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255]
-; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
-; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
-; AVX2-NEXT:    vpackusdw %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX2-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; AVX2-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX2-NEXT:    vpmovmskb %xmm0, %eax
-; AVX2-NEXT:    cmpw $-1, %ax
-; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vptest %ymm1, %ymm0
+; AVX2-NEXT:    setb %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: trunc_v16i32_v16i1:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpslld $31, %zmm0, %zmm0
-; AVX512-NEXT:    vptestmd %zmm0, %zmm0, %k0
+; AVX512-NEXT:    vptestnmd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to16}, %zmm0, %k0
 ; AVX512-NEXT:    kortestw %k0, %k0
-; AVX512-NEXT:    setb %al
+; AVX512-NEXT:    sete %al
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %a = trunc <16 x i32> %0 to <16 x i1>
   %b = call i1 @llvm.vector.reduce.and.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v32i16_v32i1(<32 x i16>) {
-; SSE-LABEL: trunc_v32i16_v32i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
-; SSE-NEXT:    pand %xmm4, %xmm3
-; SSE-NEXT:    pand %xmm4, %xmm2
-; SSE-NEXT:    packuswb %xmm3, %xmm2
-; SSE-NEXT:    pand %xmm4, %xmm1
-; SSE-NEXT:    pand %xmm4, %xmm0
-; SSE-NEXT:    packuswb %xmm1, %xmm0
-; SSE-NEXT:    pand %xmm2, %xmm0
-; SSE-NEXT:    psllw $7, %xmm0
-; SSE-NEXT:    pmovmskb %xmm0, %eax
-; SSE-NEXT:    cmpw $-1, %ax
-; SSE-NEXT:    sete %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: trunc_v32i16_v32i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    pand %xmm3, %xmm1
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    psllw $7, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    notl %eax
+; SSE2-NEXT:    testl $21845, %eax # imm = 0x5555
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; SSE41-LABEL: trunc_v32i16_v32i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    pand %xmm3, %xmm1
+; SSE41-NEXT:    pand %xmm2, %xmm0
+; SSE41-NEXT:    pand %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setb %al
+; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_v32i16_v32i1:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vandps %ymm1, %ymm0, %ymm0
-; AVX1-NEXT:    vandps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovmskb %xmm0, %eax
-; AVX1-NEXT:    cmpw $-1, %ax
-; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
+; AVX1-NEXT:    setb %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: trunc_v32i16_v32i1:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpbroadcastw {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
-; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
-; AVX2-NEXT:    vpackuswb %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
-; AVX2-NEXT:    vpsllw $7, %ymm0, %ymm0
-; AVX2-NEXT:    vpmovmskb %ymm0, %eax
-; AVX2-NEXT:    cmpl $-1, %eax
-; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
+; AVX2-NEXT:    setb %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: trunc_v32i16_v32i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512F-NEXT:    vpand %ymm1, %ymm0, %ymm0
-; AVX512F-NEXT:    vpmovsxwd %ymm0, %zmm0
-; AVX512F-NEXT:    vpslld $31, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: trunc_v32i16_v32i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpsllw $15, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpmovw2m %zmm0, %k0
-; AVX512BW-NEXT:    kortestd %k0, %k0
-; AVX512BW-NEXT:    setb %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: trunc_v32i16_v32i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpsllw $15, %zmm0, %zmm0
-; AVX512VL-NEXT:    vpmovw2m %zmm0, %k0
-; AVX512VL-NEXT:    kortestd %k0, %k0
-; AVX512VL-NEXT:    setb %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: trunc_v32i16_v32i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vmovdqa64 {{.*#+}} zmm1 = [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]
+; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
+; AVX512-NEXT:    kortestw %k0, %k0
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = trunc <32 x i16> %0 to <32 x i1>
   %b = call i1 @llvm.vector.reduce.and.v32i1(<32 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v64i8_v64i1(<64 x i8>) {
-; SSE-LABEL: trunc_v64i8_v64i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    pand %xmm3, %xmm1
-; SSE-NEXT:    pand %xmm2, %xmm0
-; SSE-NEXT:    pand %xmm1, %xmm0
-; SSE-NEXT:    psllw $7, %xmm0
-; SSE-NEXT:    pmovmskb %xmm0, %eax
-; SSE-NEXT:    cmpw $-1, %ax
-; SSE-NEXT:    sete %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: trunc_v64i8_v64i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    pand %xmm3, %xmm1
+; SSE2-NEXT:    pand %xmm2, %xmm0
+; SSE2-NEXT:    pand %xmm1, %xmm0
+; SSE2-NEXT:    psllw $7, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; SSE41-LABEL: trunc_v64i8_v64i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    pand %xmm3, %xmm1
+; SSE41-NEXT:    pand %xmm2, %xmm0
+; SSE41-NEXT:    pand %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setb %al
+; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_v64i8_v64i1:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vandps %ymm1, %ymm0, %ymm0
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vandps %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovmskb %xmm0, %eax
-; AVX1-NEXT:    cmpw $-1, %ax
-; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
+; AVX1-NEXT:    setb %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: trunc_v64i8_v64i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpsllw $7, %ymm0, %ymm0
-; AVX2-NEXT:    vpmovmskb %ymm0, %eax
-; AVX2-NEXT:    cmpl $-1, %eax
-; AVX2-NEXT:    sete %al
+; AVX2-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
+; AVX2-NEXT:    setb %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: trunc_v64i8_v64i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512F-NEXT:    vextracti128 $1, %ymm1, %xmm2
-; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm3
-; AVX512F-NEXT:    vpand %xmm2, %xmm3, %xmm2
-; AVX512F-NEXT:    vpand %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxbd %xmm0, %zmm0
-; AVX512F-NEXT:    vpslld $31, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: trunc_v64i8_v64i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpsllw $7, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpmovb2m %zmm0, %k0
-; AVX512BW-NEXT:    kortestq %k0, %k0
-; AVX512BW-NEXT:    setb %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: trunc_v64i8_v64i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpsllw $7, %zmm0, %zmm0
-; AVX512VL-NEXT:    vpmovb2m %zmm0, %k0
-; AVX512VL-NEXT:    kortestq %k0, %k0
-; AVX512VL-NEXT:    setb %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: trunc_v64i8_v64i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vmovdqa64 {{.*#+}} zmm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX512-NEXT:    vpandq %zmm1, %zmm0, %zmm0
+; AVX512-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
+; AVX512-NEXT:    kortestw %k0, %k0
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = trunc <64 x i8> %0 to <64 x i1>
   %b = call i1 @llvm.vector.reduce.and.v64i1(<64 x i1> %a)
   ret i1 %b
 }
 
 ;
 ; Comparison With Zero
 ;
 
 define i1 @icmp0_v2i64_v2i1(<2 x i64>) {
 ; SSE2-LABEL: icmp0_v2i64_v2i1:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    pxor %xmm1, %xmm1
-; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
-; SSE2-NEXT:    pmovmskb %xmm1, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp0_v2i64_v2i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
-; AVX1OR2-LABEL: icmp0_v2i64_v2i1:
-; AVX1OR2:       # %bb.0:
-; AVX1OR2-NEXT:    vptest %xmm0, %xmm0
-; AVX1OR2-NEXT:    sete %al
-; AVX1OR2-NEXT:    retq
-;
-; AVX512F-LABEL: icmp0_v2i64_v2i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
-; AVX512F-NEXT:    vptestmq %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    testb $3, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp0_v2i64_v2i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
-; AVX512BW-NEXT:    vptestmq %zmm0, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb $3, %al
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp0_v2i64_v2i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vptestnmq %xmm0, %xmm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    cmpb $3, %al
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    retq
+; AVX-LABEL: icmp0_v2i64_v2i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vptest %xmm0, %xmm0
+; AVX-NEXT:    sete %al
+; AVX-NEXT:    retq
   %a = icmp eq <2 x i64> %0, zeroinitializer
   %b = call i1 @llvm.vector.reduce.and.v2i1(<2 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp0_v4i32_v4i1(<4 x i32>) {
 ; SSE2-LABEL: icmp0_v4i32_v4i1:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    pxor %xmm1, %xmm1
-; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
-; SSE2-NEXT:    pmovmskb %xmm1, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp0_v4i32_v4i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
-; AVX1OR2-LABEL: icmp0_v4i32_v4i1:
-; AVX1OR2:       # %bb.0:
-; AVX1OR2-NEXT:    vptest %xmm0, %xmm0
-; AVX1OR2-NEXT:    sete %al
-; AVX1OR2-NEXT:    retq
-;
-; AVX512F-LABEL: icmp0_v4i32_v4i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    testb $15, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp0_v4i32_v4i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
-; AVX512BW-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb $15, %al
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp0_v4i32_v4i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vptestnmd %xmm0, %xmm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    cmpb $15, %al
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    retq
+; AVX-LABEL: icmp0_v4i32_v4i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vptest %xmm0, %xmm0
+; AVX-NEXT:    sete %al
+; AVX-NEXT:    retq
   %a = icmp eq <4 x i32> %0, zeroinitializer
   %b = call i1 @llvm.vector.reduce.and.v4i1(<4 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp0_v8i16_v8i1(<8 x i16>) {
 ; SSE2-LABEL: icmp0_v8i16_v8i1:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    pxor %xmm1, %xmm1
 ; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
 ; SSE2-NEXT:    pmovmskb %xmm1, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp0_v8i16_v8i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
-; AVX1OR2-LABEL: icmp0_v8i16_v8i1:
-; AVX1OR2:       # %bb.0:
-; AVX1OR2-NEXT:    vptest %xmm0, %xmm0
-; AVX1OR2-NEXT:    sete %al
-; AVX1OR2-NEXT:    retq
-;
-; AVX512F-LABEL: icmp0_v8i16_v8i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512F-NEXT:    vpcmpeqw %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxwq %xmm0, %zmm0
-; AVX512F-NEXT:    vptestmq %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    cmpb $-1, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp0_v8i16_v8i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vptest %xmm0, %xmm0
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp0_v8i16_v8i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vptest %xmm0, %xmm0
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    retq
+; AVX-LABEL: icmp0_v8i16_v8i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vptest %xmm0, %xmm0
+; AVX-NEXT:    sete %al
+; AVX-NEXT:    retq
   %a = icmp eq <8 x i16> %0, zeroinitializer
   %b = call i1 @llvm.vector.reduce.and.v8i1(<8 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp0_v16i8_v16i1(<16 x i8>) {
 ; SSE2-LABEL: icmp0_v16i8_v16i1:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    pxor %xmm1, %xmm1
 ; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
 ; SSE2-NEXT:    pmovmskb %xmm1, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp0_v16i8_v16i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: icmp0_v16i8_v16i1:
 ; AVX:       # %bb.0:
 ; AVX-NEXT:    vptest %xmm0, %xmm0
 ; AVX-NEXT:    sete %al
 ; AVX-NEXT:    retq
   %a = icmp eq <16 x i8> %0, zeroinitializer
   %b = call i1 @llvm.vector.reduce.and.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp0_v4i64_v4i1(<4 x i64>) {
 ; SSE2-LABEL: icmp0_v4i64_v4i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pxor %xmm2, %xmm2
 ; SSE2-NEXT:    por %xmm1, %xmm0
-; SSE2-NEXT:    pcmpeqb %xmm2, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp0_v4i64_v4i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    por %xmm1, %xmm0
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
-; AVX1OR2-LABEL: icmp0_v4i64_v4i1:
-; AVX1OR2:       # %bb.0:
-; AVX1OR2-NEXT:    vptest %ymm0, %ymm0
-; AVX1OR2-NEXT:    sete %al
-; AVX1OR2-NEXT:    vzeroupper
-; AVX1OR2-NEXT:    retq
-;
-; AVX512F-LABEL: icmp0_v4i64_v4i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
-; AVX512F-NEXT:    vptestmq %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    testb $15, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp0_v4i64_v4i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
-; AVX512BW-NEXT:    vptestmq %zmm0, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb $15, %al
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp0_v4i64_v4i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vptestnmq %ymm0, %ymm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    cmpb $15, %al
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX-LABEL: icmp0_v4i64_v4i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vptest %ymm0, %ymm0
+; AVX-NEXT:    sete %al
+; AVX-NEXT:    vzeroupper
+; AVX-NEXT:    retq
   %a = icmp eq <4 x i64> %0, zeroinitializer
   %b = call i1 @llvm.vector.reduce.and.v4i1(<4 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp0_v8i32_v8i1(<8 x i32>) {
 ; SSE2-LABEL: icmp0_v8i32_v8i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pxor %xmm2, %xmm2
 ; SSE2-NEXT:    por %xmm1, %xmm0
-; SSE2-NEXT:    pcmpeqb %xmm2, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp0_v8i32_v8i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    por %xmm1, %xmm0
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: icmp0_v8i32_v8i1:
 ; AVX:       # %bb.0:
 ; AVX-NEXT:    vptest %ymm0, %ymm0
 ; AVX-NEXT:    sete %al
 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
   %a = icmp eq <8 x i32> %0, zeroinitializer
   %b = call i1 @llvm.vector.reduce.and.v8i1(<8 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp0_v16i16_v16i1(<16 x i16>) {
 ; SSE2-LABEL: icmp0_v16i16_v16i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pxor %xmm2, %xmm2
 ; SSE2-NEXT:    por %xmm1, %xmm0
-; SSE2-NEXT:    pcmpeqb %xmm2, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %eax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp0_v16i16_v16i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    por %xmm1, %xmm0
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
-; AVX1OR2-LABEL: icmp0_v16i16_v16i1:
-; AVX1OR2:       # %bb.0:
-; AVX1OR2-NEXT:    vptest %ymm0, %ymm0
-; AVX1OR2-NEXT:    sete %al
-; AVX1OR2-NEXT:    vzeroupper
-; AVX1OR2-NEXT:    retq
-;
-; AVX512F-LABEL: icmp0_v16i16_v16i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512F-NEXT:    vpcmpeqw %ymm1, %ymm0, %ymm0
-; AVX512F-NEXT:    vpmovsxwd %ymm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kortestw %k0, %k0
-; AVX512F-NEXT:    setb %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp0_v16i16_v16i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vptest %ymm0, %ymm0
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp0_v16i16_v16i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vptest %ymm0, %ymm0
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX-LABEL: icmp0_v16i16_v16i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vptest %ymm0, %ymm0
+; AVX-NEXT:    sete %al
+; AVX-NEXT:    vzeroupper
+; AVX-NEXT:    retq
   %a = icmp eq <16 x i16> %0, zeroinitializer
   %b = call i1 @llvm.vector.reduce.and.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp0_v32i8_v32i1(<32 x i8>) {
 ; SSE2-LABEL: icmp0_v32i8_v32i1:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    por %xmm1, %xmm0
 ; SSE2-NEXT:    pxor %xmm1, %xmm1
 ; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
 ; SSE2-NEXT:    pmovmskb %xmm1, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp0_v32i8_v32i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    por %xmm1, %xmm0
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
-; AVX1OR2-LABEL: icmp0_v32i8_v32i1:
-; AVX1OR2:       # %bb.0:
-; AVX1OR2-NEXT:    vptest %ymm0, %ymm0
-; AVX1OR2-NEXT:    sete %al
-; AVX1OR2-NEXT:    vzeroupper
-; AVX1OR2-NEXT:    retq
-;
-; AVX512F-LABEL: icmp0_v32i8_v32i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512F-NEXT:    vpcmpeqb %ymm1, %ymm0, %ymm0
-; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512F-NEXT:    vpand %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxbd %xmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp0_v32i8_v32i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vptest %ymm0, %ymm0
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp0_v32i8_v32i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vptest %ymm0, %ymm0
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX-LABEL: icmp0_v32i8_v32i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vptest %ymm0, %ymm0
+; AVX-NEXT:    sete %al
+; AVX-NEXT:    vzeroupper
+; AVX-NEXT:    retq
   %a = icmp eq <32 x i8> %0, zeroinitializer
   %b = call i1 @llvm.vector.reduce.and.v32i1(<32 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp0_v8i64_v8i1(<8 x i64>) {
 ; SSE2-LABEL: icmp0_v8i64_v8i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pxor %xmm4, %xmm4
 ; SSE2-NEXT:    por %xmm3, %xmm1
 ; SSE2-NEXT:    por %xmm2, %xmm0
 ; SSE2-NEXT:    por %xmm1, %xmm0
-; SSE2-NEXT:    pcmpeqb %xmm4, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp0_v8i64_v8i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    por %xmm3, %xmm1
 ; SSE41-NEXT:    por %xmm2, %xmm0
 ; SSE41-NEXT:    por %xmm1, %xmm0
; AVX512-NEXT:    retq
   %a = icmp eq <8 x i64> %0, zeroinitializer
   %b = call i1 @llvm.vector.reduce.and.v8i1(<8 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp0_v16i32_v16i1(<16 x i32>) {
 ; SSE2-LABEL: icmp0_v16i32_v16i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pxor %xmm4, %xmm4
 ; SSE2-NEXT:    por %xmm3, %xmm1
 ; SSE2-NEXT:    por %xmm2, %xmm0
 ; SSE2-NEXT:    por %xmm1, %xmm0
-; SSE2-NEXT:    pcmpeqb %xmm4, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp0_v16i32_v16i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    por %xmm3, %xmm1
 ; SSE41-NEXT:    por %xmm2, %xmm0
 ; SSE41-NEXT:    por %xmm1, %xmm0
; AVX512-NEXT:    retq
   %a = icmp eq <16 x i32> %0, zeroinitializer
   %b = call i1 @llvm.vector.reduce.and.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp0_v32i16_v32i1(<32 x i16>) {
 ; SSE2-LABEL: icmp0_v32i16_v32i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pxor %xmm4, %xmm4
 ; SSE2-NEXT:    por %xmm3, %xmm1
 ; SSE2-NEXT:    por %xmm2, %xmm0
 ; SSE2-NEXT:    por %xmm1, %xmm0
-; SSE2-NEXT:    pcmpeqb %xmm4, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %eax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp0_v32i16_v32i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    por %xmm3, %xmm1
 ; SSE41-NEXT:    por %xmm2, %xmm0
 ; SSE41-NEXT:    por %xmm1, %xmm0

 ; AVX2-LABEL: icmp0_v32i16_v32i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpor %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vptest %ymm0, %ymm0
 ; AVX2-NEXT:    sete %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: icmp0_v32i16_v32i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512F-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX512F-NEXT:    vpcmpeqw %ymm2, %ymm1, %ymm1
-; AVX512F-NEXT:    vpcmpeqw %ymm2, %ymm0, %ymm0
-; AVX512F-NEXT:    vpand %ymm1, %ymm0, %ymm0
-; AVX512F-NEXT:    vpmovsxwd %ymm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp0_v32i16_v32i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512BW-NEXT:    kortestw %k0, %k0
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp0_v32i16_v32i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512VL-NEXT:    kortestw %k0, %k0
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: icmp0_v32i16_v32i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vptestmd %zmm0, %zmm0, %k0
+; AVX512-NEXT:    kortestw %k0, %k0
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = icmp eq <32 x i16> %0, zeroinitializer
   %b = call i1 @llvm.vector.reduce.and.v32i1(<32 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp0_v64i8_v64i1(<64 x i8>) {
 ; SSE2-LABEL: icmp0_v64i8_v64i1:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    por %xmm3, %xmm1
 ; SSE2-NEXT:    por %xmm2, %xmm0
 ; SSE2-NEXT:    por %xmm1, %xmm0
 ; SSE2-NEXT:    pxor %xmm1, %xmm1
 ; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
 ; SSE2-NEXT:    pmovmskb %xmm1, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp0_v64i8_v64i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    por %xmm3, %xmm1
 ; SSE41-NEXT:    por %xmm2, %xmm0
 ; SSE41-NEXT:    por %xmm1, %xmm0

 ; AVX2-LABEL: icmp0_v64i8_v64i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpor %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vptest %ymm0, %ymm0
 ; AVX2-NEXT:    sete %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: icmp0_v64i8_v64i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512F-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX512F-NEXT:    vpcmpeqb %ymm2, %ymm1, %ymm1
-; AVX512F-NEXT:    vpcmpeqb %ymm2, %ymm0, %ymm0
-; AVX512F-NEXT:    vextracti128 $1, %ymm1, %xmm2
-; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm3
-; AVX512F-NEXT:    vpand %xmm2, %xmm3, %xmm2
-; AVX512F-NEXT:    vpand %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxbd %xmm0, %zmm0
-; AVX512F-NEXT:    vpslld $31, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp0_v64i8_v64i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512BW-NEXT:    kortestw %k0, %k0
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp0_v64i8_v64i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512VL-NEXT:    kortestw %k0, %k0
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: icmp0_v64i8_v64i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vptestmd %zmm0, %zmm0, %k0
+; AVX512-NEXT:    kortestw %k0, %k0
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = icmp eq <64 x i8> %0, zeroinitializer
   %b = call i1 @llvm.vector.reduce.and.v64i1(<64 x i1> %a)
   ret i1 %b
 }
 
 define i8 @icmp0_v8i1(<8 x i8>) {
 ; SSE2-LABEL: icmp0_v8i1:
 ; SSE2:       # %bb.0:

 ;
 ; Comparison With All Ones
 ;
 
 define i1 @icmp1_v2i64_v2i1(<2 x i64>) {
 ; SSE2-LABEL: icmp1_v2i64_v2i1:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    pcmpeqd %xmm1, %xmm1
-; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
-; SSE2-NEXT:    pmovmskb %xmm1, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp1_v2i64_v2i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE41-NEXT:    ptest %xmm1, %xmm0
 ; SSE41-NEXT:    setb %al
 ; SSE41-NEXT:    retq
 ;
-; AVX1OR2-LABEL: icmp1_v2i64_v2i1:
-; AVX1OR2:       # %bb.0:
-; AVX1OR2-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
-; AVX1OR2-NEXT:    vptest %xmm1, %xmm0
-; AVX1OR2-NEXT:    setb %al
-; AVX1OR2-NEXT:    retq
-;
-; AVX512F-LABEL: icmp1_v2i64_v2i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
-; AVX512F-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
-; AVX512F-NEXT:    vpcmpneqq %zmm1, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    testb $3, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp1_v2i64_v2i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
-; AVX512BW-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
-; AVX512BW-NEXT:    vpcmpneqq %zmm1, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb $3, %al
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp1_v2i64_v2i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
-; AVX512VL-NEXT:    vpcmpeqq %xmm1, %xmm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    cmpb $3, %al
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    retq
+; AVX-LABEL: icmp1_v2i64_v2i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vptest %xmm1, %xmm0
+; AVX-NEXT:    setb %al
+; AVX-NEXT:    retq
   %a = icmp eq <2 x i64> %0, <i64 -1, i64 -1>
   %b = call i1 @llvm.vector.reduce.and.v2i1(<2 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp1_v4i32_v4i1(<4 x i32>) {
 ; SSE2-LABEL: icmp1_v4i32_v4i1:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    pcmpeqd %xmm1, %xmm1
-; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
-; SSE2-NEXT:    pmovmskb %xmm1, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp1_v4i32_v4i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE41-NEXT:    ptest %xmm1, %xmm0
 ; SSE41-NEXT:    setb %al
 ; SSE41-NEXT:    retq
 ;
-; AVX1OR2-LABEL: icmp1_v4i32_v4i1:
-; AVX1OR2:       # %bb.0:
-; AVX1OR2-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
-; AVX1OR2-NEXT:    vptest %xmm1, %xmm0
-; AVX1OR2-NEXT:    setb %al
-; AVX1OR2-NEXT:    retq
-;
-; AVX512F-LABEL: icmp1_v4i32_v4i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
-; AVX512F-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
-; AVX512F-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    testb $15, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp1_v4i32_v4i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
-; AVX512BW-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
-; AVX512BW-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb $15, %al
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp1_v4i32_v4i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
-; AVX512VL-NEXT:    vpcmpeqd %xmm1, %xmm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    cmpb $15, %al
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    retq
+; AVX-LABEL: icmp1_v4i32_v4i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vptest %xmm1, %xmm0
+; AVX-NEXT:    setb %al
+; AVX-NEXT:    retq
   %a = icmp eq <4 x i32> %0, <i32 -1, i32 -1, i32 -1, i32 -1>
   %b = call i1 @llvm.vector.reduce.and.v4i1(<4 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp1_v8i16_v8i1(<8 x i16>) {
 ; SSE2-LABEL: icmp1_v8i16_v8i1:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
 ; SSE2-NEXT:    pmovmskb %xmm1, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp1_v8i16_v8i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE41-NEXT:    ptest %xmm1, %xmm0
 ; SSE41-NEXT:    setb %al
 ; SSE41-NEXT:    retq
 ;
-; AVX1OR2-LABEL: icmp1_v8i16_v8i1:
-; AVX1OR2:       # %bb.0:
-; AVX1OR2-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
-; AVX1OR2-NEXT:    vptest %xmm1, %xmm0
-; AVX1OR2-NEXT:    setb %al
-; AVX1OR2-NEXT:    retq
-;
-; AVX512F-LABEL: icmp1_v8i16_v8i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
-; AVX512F-NEXT:    vpcmpeqw %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxwq %xmm0, %zmm0
-; AVX512F-NEXT:    vptestmq %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    cmpb $-1, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp1_v8i16_v8i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
-; AVX512BW-NEXT:    vptest %xmm1, %xmm0
-; AVX512BW-NEXT:    setb %al
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp1_v8i16_v8i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
-; AVX512VL-NEXT:    vptest %xmm1, %xmm0
-; AVX512VL-NEXT:    setb %al
-; AVX512VL-NEXT:    retq
+; AVX-LABEL: icmp1_v8i16_v8i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
+; AVX-NEXT:    vptest %xmm1, %xmm0
+; AVX-NEXT:    setb %al
+; AVX-NEXT:    retq
   %a = icmp eq <8 x i16> %0, <i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1, i16 -1>
   %b = call i1 @llvm.vector.reduce.and.v8i1(<8 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp1_v16i8_v16i1(<16 x i8>) {
 ; SSE2-LABEL: icmp1_v16i8_v16i1:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
 ; SSE2-NEXT:    pmovmskb %xmm1, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp1_v16i8_v16i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE41-NEXT:    ptest %xmm1, %xmm0
 ; SSE41-NEXT:    setb %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: icmp1_v16i8_v16i1:
 ; AVX:       # %bb.0:
 ; AVX-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
 ; AVX-NEXT:    vptest %xmm1, %xmm0
 ; AVX-NEXT:    setb %al
 ; AVX-NEXT:    retq
   %a = icmp eq <16 x i8> %0, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
   %b = call i1 @llvm.vector.reduce.and.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp1_v4i64_v4i1(<4 x i64>) {
 ; SSE2-LABEL: icmp1_v4i64_v4i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pcmpeqd %xmm2, %xmm2
 ; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    pcmpeqb %xmm2, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp1_v4i64_v4i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pand %xmm1, %xmm0
 ; SSE41-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE41-NEXT:    ptest %xmm1, %xmm0
 ; SSE41-NEXT:    setb %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp1_v4i64_v4i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpxor %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vptest %xmm0, %xmm0
-; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vxorps %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vcmptrueps %ymm1, %ymm1, %ymm1
+; AVX1-NEXT:    vptest %ymm1, %ymm0
+; AVX1-NEXT:    setb %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp1_v4i64_v4i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
 ; AVX2-NEXT:    vptest %ymm1, %ymm0
 ; AVX2-NEXT:    setb %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: icmp1_v4i64_v4i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
-; AVX512F-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
-; AVX512F-NEXT:    vpcmpneqq %zmm1, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    testb $15, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp1_v4i64_v4i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
-; AVX512BW-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
-; AVX512BW-NEXT:    vpcmpneqq %zmm1, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb $15, %al
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp1_v4i64_v4i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpcmpeqq %ymm1, %ymm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    cmpb $15, %al
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: icmp1_v4i64_v4i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
+; AVX512-NEXT:    vptest %ymm1, %ymm0
+; AVX512-NEXT:    setb %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = icmp eq <4 x i64> %0, <i64 -1, i64 -1, i64 -1, i64 -1>
   %b = call i1 @llvm.vector.reduce.and.v4i1(<4 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp1_v8i32_v8i1(<8 x i32>) {
 ; SSE2-LABEL: icmp1_v8i32_v8i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pcmpeqd %xmm2, %xmm2
 ; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    pcmpeqb %xmm2, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp1_v8i32_v8i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pand %xmm1, %xmm0
 ; SSE41-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE41-NEXT:    ptest %xmm1, %xmm0
 ; SSE41-NEXT:    setb %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp1_v8i32_v8i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpxor %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vptest %xmm0, %xmm0
-; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vxorps %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vcmptrueps %ymm1, %ymm1, %ymm1
+; AVX1-NEXT:    vptest %ymm1, %ymm0
+; AVX1-NEXT:    setb %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp1_v8i32_v8i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
 ; AVX2-NEXT:    vptest %ymm1, %ymm0
 ; AVX2-NEXT:    setb %al
; AVX512-NEXT:    retq
   %a = icmp eq <8 x i32> %0, <i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1>
   %b = call i1 @llvm.vector.reduce.and.v8i1(<8 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp1_v16i16_v16i1(<16 x i16>) {
 ; SSE2-LABEL: icmp1_v16i16_v16i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pcmpeqd %xmm2, %xmm2
 ; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    pcmpeqb %xmm2, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %eax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp1_v16i16_v16i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pand %xmm1, %xmm0
 ; SSE41-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE41-NEXT:    ptest %xmm1, %xmm0
 ; SSE41-NEXT:    setb %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp1_v16i16_v16i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpxor %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vptest %xmm0, %xmm0
-; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vxorps %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vcmptrueps %ymm1, %ymm1, %ymm1
+; AVX1-NEXT:    vptest %ymm1, %ymm0
+; AVX1-NEXT:    setb %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp1_v16i16_v16i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
 ; AVX2-NEXT:    vptest %ymm1, %ymm0
 ; AVX2-NEXT:    setb %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: icmp1_v16i16_v16i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
-; AVX512F-NEXT:    vpcmpeqw %ymm1, %ymm0, %ymm0
-; AVX512F-NEXT:    vpmovsxwd %ymm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kortestw %k0, %k0
-; AVX512F-NEXT:    setb %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp1_v16i16_v16i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
-; AVX512BW-NEXT:    vptest %ymm1, %ymm0
-; AVX512BW-NEXT:    setb %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp1_v16i16_v16i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
-; AVX512VL-NEXT:    vptest %ymm1, %ymm0
-; AVX512VL-NEXT:    setb %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: icmp1_v16i16_v16i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
+; AVX512-NEXT:    vptest %ymm1, %ymm0
+; AVX512-NEXT:    setb %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = icmp eq <16 x i16> %0, <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>
   %b = call i1 @llvm.vector.reduce.and.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp1_v32i8_v32i1(<32 x i8>) {
 ; SSE2-LABEL: icmp1_v32i8_v32i1:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    pand %xmm1, %xmm0
 ; SSE2-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
 ; SSE2-NEXT:    pmovmskb %xmm1, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp1_v32i8_v32i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pand %xmm1, %xmm0
 ; SSE41-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE41-NEXT:    ptest %xmm1, %xmm0
 ; SSE41-NEXT:    setb %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp1_v32i8_v32i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpxor %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vptest %xmm0, %xmm0
-; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vxorps %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vcmptrueps %ymm1, %ymm1, %ymm1
+; AVX1-NEXT:    vptest %ymm1, %ymm0
+; AVX1-NEXT:    setb %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp1_v32i8_v32i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
 ; AVX2-NEXT:    vptest %ymm1, %ymm0
 ; AVX2-NEXT:    setb %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: icmp1_v32i8_v32i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
-; AVX512F-NEXT:    vpcmpeqb %ymm1, %ymm0, %ymm0
-; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512F-NEXT:    vpand %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxbd %xmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp1_v32i8_v32i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
-; AVX512BW-NEXT:    vptest %ymm1, %ymm0
-; AVX512BW-NEXT:    setb %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp1_v32i8_v32i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
-; AVX512VL-NEXT:    vptest %ymm1, %ymm0
-; AVX512VL-NEXT:    setb %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: icmp1_v32i8_v32i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
+; AVX512-NEXT:    vptest %ymm1, %ymm0
+; AVX512-NEXT:    setb %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = icmp eq <32 x i8> %0, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
   %b = call i1 @llvm.vector.reduce.and.v32i1(<32 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp1_v8i64_v8i1(<8 x i64>) {
 ; SSE2-LABEL: icmp1_v8i64_v8i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pcmpeqd %xmm4, %xmm4
 ; SSE2-NEXT:    pand %xmm3, %xmm1
 ; SSE2-NEXT:    pand %xmm2, %xmm0
 ; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    pcmpeqb %xmm4, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp1_v8i64_v8i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pand %xmm3, %xmm1
 ; SSE41-NEXT:    pand %xmm2, %xmm0
 ; SSE41-NEXT:    pand %xmm1, %xmm0
 ; SSE41-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE41-NEXT:    ptest %xmm1, %xmm0
 ; SSE41-NEXT:    setb %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp1_v8i64_v8i1:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vandps %ymm1, %ymm0, %ymm0
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpxor %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vptest %xmm0, %xmm0
-; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vxorps %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vcmptrueps %ymm1, %ymm1, %ymm1
+; AVX1-NEXT:    vptest %ymm1, %ymm0
+; AVX1-NEXT:    setb %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp1_v8i64_v8i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
 ; AVX2-NEXT:    vptest %ymm1, %ymm0
; AVX512-NEXT:    retq
   %a = icmp eq <8 x i64> %0, <i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1, i64 -1>
   %b = call i1 @llvm.vector.reduce.and.v8i1(<8 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp1_v16i32_v16i1(<16 x i32>) {
 ; SSE2-LABEL: icmp1_v16i32_v16i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pcmpeqd %xmm4, %xmm4
 ; SSE2-NEXT:    pand %xmm3, %xmm1
 ; SSE2-NEXT:    pand %xmm2, %xmm0
 ; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    pcmpeqb %xmm4, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp1_v16i32_v16i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pand %xmm3, %xmm1
 ; SSE41-NEXT:    pand %xmm2, %xmm0
 ; SSE41-NEXT:    pand %xmm1, %xmm0
 ; SSE41-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE41-NEXT:    ptest %xmm1, %xmm0
 ; SSE41-NEXT:    setb %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp1_v16i32_v16i1:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vandps %ymm1, %ymm0, %ymm0
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpxor %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vptest %xmm0, %xmm0
-; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vxorps %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vcmptrueps %ymm1, %ymm1, %ymm1
+; AVX1-NEXT:    vptest %ymm1, %ymm0
+; AVX1-NEXT:    setb %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp1_v16i32_v16i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
 ; AVX2-NEXT:    vptest %ymm1, %ymm0
; AVX512-NEXT:    retq
   %a = icmp eq <16 x i32> %0, <i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1, i32 -1>
   %b = call i1 @llvm.vector.reduce.and.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp1_v32i16_v32i1(<32 x i16>) {
 ; SSE2-LABEL: icmp1_v32i16_v32i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pcmpeqd %xmm4, %xmm4
 ; SSE2-NEXT:    pand %xmm3, %xmm1
 ; SSE2-NEXT:    pand %xmm2, %xmm0
 ; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    pcmpeqb %xmm4, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %eax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp1_v32i16_v32i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pand %xmm3, %xmm1
 ; SSE41-NEXT:    pand %xmm2, %xmm0
 ; SSE41-NEXT:    pand %xmm1, %xmm0
 ; SSE41-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE41-NEXT:    ptest %xmm1, %xmm0
 ; SSE41-NEXT:    setb %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp1_v32i16_v32i1:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vandps %ymm1, %ymm0, %ymm0
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpxor %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vptest %xmm0, %xmm0
-; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vxorps %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vcmptrueps %ymm1, %ymm1, %ymm1
+; AVX1-NEXT:    vptest %ymm1, %ymm0
+; AVX1-NEXT:    setb %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp1_v32i16_v32i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
 ; AVX2-NEXT:    vptest %ymm1, %ymm0
 ; AVX2-NEXT:    setb %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: icmp1_v32i16_v32i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512F-NEXT:    vpcmpeqd %ymm2, %ymm2, %ymm2
-; AVX512F-NEXT:    vpcmpeqw %ymm2, %ymm1, %ymm1
-; AVX512F-NEXT:    vpcmpeqw %ymm2, %ymm0, %ymm0
-; AVX512F-NEXT:    vpand %ymm1, %ymm0, %ymm0
-; AVX512F-NEXT:    vpmovsxwd %ymm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp1_v32i16_v32i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpternlogd $255, %zmm1, %zmm1, %zmm1
-; AVX512BW-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
-; AVX512BW-NEXT:    kortestw %k0, %k0
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp1_v32i16_v32i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpternlogd $255, %zmm1, %zmm1, %zmm1
-; AVX512VL-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
-; AVX512VL-NEXT:    kortestw %k0, %k0
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: icmp1_v32i16_v32i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpternlogd $255, %zmm1, %zmm1, %zmm1
+; AVX512-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
+; AVX512-NEXT:    kortestw %k0, %k0
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = icmp eq <32 x i16> %0, <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>
   %b = call i1 @llvm.vector.reduce.and.v32i1(<32 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp1_v64i8_v64i1(<64 x i8>) {
 ; SSE2-LABEL: icmp1_v64i8_v64i1:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    pand %xmm3, %xmm1
 ; SSE2-NEXT:    pand %xmm2, %xmm0
 ; SSE2-NEXT:    pand %xmm1, %xmm0
 ; SSE2-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
 ; SSE2-NEXT:    pmovmskb %xmm1, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp1_v64i8_v64i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pand %xmm3, %xmm1
 ; SSE41-NEXT:    pand %xmm2, %xmm0
 ; SSE41-NEXT:    pand %xmm1, %xmm0
 ; SSE41-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE41-NEXT:    ptest %xmm1, %xmm0
 ; SSE41-NEXT:    setb %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp1_v64i8_v64i1:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vandps %ymm1, %ymm0, %ymm0
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpxor %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vptest %xmm0, %xmm0
-; AVX1-NEXT:    sete %al
+; AVX1-NEXT:    vxorps %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vcmptrueps %ymm1, %ymm1, %ymm1
+; AVX1-NEXT:    vptest %ymm1, %ymm0
+; AVX1-NEXT:    setb %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp1_v64i8_v64i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpand %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
 ; AVX2-NEXT:    vptest %ymm1, %ymm0
 ; AVX2-NEXT:    setb %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: icmp1_v64i8_v64i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512F-NEXT:    vpcmpeqd %ymm2, %ymm2, %ymm2
-; AVX512F-NEXT:    vpcmpeqb %ymm2, %ymm1, %ymm1
-; AVX512F-NEXT:    vpcmpeqb %ymm2, %ymm0, %ymm0
-; AVX512F-NEXT:    vextracti128 $1, %ymm1, %xmm2
-; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm3
-; AVX512F-NEXT:    vpand %xmm2, %xmm3, %xmm2
-; AVX512F-NEXT:    vpand %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxbd %xmm0, %zmm0
-; AVX512F-NEXT:    vpslld $31, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp1_v64i8_v64i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpternlogd $255, %zmm1, %zmm1, %zmm1
-; AVX512BW-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
-; AVX512BW-NEXT:    kortestw %k0, %k0
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp1_v64i8_v64i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpternlogd $255, %zmm1, %zmm1, %zmm1
-; AVX512VL-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
-; AVX512VL-NEXT:    kortestw %k0, %k0
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: icmp1_v64i8_v64i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpternlogd $255, %zmm1, %zmm1, %zmm1
+; AVX512-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
+; AVX512-NEXT:    kortestw %k0, %k0
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = icmp eq <64 x i8> %0, <i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>
   %b = call i1 @llvm.vector.reduce.and.v64i1(<64 x i1> %a)
   ret i1 %b
 }
 
 define i8 @icmp1_v8i1(<8 x i8>) {
 ; SSE2-LABEL: icmp1_v8i1:
 ; SSE2:       # %bb.0:

 
 ;
 ; Comparison
 ;
 
 define i1 @icmp_v2i64_v2i1(<2 x i64>, <2 x i64>) {
 ; SSE2-LABEL: icmp_v2i64_v2i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pcmpeqb %xmm1, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSE2-NEXT:    movmskps %xmm0, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp_v2i64_v2i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pxor %xmm1, %xmm0
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
-; AVX1OR2-LABEL: icmp_v2i64_v2i1:
-; AVX1OR2:       # %bb.0:
-; AVX1OR2-NEXT:    vpxor %xmm1, %xmm0, %xmm0
-; AVX1OR2-NEXT:    vptest %xmm0, %xmm0
-; AVX1OR2-NEXT:    sete %al
-; AVX1OR2-NEXT:    retq
-;
-; AVX512F-LABEL: icmp_v2i64_v2i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    # kill: def $xmm1 killed $xmm1 def $zmm1
-; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
-; AVX512F-NEXT:    vpcmpneqq %zmm1, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    testb $3, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp_v2i64_v2i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    # kill: def $xmm1 killed $xmm1 def $zmm1
-; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
-; AVX512BW-NEXT:    vpcmpneqq %zmm1, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb $3, %al
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp_v2i64_v2i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpcmpeqq %xmm1, %xmm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    cmpb $3, %al
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    retq
+; AVX-LABEL: icmp_v2i64_v2i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpxor %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vptest %xmm0, %xmm0
+; AVX-NEXT:    sete %al
+; AVX-NEXT:    retq
   %a = icmp eq <2 x i64> %0, %1
   %b = call i1 @llvm.vector.reduce.and.v2i1(<2 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp_v4i32_v4i1(<4 x i32>, <4 x i32>) {
 ; SSE2-LABEL: icmp_v4i32_v4i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pcmpeqb %xmm1, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
+; SSE2-NEXT:    movmskps %xmm0, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp_v4i32_v4i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pxor %xmm1, %xmm0
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
-; AVX1OR2-LABEL: icmp_v4i32_v4i1:
-; AVX1OR2:       # %bb.0:
-; AVX1OR2-NEXT:    vpxor %xmm1, %xmm0, %xmm0
-; AVX1OR2-NEXT:    vptest %xmm0, %xmm0
-; AVX1OR2-NEXT:    sete %al
-; AVX1OR2-NEXT:    retq
-;
-; AVX512F-LABEL: icmp_v4i32_v4i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    # kill: def $xmm1 killed $xmm1 def $zmm1
-; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
-; AVX512F-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    testb $15, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp_v4i32_v4i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    # kill: def $xmm1 killed $xmm1 def $zmm1
-; AVX512BW-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
-; AVX512BW-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb $15, %al
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp_v4i32_v4i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpcmpeqd %xmm1, %xmm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    cmpb $15, %al
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    retq
+; AVX-LABEL: icmp_v4i32_v4i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpxor %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vptest %xmm0, %xmm0
+; AVX-NEXT:    sete %al
+; AVX-NEXT:    retq
   %a = icmp eq <4 x i32> %0, %1
   %b = call i1 @llvm.vector.reduce.and.v4i1(<4 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp_v8i16_v8i1(<8 x i16>, <8 x i16>) {
 ; SSE2-LABEL: icmp_v8i16_v8i1:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    pcmpeqb %xmm1, %xmm0
 ; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp_v8i16_v8i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pxor %xmm1, %xmm0
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
-; AVX1OR2-LABEL: icmp_v8i16_v8i1:
-; AVX1OR2:       # %bb.0:
-; AVX1OR2-NEXT:    vpxor %xmm1, %xmm0, %xmm0
-; AVX1OR2-NEXT:    vptest %xmm0, %xmm0
-; AVX1OR2-NEXT:    sete %al
-; AVX1OR2-NEXT:    retq
-;
-; AVX512F-LABEL: icmp_v8i16_v8i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpcmpeqw %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxwq %xmm0, %zmm0
-; AVX512F-NEXT:    vptestmq %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    cmpb $-1, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp_v8i16_v8i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpxor %xmm1, %xmm0, %xmm0
-; AVX512BW-NEXT:    vptest %xmm0, %xmm0
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp_v8i16_v8i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpxor %xmm1, %xmm0, %xmm0
-; AVX512VL-NEXT:    vptest %xmm0, %xmm0
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    retq
+; AVX-LABEL: icmp_v8i16_v8i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpxor %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vptest %xmm0, %xmm0
+; AVX-NEXT:    sete %al
+; AVX-NEXT:    retq
   %a = icmp eq <8 x i16> %0, %1
   %b = call i1 @llvm.vector.reduce.and.v8i1(<8 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp_v16i8_v16i1(<16 x i8>, <16 x i8>) {
 ; SSE2-LABEL: icmp_v16i8_v16i1:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    pcmpeqb %xmm1, %xmm0
 ; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp_v16i8_v16i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pxor %xmm1, %xmm0
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: icmp_v16i8_v16i1:
 ; AVX:       # %bb.0:
 ; AVX-NEXT:    vpxor %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    vptest %xmm0, %xmm0
 ; AVX-NEXT:    sete %al
 ; AVX-NEXT:    retq
   %a = icmp eq <16 x i8> %0, %1
   %b = call i1 @llvm.vector.reduce.and.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp_v4i64_v4i1(<4 x i64>, <4 x i64>) {
 ; SSE2-LABEL: icmp_v4i64_v4i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pcmpeqb %xmm3, %xmm1
-; SSE2-NEXT:    pcmpeqb %xmm2, %xmm0
-; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pxor %xmm3, %xmm1
+; SSE2-NEXT:    pxor %xmm2, %xmm0
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp_v4i64_v4i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pxor %xmm3, %xmm1
 ; SSE41-NEXT:    pxor %xmm2, %xmm0
 ; SSE41-NEXT:    por %xmm1, %xmm0
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp_v4i64_v4i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm3
-; AVX1-NEXT:    vpxor %xmm2, %xmm3, %xmm2
-; AVX1-NEXT:    vpxor %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vptest %xmm0, %xmm0
+; AVX1-NEXT:    vxorps %ymm1, %ymm0, %ymm0
+; AVX1-NEXT:    vptest %ymm0, %ymm0
 ; AVX1-NEXT:    sete %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp_v4i64_v4i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpxor %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vptest %ymm0, %ymm0
 ; AVX2-NEXT:    sete %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: icmp_v4i64_v4i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    # kill: def $ymm1 killed $ymm1 def $zmm1
-; AVX512F-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
-; AVX512F-NEXT:    vpcmpneqq %zmm1, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    testb $15, %al
-; AVX512F-NEXT:    sete %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp_v4i64_v4i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    # kill: def $ymm1 killed $ymm1 def $zmm1
-; AVX512BW-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
-; AVX512BW-NEXT:    vpcmpneqq %zmm1, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb $15, %al
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp_v4i64_v4i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpcmpeqq %ymm1, %ymm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    cmpb $15, %al
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: icmp_v4i64_v4i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpxor %ymm1, %ymm0, %ymm0
+; AVX512-NEXT:    vptest %ymm0, %ymm0
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = icmp eq <4 x i64> %0, %1
   %b = call i1 @llvm.vector.reduce.and.v4i1(<4 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp_v8i32_v8i1(<8 x i32>, <8 x i32>) {
 ; SSE2-LABEL: icmp_v8i32_v8i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pcmpeqb %xmm3, %xmm1
-; SSE2-NEXT:    pcmpeqb %xmm2, %xmm0
-; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pxor %xmm3, %xmm1
+; SSE2-NEXT:    pxor %xmm2, %xmm0
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp_v8i32_v8i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pxor %xmm3, %xmm1
 ; SSE41-NEXT:    pxor %xmm2, %xmm0
 ; SSE41-NEXT:    por %xmm1, %xmm0
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp_v8i32_v8i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm3
-; AVX1-NEXT:    vpxor %xmm2, %xmm3, %xmm2
-; AVX1-NEXT:    vpxor %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vptest %xmm0, %xmm0
+; AVX1-NEXT:    vxorps %ymm1, %ymm0, %ymm0
+; AVX1-NEXT:    vptest %ymm0, %ymm0
 ; AVX1-NEXT:    sete %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp_v8i32_v8i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpxor %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vptest %ymm0, %ymm0
; AVX512-NEXT:    retq
   %a = icmp eq <8 x i32> %0, %1
   %b = call i1 @llvm.vector.reduce.and.v8i1(<8 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp_v16i16_v16i1(<16 x i16>, <16 x i16>) {
 ; SSE2-LABEL: icmp_v16i16_v16i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pcmpeqb %xmm3, %xmm1
-; SSE2-NEXT:    pcmpeqb %xmm2, %xmm0
-; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pxor %xmm3, %xmm1
+; SSE2-NEXT:    pxor %xmm2, %xmm0
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %eax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp_v16i16_v16i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pxor %xmm3, %xmm1
 ; SSE41-NEXT:    pxor %xmm2, %xmm0
 ; SSE41-NEXT:    por %xmm1, %xmm0
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp_v16i16_v16i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm3
-; AVX1-NEXT:    vpxor %xmm2, %xmm3, %xmm2
-; AVX1-NEXT:    vpxor %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vptest %xmm0, %xmm0
+; AVX1-NEXT:    vxorps %ymm1, %ymm0, %ymm0
+; AVX1-NEXT:    vptest %ymm0, %ymm0
 ; AVX1-NEXT:    sete %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp_v16i16_v16i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpxor %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vptest %ymm0, %ymm0
 ; AVX2-NEXT:    sete %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: icmp_v16i16_v16i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpcmpeqw %ymm1, %ymm0, %ymm0
-; AVX512F-NEXT:    vpmovsxwd %ymm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kortestw %k0, %k0
-; AVX512F-NEXT:    setb %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp_v16i16_v16i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpxor %ymm1, %ymm0, %ymm0
-; AVX512BW-NEXT:    vptest %ymm0, %ymm0
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp_v16i16_v16i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpxor %ymm1, %ymm0, %ymm0
-; AVX512VL-NEXT:    vptest %ymm0, %ymm0
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: icmp_v16i16_v16i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpxor %ymm1, %ymm0, %ymm0
+; AVX512-NEXT:    vptest %ymm0, %ymm0
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = icmp eq <16 x i16> %0, %1
   %b = call i1 @llvm.vector.reduce.and.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp_v32i8_v32i1(<32 x i8>, <32 x i8>) {
 ; SSE2-LABEL: icmp_v32i8_v32i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pcmpeqb %xmm3, %xmm1
-; SSE2-NEXT:    pcmpeqb %xmm2, %xmm0
-; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    cmpw $-1, %ax
+; SSE2-NEXT:    pxor %xmm3, %xmm1
+; SSE2-NEXT:    pxor %xmm2, %xmm0
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %eax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
 ; SSE2-NEXT:    sete %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: icmp_v32i8_v32i1:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pxor %xmm3, %xmm1
 ; SSE41-NEXT:    pxor %xmm2, %xmm0
 ; SSE41-NEXT:    por %xmm1, %xmm0
 ; SSE41-NEXT:    ptest %xmm0, %xmm0
 ; SSE41-NEXT:    sete %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp_v32i8_v32i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm3
-; AVX1-NEXT:    vpxor %xmm2, %xmm3, %xmm2
-; AVX1-NEXT:    vpxor %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vptest %xmm0, %xmm0
+; AVX1-NEXT:    vxorps %ymm1, %ymm0, %ymm0
+; AVX1-NEXT:    vptest %ymm0, %ymm0
 ; AVX1-NEXT:    sete %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp_v32i8_v32i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpxor %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vptest %ymm0, %ymm0
 ; AVX2-NEXT:    sete %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: icmp_v32i8_v32i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpcmpeqb %ymm1, %ymm0, %ymm0
-; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512F-NEXT:    vpand %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxbd %xmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp_v32i8_v32i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpxor %ymm1, %ymm0, %ymm0
-; AVX512BW-NEXT:    vptest %ymm0, %ymm0
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp_v32i8_v32i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpxor %ymm1, %ymm0, %ymm0
-; AVX512VL-NEXT:    vptest %ymm0, %ymm0
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: icmp_v32i8_v32i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpxor %ymm1, %ymm0, %ymm0
+; AVX512-NEXT:    vptest %ymm0, %ymm0
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = icmp eq <32 x i8> %0, %1
   %b = call i1 @llvm.vector.reduce.and.v32i1(<32 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp_v8i64_v8i1(<8 x i64>, <8 x i64>) {
-; SSE-LABEL: icmp_v8i64_v8i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    pcmpeqb %xmm7, %xmm3
-; SSE-NEXT:    pcmpeqb %xmm5, %xmm1
-; SSE-NEXT:    pand %xmm3, %xmm1
-; SSE-NEXT:    pcmpeqb %xmm6, %xmm2
-; SSE-NEXT:    pcmpeqb %xmm4, %xmm0
-; SSE-NEXT:    pand %xmm2, %xmm0
-; SSE-NEXT:    pand %xmm1, %xmm0
-; SSE-NEXT:    pmovmskb %xmm0, %eax
-; SSE-NEXT:    cmpw $-1, %ax
-; SSE-NEXT:    sete %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: icmp_v8i64_v8i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    pxor %xmm7, %xmm3
+; SSE2-NEXT:    pxor %xmm5, %xmm1
+; SSE2-NEXT:    por %xmm3, %xmm1
+; SSE2-NEXT:    pxor %xmm6, %xmm2
+; SSE2-NEXT:    pxor %xmm4, %xmm0
+; SSE2-NEXT:    por %xmm2, %xmm0
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; SSE41-LABEL: icmp_v8i64_v8i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    pxor %xmm7, %xmm3
+; SSE41-NEXT:    pxor %xmm5, %xmm1
+; SSE41-NEXT:    por %xmm3, %xmm1
+; SSE41-NEXT:    pxor %xmm6, %xmm2
+; SSE41-NEXT:    pxor %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm2, %xmm0
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    ptest %xmm0, %xmm0
+; SSE41-NEXT:    sete %al
+; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp_v8i64_v8i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm3, %xmm4
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm5
-; AVX1-NEXT:    vpcmpeqb %xmm4, %xmm5, %xmm4
-; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm5
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm6
-; AVX1-NEXT:    vpcmpeqb %xmm5, %xmm6, %xmm5
-; AVX1-NEXT:    vpand %xmm4, %xmm5, %xmm4
-; AVX1-NEXT:    vpcmpeqb %xmm3, %xmm1, %xmm1
-; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm4, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovmskb %xmm0, %eax
-; AVX1-NEXT:    cmpw $-1, %ax
+; AVX1-NEXT:    vxorps %ymm3, %ymm1, %ymm1
+; AVX1-NEXT:    vxorps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vorps %ymm1, %ymm0, %ymm0
+; AVX1-NEXT:    vptest %ymm0, %ymm0
 ; AVX1-NEXT:    sete %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp_v8i64_v8i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpxor %ymm3, %ymm1, %ymm1
 ; AVX2-NEXT:    vpxor %ymm2, %ymm0, %ymm0

 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %a = icmp eq <8 x i64> %0, %1
   %b = call i1 @llvm.vector.reduce.and.v8i1(<8 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp_v16i32_v16i1(<16 x i32>, <16 x i32>) {
-; SSE-LABEL: icmp_v16i32_v16i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    pcmpeqb %xmm7, %xmm3
-; SSE-NEXT:    pcmpeqb %xmm5, %xmm1
-; SSE-NEXT:    pand %xmm3, %xmm1
-; SSE-NEXT:    pcmpeqb %xmm6, %xmm2
-; SSE-NEXT:    pcmpeqb %xmm4, %xmm0
-; SSE-NEXT:    pand %xmm2, %xmm0
-; SSE-NEXT:    pand %xmm1, %xmm0
-; SSE-NEXT:    pmovmskb %xmm0, %eax
-; SSE-NEXT:    cmpw $-1, %ax
-; SSE-NEXT:    sete %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: icmp_v16i32_v16i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    pxor %xmm7, %xmm3
+; SSE2-NEXT:    pxor %xmm5, %xmm1
+; SSE2-NEXT:    por %xmm3, %xmm1
+; SSE2-NEXT:    pxor %xmm6, %xmm2
+; SSE2-NEXT:    pxor %xmm4, %xmm0
+; SSE2-NEXT:    por %xmm2, %xmm0
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    movmskps %xmm1, %eax
+; SSE2-NEXT:    xorl $15, %eax
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; SSE41-LABEL: icmp_v16i32_v16i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    pxor %xmm7, %xmm3
+; SSE41-NEXT:    pxor %xmm5, %xmm1
+; SSE41-NEXT:    por %xmm3, %xmm1
+; SSE41-NEXT:    pxor %xmm6, %xmm2
+; SSE41-NEXT:    pxor %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm2, %xmm0
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    ptest %xmm0, %xmm0
+; SSE41-NEXT:    sete %al
+; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp_v16i32_v16i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm3, %xmm4
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm5
-; AVX1-NEXT:    vpcmpeqb %xmm4, %xmm5, %xmm4
-; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm5
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm6
-; AVX1-NEXT:    vpcmpeqb %xmm5, %xmm6, %xmm5
-; AVX1-NEXT:    vpand %xmm4, %xmm5, %xmm4
-; AVX1-NEXT:    vpcmpeqb %xmm3, %xmm1, %xmm1
-; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm4, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovmskb %xmm0, %eax
-; AVX1-NEXT:    cmpw $-1, %ax
+; AVX1-NEXT:    vxorps %ymm3, %ymm1, %ymm1
+; AVX1-NEXT:    vxorps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vorps %ymm1, %ymm0, %ymm0
+; AVX1-NEXT:    vptest %ymm0, %ymm0
 ; AVX1-NEXT:    sete %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp_v16i32_v16i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpxor %ymm3, %ymm1, %ymm1
 ; AVX2-NEXT:    vpxor %ymm2, %ymm0, %ymm0

 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %a = icmp eq <16 x i32> %0, %1
   %b = call i1 @llvm.vector.reduce.and.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp_v32i16_v32i1(<32 x i16>, <32 x i16>) {
-; SSE-LABEL: icmp_v32i16_v32i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    pcmpeqb %xmm7, %xmm3
-; SSE-NEXT:    pcmpeqb %xmm5, %xmm1
-; SSE-NEXT:    pand %xmm3, %xmm1
-; SSE-NEXT:    pcmpeqb %xmm6, %xmm2
-; SSE-NEXT:    pcmpeqb %xmm4, %xmm0
-; SSE-NEXT:    pand %xmm2, %xmm0
-; SSE-NEXT:    pand %xmm1, %xmm0
-; SSE-NEXT:    pmovmskb %xmm0, %eax
-; SSE-NEXT:    cmpw $-1, %ax
-; SSE-NEXT:    sete %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: icmp_v32i16_v32i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    pxor %xmm7, %xmm3
+; SSE2-NEXT:    pxor %xmm5, %xmm1
+; SSE2-NEXT:    por %xmm3, %xmm1
+; SSE2-NEXT:    pxor %xmm6, %xmm2
+; SSE2-NEXT:    pxor %xmm4, %xmm0
+; SSE2-NEXT:    por %xmm2, %xmm0
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %eax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; SSE41-LABEL: icmp_v32i16_v32i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    pxor %xmm7, %xmm3
+; SSE41-NEXT:    pxor %xmm5, %xmm1
+; SSE41-NEXT:    por %xmm3, %xmm1
+; SSE41-NEXT:    pxor %xmm6, %xmm2
+; SSE41-NEXT:    pxor %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm2, %xmm0
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    ptest %xmm0, %xmm0
+; SSE41-NEXT:    sete %al
+; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp_v32i16_v32i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm3, %xmm4
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm5
-; AVX1-NEXT:    vpcmpeqb %xmm4, %xmm5, %xmm4
-; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm5
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm6
-; AVX1-NEXT:    vpcmpeqb %xmm5, %xmm6, %xmm5
-; AVX1-NEXT:    vpand %xmm4, %xmm5, %xmm4
-; AVX1-NEXT:    vpcmpeqb %xmm3, %xmm1, %xmm1
-; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm4, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovmskb %xmm0, %eax
-; AVX1-NEXT:    cmpw $-1, %ax
+; AVX1-NEXT:    vxorps %ymm3, %ymm1, %ymm1
+; AVX1-NEXT:    vxorps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vorps %ymm1, %ymm0, %ymm0
+; AVX1-NEXT:    vptest %ymm0, %ymm0
 ; AVX1-NEXT:    sete %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp_v32i16_v32i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpxor %ymm3, %ymm1, %ymm1
 ; AVX2-NEXT:    vpxor %ymm2, %ymm0, %ymm0
 ; AVX2-NEXT:    vpor %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vptest %ymm0, %ymm0
 ; AVX2-NEXT:    sete %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: icmp_v32i16_v32i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vextracti64x4 $1, %zmm1, %ymm2
-; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm3
-; AVX512F-NEXT:    vpcmpeqw %ymm2, %ymm3, %ymm2
-; AVX512F-NEXT:    vpcmpeqw %ymm1, %ymm0, %ymm0
-; AVX512F-NEXT:    vpand %ymm2, %ymm0, %ymm0
-; AVX512F-NEXT:    vpmovsxwd %ymm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp_v32i16_v32i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
-; AVX512BW-NEXT:    kortestw %k0, %k0
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp_v32i16_v32i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
-; AVX512VL-NEXT:    kortestw %k0, %k0
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: icmp_v32i16_v32i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
+; AVX512-NEXT:    kortestw %k0, %k0
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = icmp eq <32 x i16> %0, %1
   %b = call i1 @llvm.vector.reduce.and.v32i1(<32 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp_v64i8_v64i1(<64 x i8>, <64 x i8>) {
-; SSE-LABEL: icmp_v64i8_v64i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    pcmpeqb %xmm6, %xmm2
-; SSE-NEXT:    pcmpeqb %xmm4, %xmm0
-; SSE-NEXT:    pand %xmm2, %xmm0
-; SSE-NEXT:    pcmpeqb %xmm7, %xmm3
-; SSE-NEXT:    pcmpeqb %xmm5, %xmm1
-; SSE-NEXT:    pand %xmm3, %xmm1
-; SSE-NEXT:    pand %xmm0, %xmm1
-; SSE-NEXT:    pmovmskb %xmm1, %eax
-; SSE-NEXT:    cmpw $-1, %ax
-; SSE-NEXT:    sete %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: icmp_v64i8_v64i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    pxor %xmm7, %xmm3
+; SSE2-NEXT:    pxor %xmm5, %xmm1
+; SSE2-NEXT:    por %xmm3, %xmm1
+; SSE2-NEXT:    pxor %xmm6, %xmm2
+; SSE2-NEXT:    pxor %xmm4, %xmm0
+; SSE2-NEXT:    por %xmm2, %xmm0
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pxor %xmm1, %xmm1
+; SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
+; SSE2-NEXT:    pmovmskb %xmm1, %eax
+; SSE2-NEXT:    xorl $65535, %eax # imm = 0xFFFF
+; SSE2-NEXT:    sete %al
+; SSE2-NEXT:    retq
+;
+; SSE41-LABEL: icmp_v64i8_v64i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    pxor %xmm7, %xmm3
+; SSE41-NEXT:    pxor %xmm5, %xmm1
+; SSE41-NEXT:    por %xmm3, %xmm1
+; SSE41-NEXT:    pxor %xmm6, %xmm2
+; SSE41-NEXT:    pxor %xmm4, %xmm0
+; SSE41-NEXT:    por %xmm2, %xmm0
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    ptest %xmm0, %xmm0
+; SSE41-NEXT:    sete %al
+; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp_v64i8_v64i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpcmpeqb %xmm3, %xmm1, %xmm4
-; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm5
-; AVX1-NEXT:    vpand %xmm4, %xmm5, %xmm4
-; AVX1-NEXT:    vextractf128 $1, %ymm3, %xmm3
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
-; AVX1-NEXT:    vpcmpeqb %xmm3, %xmm1, %xmm1
-; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm2
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; AVX1-NEXT:    vpcmpeqb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm0, %xmm4, %xmm0
-; AVX1-NEXT:    vpmovmskb %xmm0, %eax
-; AVX1-NEXT:    cmpw $-1, %ax
+; AVX1-NEXT:    vxorps %ymm3, %ymm1, %ymm1
+; AVX1-NEXT:    vxorps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vorps %ymm1, %ymm0, %ymm0
+; AVX1-NEXT:    vptest %ymm0, %ymm0
 ; AVX1-NEXT:    sete %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: icmp_v64i8_v64i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpxor %ymm3, %ymm1, %ymm1
 ; AVX2-NEXT:    vpxor %ymm2, %ymm0, %ymm0
 ; AVX2-NEXT:    vpor %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vptest %ymm0, %ymm0
 ; AVX2-NEXT:    sete %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: icmp_v64i8_v64i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vextracti64x4 $1, %zmm1, %ymm2
-; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm3
-; AVX512F-NEXT:    vpcmpeqb %ymm2, %ymm3, %ymm2
-; AVX512F-NEXT:    vpcmpeqb %ymm1, %ymm0, %ymm0
-; AVX512F-NEXT:    vextracti128 $1, %ymm2, %xmm1
-; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm3
-; AVX512F-NEXT:    vpand %xmm1, %xmm3, %xmm1
-; AVX512F-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX512F-NEXT:    vpand %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxbd %xmm0, %zmm0
-; AVX512F-NEXT:    vpslld $31, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    kandw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: icmp_v64i8_v64i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
-; AVX512BW-NEXT:    kortestw %k0, %k0
-; AVX512BW-NEXT:    sete %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: icmp_v64i8_v64i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
-; AVX512VL-NEXT:    kortestw %k0, %k0
-; AVX512VL-NEXT:    sete %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: icmp_v64i8_v64i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpcmpneqd %zmm1, %zmm0, %k0
+; AVX512-NEXT:    kortestw %k0, %k0
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = icmp eq <64 x i8> %0, %1
   %b = call i1 @llvm.vector.reduce.and.v64i1(<64 x i1> %a)
   ret i1 %b
 }
 
 declare i1 @llvm.vector.reduce.and.v2i1(<2 x i1>)
 declare i1 @llvm.vector.reduce.and.v4i1(<4 x i1>)
 declare i1 @llvm.vector.reduce.and.v8i1(<8 x i1>)
 declare i1 @llvm.vector.reduce.and.v16i1(<16 x i1>)
 declare i1 @llvm.vector.reduce.and.v32i1(<32 x i1>)
 declare i1 @llvm.vector.reduce.and.v64i1(<64 x i1>)
+;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
+; SSE: {{.*}}

llvm/test/CodeGen/X86/vector-reduce-or-bool.ll

 ; 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-prefixes=SSE,SSE2
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse4.1 | FileCheck %s --check-prefixes=SSE,SSE41
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefixes=AVX,AVX1
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX2
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f | FileCheck %s --check-prefixes=AVX512,AVX512F
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw | FileCheck %s --check-prefixes=AVX512,AVX512BW
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw,+avx512vl | FileCheck %s --check-prefixes=AVX512,AVX512VL
 
 ;
 ; Truncate
 ;
 
 define i1 @trunc_v2i64_v2i1(<2 x i64>) {
-; SSE-LABEL: trunc_v2i64_v2i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    psllq $63, %xmm0
-; SSE-NEXT:    movmskpd %xmm0, %eax
-; SSE-NEXT:    testl %eax, %eax
-; SSE-NEXT:    setne %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: trunc_v2i64_v2i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    psllq $63, %xmm0
+; SSE2-NEXT:    movmskpd %xmm0, %eax
+; SSE2-NEXT:    testl %eax, %eax
+; SSE2-NEXT:    setne %al
+; SSE2-NEXT:    retq
+;
+; SSE41-LABEL: trunc_v2i64_v2i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setne %al
+; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: trunc_v2i64_v2i1:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vpsllq $63, %xmm0, %xmm0
-; AVX-NEXT:    vmovmskpd %xmm0, %eax
-; AVX-NEXT:    testl %eax, %eax
+; AVX-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
 ; AVX-NEXT:    setne %al
 ; AVX-NEXT:    retq
 ;
 ; AVX512F-LABEL: trunc_v2i64_v2i1:
 ; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpsllq $63, %xmm0, %xmm0
-; AVX512F-NEXT:    vptestmq %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    testb $3, %al
+; AVX512F-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
 ; AVX512F-NEXT:    setne %al
-; AVX512F-NEXT:    vzeroupper
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: trunc_v2i64_v2i1:
 ; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpsllq $63, %xmm0, %xmm0
-; AVX512BW-NEXT:    vptestmq %zmm0, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb $3, %al
+; AVX512BW-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
 ; AVX512BW-NEXT:    setne %al
-; AVX512BW-NEXT:    vzeroupper
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512VL-LABEL: trunc_v2i64_v2i1:
 ; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpsllq $63, %xmm0, %xmm0
-; AVX512VL-NEXT:    vptestmq %xmm0, %xmm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    testb %al, %al
+; AVX512VL-NEXT:    vpbroadcastq {{.*#+}} xmm1 = [1,1]
+; AVX512VL-NEXT:    vptest %xmm1, %xmm0
 ; AVX512VL-NEXT:    setne %al
 ; AVX512VL-NEXT:    retq
   %a = trunc <2 x i64> %0 to <2 x i1>
   %b = call i1 @llvm.vector.reduce.or.v2i1(<2 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v4i32_v4i1(<4 x i32>) {
-; SSE-LABEL: trunc_v4i32_v4i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    pslld $31, %xmm0
-; SSE-NEXT:    movmskps %xmm0, %eax
-; SSE-NEXT:    testl %eax, %eax
-; SSE-NEXT:    setne %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: trunc_v4i32_v4i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    pslld $31, %xmm0
+; SSE2-NEXT:    movmskps %xmm0, %eax
+; SSE2-NEXT:    testl %eax, %eax
+; SSE2-NEXT:    setne %al
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: trunc_v4i32_v4i1:
-; AVX:       # %bb.0:
-; AVX-NEXT:    vpslld $31, %xmm0, %xmm0
-; AVX-NEXT:    vmovmskps %xmm0, %eax
-; AVX-NEXT:    testl %eax, %eax
-; AVX-NEXT:    setne %al
-; AVX-NEXT:    retq
+; SSE41-LABEL: trunc_v4i32_v4i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setne %al
+; SSE41-NEXT:    retq
 ;
-; AVX512F-LABEL: trunc_v4i32_v4i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpslld $31, %xmm0, %xmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    testb $15, %al
-; AVX512F-NEXT:    setne %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
+; AVX1-LABEL: trunc_v4i32_v4i1:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; AVX1-NEXT:    setne %al
+; AVX1-NEXT:    retq
 ;
-; AVX512BW-LABEL: trunc_v4i32_v4i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpslld $31, %xmm0, %xmm0
-; AVX512BW-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb $15, %al
-; AVX512BW-NEXT:    setne %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
+; AVX2-LABEL: trunc_v4i32_v4i1:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [1,1,1,1]
+; AVX2-NEXT:    vptest %xmm1, %xmm0
+; AVX2-NEXT:    setne %al
+; AVX2-NEXT:    retq
 ;
-; AVX512VL-LABEL: trunc_v4i32_v4i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpslld $31, %xmm0, %xmm0
-; AVX512VL-NEXT:    vptestmd %xmm0, %xmm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    testb %al, %al
-; AVX512VL-NEXT:    setne %al
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: trunc_v4i32_v4i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [1,1,1,1]
+; AVX512-NEXT:    vptest %xmm1, %xmm0
+; AVX512-NEXT:    setne %al
+; AVX512-NEXT:    retq
   %a = trunc <4 x i32> %0 to <4 x i1>
   %b = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v8i16_v8i1(<8 x i16>) {
-; SSE-LABEL: trunc_v8i16_v8i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    psllw $15, %xmm0
-; SSE-NEXT:    pmovmskb %xmm0, %eax
-; SSE-NEXT:    testl $43690, %eax # imm = 0xAAAA
-; SSE-NEXT:    setne %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: trunc_v8i16_v8i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    psllw $7, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    testl $21845, %eax # imm = 0x5555
+; SSE2-NEXT:    setne %al
+; SSE2-NEXT:    retq
+;
+; SSE41-LABEL: trunc_v8i16_v8i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setne %al
+; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: trunc_v8i16_v8i1:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vpsllw $15, %xmm0, %xmm0
-; AVX-NEXT:    vpmovmskb %xmm0, %eax
-; AVX-NEXT:    testl $43690, %eax # imm = 0xAAAA
+; AVX-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
 ; AVX-NEXT:    setne %al
 ; AVX-NEXT:    retq
 ;
-; AVX512F-LABEL: trunc_v8i16_v8i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpmovsxwq %xmm0, %zmm0
-; AVX512F-NEXT:    vpsllq $63, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kortestw %k0, %k0
-; AVX512F-NEXT:    setne %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: trunc_v8i16_v8i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpsllw $15, %xmm0, %xmm0
-; AVX512BW-NEXT:    vpmovw2m %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb %al, %al
-; AVX512BW-NEXT:    setne %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: trunc_v8i16_v8i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpsllw $15, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpmovw2m %xmm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    testb %al, %al
-; AVX512VL-NEXT:    setne %al
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: trunc_v8i16_v8i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; AVX512-NEXT:    setne %al
+; AVX512-NEXT:    retq
   %a = trunc <8 x i16> %0 to <8 x i1>
   %b = call i1 @llvm.vector.reduce.or.v8i1(<8 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v16i8_v16i1(<16 x i8>) {
-; SSE-LABEL: trunc_v16i8_v16i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    psllw $7, %xmm0
-; SSE-NEXT:    pmovmskb %xmm0, %eax
-; SSE-NEXT:    testl %eax, %eax
-; SSE-NEXT:    setne %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: trunc_v16i8_v16i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    psllw $7, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    testl %eax, %eax
+; SSE2-NEXT:    setne %al
+; SSE2-NEXT:    retq
+;
+; SSE41-LABEL: trunc_v16i8_v16i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setne %al
+; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: trunc_v16i8_v16i1:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX-NEXT:    vpmovmskb %xmm0, %eax
-; AVX-NEXT:    testl %eax, %eax
+; AVX-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
 ; AVX-NEXT:    setne %al
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: trunc_v16i8_v16i1:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX512-NEXT:    vpmovmskb %xmm0, %eax
-; AVX512-NEXT:    testl %eax, %eax
+; AVX512-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
 ; AVX512-NEXT:    setne %al
 ; AVX512-NEXT:    retq
   %a = trunc <16 x i8> %0 to <16 x i1>
   %b = call i1 @llvm.vector.reduce.or.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v4i64_v4i1(<4 x i64>) {
-; SSE-LABEL: trunc_v4i64_v4i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
-; SSE-NEXT:    pslld $31, %xmm0
-; SSE-NEXT:    movmskps %xmm0, %eax
-; SSE-NEXT:    testl %eax, %eax
-; SSE-NEXT:    setne %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: trunc_v4i64_v4i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
+; SSE2-NEXT:    pslld $31, %xmm0
+; SSE2-NEXT:    movmskps %xmm0, %eax
+; SSE2-NEXT:    testl %eax, %eax
+; SSE2-NEXT:    setne %al
+; SSE2-NEXT:    retq
+;
+; SSE41-LABEL: trunc_v4i64_v4i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setne %al
+; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_v4i64_v4i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
-; AVX1-NEXT:    vpslld $31, %xmm0, %xmm0
-; AVX1-NEXT:    vmovmskps %xmm0, %eax
-; AVX1-NEXT:    testl %eax, %eax
+; AVX1-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
 ; AVX1-NEXT:    setne %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: trunc_v4i64_v4i1:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpsllq $63, %ymm0, %ymm0
-; AVX2-NEXT:    vmovmskpd %ymm0, %eax
-; AVX2-NEXT:    testl %eax, %eax
+; AVX2-NEXT:    vpbroadcastq {{.*#+}} ymm1 = [1,1,1,1]
+; AVX2-NEXT:    vptest %ymm1, %ymm0
 ; AVX2-NEXT:    setne %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: trunc_v4i64_v4i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpsllq $63, %ymm0, %ymm0
-; AVX512F-NEXT:    vptestmq %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    testb $15, %al
-; AVX512F-NEXT:    setne %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: trunc_v4i64_v4i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpsllq $63, %ymm0, %ymm0
-; AVX512BW-NEXT:    vptestmq %zmm0, %zmm0, %k0
-; AVX512BW-NEXT:    kmovd %k0, %eax
-; AVX512BW-NEXT:    testb $15, %al
-; AVX512BW-NEXT:    setne %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: trunc_v4i64_v4i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpsllq $63, %ymm0, %ymm0
-; AVX512VL-NEXT:    vptestmq %ymm0, %ymm0, %k0
-; AVX512VL-NEXT:    kmovd %k0, %eax
-; AVX512VL-NEXT:    testb %al, %al
-; AVX512VL-NEXT:    setne %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: trunc_v4i64_v4i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpbroadcastq {{.*#+}} ymm1 = [1,1,1,1]
+; AVX512-NEXT:    vptest %ymm1, %ymm0
+; AVX512-NEXT:    setne %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = trunc <4 x i64> %0 to <4 x i1>
   %b = call i1 @llvm.vector.reduce.or.v4i1(<4 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v8i32_v8i1(<8 x i32>) {
 ; SSE2-LABEL: trunc_v8i32_v8i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    pslld $16, %xmm1
-; SSE2-NEXT:    psrad $16, %xmm1
-; SSE2-NEXT:    pslld $16, %xmm0
-; SSE2-NEXT:    psrad $16, %xmm0
-; SSE2-NEXT:    packssdw %xmm1, %xmm0
-; SSE2-NEXT:    psllw $15, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-NEXT:    testl $43690, %eax # imm = 0xAAAA
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pslld $31, %xmm0
+; SSE2-NEXT:    movmskps %xmm0, %eax
+; SSE2-NEXT:    testl %eax, %eax
 ; SSE2-NEXT:    setne %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: trunc_v8i32_v8i1:
 ; SSE41:       # %bb.0:
-; SSE41-NEXT:    pxor %xmm2, %xmm2
-; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0],xmm2[1],xmm1[2],xmm2[3],xmm1[4],xmm2[5],xmm1[6],xmm2[7]
-; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0],xmm2[1],xmm0[2],xmm2[3],xmm0[4],xmm2[5],xmm0[6],xmm2[7]
-; SSE41-NEXT:    packusdw %xmm1, %xmm0
-; SSE41-NEXT:    psllw $15, %xmm0
-; SSE41-NEXT:    pmovmskb %xmm0, %eax
-; SSE41-NEXT:    testl $43690, %eax # imm = 0xAAAA
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
 ; SSE41-NEXT:    setne %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_v8i32_v8i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpslld $31, %xmm0, %xmm0
-; AVX1-NEXT:    vmovmskps %xmm0, %eax
-; AVX1-NEXT:    testl %eax, %eax
+; AVX1-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
 ; AVX1-NEXT:    setne %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: trunc_v8i32_v8i1:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpslld $31, %ymm0, %ymm0
-; AVX2-NEXT:    vmovmskps %ymm0, %eax
-; AVX2-NEXT:    testl %eax, %eax
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vptest %ymm1, %ymm0
 ; AVX2-NEXT:    setne %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: trunc_v8i32_v8i1:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpslld $31, %ymm0, %ymm0
-; AVX512-NEXT:    vptest %ymm0, %ymm0
+; AVX512-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
+; AVX512-NEXT:    vptest %ymm1, %ymm0
 ; AVX512-NEXT:    setne %al
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %a = trunc <8 x i32> %0 to <8 x i1>
   %b = call i1 @llvm.vector.reduce.or.v8i1(<8 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v16i16_v16i1(<16 x i16>) {
-; SSE-LABEL: trunc_v16i16_v16i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
-; SSE-NEXT:    pand %xmm2, %xmm1
-; SSE-NEXT:    pand %xmm2, %xmm0
-; SSE-NEXT:    packuswb %xmm1, %xmm0
-; SSE-NEXT:    psllw $7, %xmm0
-; SSE-NEXT:    pmovmskb %xmm0, %eax
-; SSE-NEXT:    testl %eax, %eax
-; SSE-NEXT:    setne %al
-; SSE-NEXT:    retq
-;
-; AVX1-LABEL: trunc_v16i16_v16i1:
-; AVX1:       # %bb.0:
-; AVX1-NEXT:    vandps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovmskb %xmm0, %eax
-; AVX1-NEXT:    testl %eax, %eax
-; AVX1-NEXT:    setne %al
-; AVX1-NEXT:    vzeroupper
-; AVX1-NEXT:    retq
-;
-; AVX2-LABEL: trunc_v16i16_v16i1:
-; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX2-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX2-NEXT:    vpmovmskb %xmm0, %eax
-; AVX2-NEXT:    testl %eax, %eax
-; AVX2-NEXT:    setne %al
-; AVX2-NEXT:    vzeroupper
-; AVX2-NEXT:    retq
+; SSE2-LABEL: trunc_v16i16_v16i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    psllw $7, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    testl $21845, %eax # imm = 0x5555
+; SSE2-NEXT:    setne %al
+; SSE2-NEXT:    retq
 ;
-; AVX512F-LABEL: trunc_v16i16_v16i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpmovsxwd %ymm0, %zmm0
-; AVX512F-NEXT:    vpslld $31, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kortestw %k0, %k0
-; AVX512F-NEXT:    setne %al
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
+; SSE41-LABEL: trunc_v16i16_v16i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setne %al
+; SSE41-NEXT:    retq
 ;
-; AVX512BW-LABEL: trunc_v16i16_v16i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpsllw $15, %ymm0, %ymm0
-; AVX512BW-NEXT:    vpmovw2m %zmm0, %k0
-; AVX512BW-NEXT:    kortestw %k0, %k0
-; AVX512BW-NEXT:    setne %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
+; AVX-LABEL: trunc_v16i16_v16i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
+; AVX-NEXT:    setne %al
+; AVX-NEXT:    vzeroupper
+; AVX-NEXT:    retq
 ;
-; AVX512VL-LABEL: trunc_v16i16_v16i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpsllw $15, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpmovw2m %ymm0, %k0
-; AVX512VL-NEXT:    kortestw %k0, %k0
-; AVX512VL-NEXT:    setne %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: trunc_v16i16_v16i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
+; AVX512-NEXT:    setne %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = trunc <16 x i16> %0 to <16 x i1>
   %b = call i1 @llvm.vector.reduce.or.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v32i8_v32i1(<32 x i8>) {
-; SSE-LABEL: trunc_v32i8_v32i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    por %xmm1, %xmm0
-; SSE-NEXT:    psllw $7, %xmm0
-; SSE-NEXT:    pmovmskb %xmm0, %eax
-; SSE-NEXT:    testl %eax, %eax
-; SSE-NEXT:    setne %al
-; SSE-NEXT:    retq
-;
-; AVX1-LABEL: trunc_v32i8_v32i1:
-; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovmskb %xmm0, %eax
-; AVX1-NEXT:    testl %eax, %eax
-; AVX1-NEXT:    setne %al
-; AVX1-NEXT:    vzeroupper
-; AVX1-NEXT:    retq
-;
-; AVX2-LABEL: trunc_v32i8_v32i1:
-; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpsllw $7, %ymm0, %ymm0
-; AVX2-NEXT:    vpmovmskb %ymm0, %eax
-; AVX2-NEXT:    testl %eax, %eax
-; AVX2-NEXT:    setne %al
-; AVX2-NEXT:    vzeroupper
-; AVX2-NEXT:    retq
+; SSE2-LABEL: trunc_v32i8_v32i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    psllw $7, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    testl %eax, %eax
+; SSE2-NEXT:    setne %al
+; SSE2-NEXT:    retq
 ;
-; AVX512F-LABEL: trunc_v32i8_v32i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512F-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxbd %xmm0, %zmm0
-; AVX512F-NEXT:    vpslld $31, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
+; SSE41-LABEL: trunc_v32i8_v32i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setne %al
+; SSE41-NEXT:    retq
 ;
-; AVX512BW-LABEL: trunc_v32i8_v32i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpsllw $7, %ymm0, %ymm0
-; AVX512BW-NEXT:    vpmovmskb %ymm0, %eax
-; AVX512BW-NEXT:    testl %eax, %eax
-; AVX512BW-NEXT:    setne %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
+; AVX-LABEL: trunc_v32i8_v32i1:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
+; AVX-NEXT:    setne %al
+; AVX-NEXT:    vzeroupper
+; AVX-NEXT:    retq
 ;
-; AVX512VL-LABEL: trunc_v32i8_v32i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpsllw $7, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpmovmskb %ymm0, %eax
-; AVX512VL-NEXT:    testl %eax, %eax
-; AVX512VL-NEXT:    setne %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: trunc_v32i8_v32i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
+; AVX512-NEXT:    setne %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = trunc <32 x i8> %0 to <32 x i1>
   %b = call i1 @llvm.vector.reduce.or.v32i1(<32 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v8i64_v8i1(<8 x i64>) {
 ; SSE2-LABEL: trunc_v8i64_v8i1:
 ; SSE2:       # %bb.0:

 ; SSE2-NEXT:    psllw $15, %xmm2
 ; SSE2-NEXT:    pmovmskb %xmm2, %eax
 ; SSE2-NEXT:    testl $43690, %eax # imm = 0xAAAA
 ; SSE2-NEXT:    setne %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: trunc_v8i64_v8i1:
 ; SSE41:       # %bb.0:
-; SSE41-NEXT:    pxor %xmm4, %xmm4
-; SSE41-NEXT:    pblendw {{.*#+}} xmm3 = xmm3[0],xmm4[1,2,3],xmm3[4],xmm4[5,6,7]
-; SSE41-NEXT:    pblendw {{.*#+}} xmm2 = xmm2[0],xmm4[1,2,3],xmm2[4],xmm4[5,6,7]
-; SSE41-NEXT:    packusdw %xmm3, %xmm2
-; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0],xmm4[1,2,3],xmm1[4],xmm4[5,6,7]
-; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0],xmm4[1,2,3],xmm0[4],xmm4[5,6,7]
-; SSE41-NEXT:    packusdw %xmm1, %xmm0
-; SSE41-NEXT:    packusdw %xmm2, %xmm0
-; SSE41-NEXT:    psllw $15, %xmm0
-; SSE41-NEXT:    pmovmskb %xmm0, %eax
-; SSE41-NEXT:    testl $43690, %eax # imm = 0xAAAA
+; SSE41-NEXT:    por %xmm3, %xmm1
+; SSE41-NEXT:    por %xmm2, %xmm0
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
 ; SSE41-NEXT:    setne %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_v8i64_v8i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
-; AVX1-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,2],xmm2[0,2]
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
-; AVX1-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[0,2]
-; AVX1-NEXT:    vorps %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpslld $31, %xmm0, %xmm0
-; AVX1-NEXT:    vmovmskps %xmm0, %eax
-; AVX1-NEXT:    testl %eax, %eax
+; AVX1-NEXT:    vorps %ymm1, %ymm0, %ymm0
+; AVX1-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
 ; AVX1-NEXT:    setne %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: trunc_v8i64_v8i1:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vperm2f128 {{.*#+}} ymm2 = ymm0[2,3],ymm1[2,3]
-; AVX2-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
-; AVX2-NEXT:    vshufps {{.*#+}} ymm0 = ymm0[0,2],ymm2[0,2],ymm0[4,6],ymm2[4,6]
-; AVX2-NEXT:    vpslld $31, %ymm0, %ymm0
-; AVX2-NEXT:    vmovmskps %ymm0, %eax
-; AVX2-NEXT:    testl %eax, %eax
+; AVX2-NEXT:    vpor %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpbroadcastq {{.*#+}} ymm1 = [1,1,1,1]
+; AVX2-NEXT:    vptest %ymm1, %ymm0
 ; AVX2-NEXT:    setne %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: trunc_v8i64_v8i1:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpsllq $63, %zmm0, %zmm0
-; AVX512-NEXT:    vptestmd %zmm0, %zmm0, %k0
+; AVX512-NEXT:    vpbroadcastq {{.*#+}} zmm1 = [1,1,1,1,1,1,1,1]
+; AVX512-NEXT:    vptestmd %zmm1, %zmm0, %k0
 ; AVX512-NEXT:    kortestw %k0, %k0
 ; AVX512-NEXT:    setne %al
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %a = trunc <8 x i64> %0 to <8 x i1>
   %b = call i1 @llvm.vector.reduce.or.v8i1(<8 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v16i32_v16i1(<16 x i32>) {
 ; SSE2-LABEL: trunc_v16i32_v16i1:
 ; SSE2:       # %bb.0:
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; SSE2-NEXT:    pand %xmm4, %xmm3
-; SSE2-NEXT:    pand %xmm4, %xmm2
-; SSE2-NEXT:    packuswb %xmm3, %xmm2
-; SSE2-NEXT:    pand %xmm4, %xmm1
-; SSE2-NEXT:    pand %xmm4, %xmm0
-; SSE2-NEXT:    packuswb %xmm1, %xmm0
-; SSE2-NEXT:    packuswb %xmm2, %xmm0
-; SSE2-NEXT:    psllw $7, %xmm0
-; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    por %xmm3, %xmm1
+; SSE2-NEXT:    por %xmm2, %xmm0
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    pslld $31, %xmm0
+; SSE2-NEXT:    movmskps %xmm0, %eax
 ; SSE2-NEXT:    testl %eax, %eax
 ; SSE2-NEXT:    setne %al
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: trunc_v16i32_v16i1:
 ; SSE41:       # %bb.0:
-; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; SSE41-NEXT:    pand %xmm4, %xmm3
-; SSE41-NEXT:    pand %xmm4, %xmm2
-; SSE41-NEXT:    packusdw %xmm3, %xmm2
-; SSE41-NEXT:    pand %xmm4, %xmm1
-; SSE41-NEXT:    pand %xmm4, %xmm0
-; SSE41-NEXT:    packusdw %xmm1, %xmm0
-; SSE41-NEXT:    packuswb %xmm2, %xmm0
-; SSE41-NEXT:    psllw $7, %xmm0
-; SSE41-NEXT:    pmovmskb %xmm0, %eax
-; SSE41-NEXT:    testl %eax, %eax
+; SSE41-NEXT:    por %xmm3, %xmm1
+; SSE41-NEXT:    por %xmm2, %xmm0
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
 ; SSE41-NEXT:    setne %al
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_v16i32_v16i1:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vmovaps {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm3
-; AVX1-NEXT:    vpackusdw %xmm3, %xmm1, %xmm1
-; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
-; AVX1-NEXT:    vpackusdw %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovmskb %xmm0, %eax
-; AVX1-NEXT:    testl %eax, %eax
+; AVX1-NEXT:    vorps %ymm1, %ymm0, %ymm0
+; AVX1-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
 ; AVX1-NEXT:    setne %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: trunc_v16i32_v16i1:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255]
-; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
-; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
-; AVX2-NEXT:    vpackusdw %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX2-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; AVX2-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX2-NEXT:    vpmovmskb %xmm0, %eax
-; AVX2-NEXT:    testl %eax, %eax
+; AVX2-NEXT:    vpor %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
+; AVX2-NEXT:    vptest %ymm1, %ymm0
 ; AVX2-NEXT:    setne %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: trunc_v16i32_v16i1:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpslld $31, %zmm0, %zmm0
-; AVX512-NEXT:    vptestmd %zmm0, %zmm0, %k0
+; AVX512-NEXT:    vptestmd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to16}, %zmm0, %k0
 ; AVX512-NEXT:    kortestw %k0, %k0
 ; AVX512-NEXT:    setne %al
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
   %a = trunc <16 x i32> %0 to <16 x i1>
   %b = call i1 @llvm.vector.reduce.or.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v32i16_v32i1(<32 x i16>) {
-; SSE-LABEL: trunc_v32i16_v32i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
-; SSE-NEXT:    pand %xmm4, %xmm3
-; SSE-NEXT:    pand %xmm4, %xmm2
-; SSE-NEXT:    packuswb %xmm3, %xmm2
-; SSE-NEXT:    pand %xmm4, %xmm1
-; SSE-NEXT:    pand %xmm4, %xmm0
-; SSE-NEXT:    packuswb %xmm1, %xmm0
-; SSE-NEXT:    por %xmm2, %xmm0
-; SSE-NEXT:    psllw $7, %xmm0
-; SSE-NEXT:    pmovmskb %xmm0, %eax
-; SSE-NEXT:    testl %eax, %eax
-; SSE-NEXT:    setne %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: trunc_v32i16_v32i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    por %xmm3, %xmm1
+; SSE2-NEXT:    por %xmm2, %xmm0
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    psllw $7, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    testl $21845, %eax # imm = 0x5555
+; SSE2-NEXT:    setne %al
+; SSE2-NEXT:    retq
+;
+; SSE41-LABEL: trunc_v32i16_v32i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    por %xmm3, %xmm1
+; SSE41-NEXT:    por %xmm2, %xmm0
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setne %al
+; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_v32i16_v32i1:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vorps %ymm1, %ymm0, %ymm0
-; AVX1-NEXT:    vandps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovmskb %xmm0, %eax
-; AVX1-NEXT:    testl %eax, %eax
+; AVX1-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
 ; AVX1-NEXT:    setne %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: trunc_v32i16_v32i1:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpbroadcastw {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX2-NEXT:    vpand %ymm2, %ymm1, %ymm1
-; AVX2-NEXT:    vpand %ymm2, %ymm0, %ymm0
-; AVX2-NEXT:    vpackuswb %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,1,3]
-; AVX2-NEXT:    vpsllw $7, %ymm0, %ymm0
-; AVX2-NEXT:    vpmovmskb %ymm0, %eax
-; AVX2-NEXT:    testl %eax, %eax
+; AVX2-NEXT:    vpor %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
 ; AVX2-NEXT:    setne %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: trunc_v32i16_v32i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512F-NEXT:    vpor %ymm1, %ymm0, %ymm0
-; AVX512F-NEXT:    vpmovsxwd %ymm0, %zmm0
-; AVX512F-NEXT:    vpslld $31, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: trunc_v32i16_v32i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpsllw $15, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpmovw2m %zmm0, %k0
-; AVX512BW-NEXT:    kortestd %k0, %k0
-; AVX512BW-NEXT:    setne %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: trunc_v32i16_v32i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpsllw $15, %zmm0, %zmm0
-; AVX512VL-NEXT:    vpmovw2m %zmm0, %k0
-; AVX512VL-NEXT:    kortestd %k0, %k0
-; AVX512VL-NEXT:    setne %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: trunc_v32i16_v32i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vptestmd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %k0
+; AVX512-NEXT:    kortestw %k0, %k0
+; AVX512-NEXT:    setne %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = trunc <32 x i16> %0 to <32 x i1>
   %b = call i1 @llvm.vector.reduce.or.v32i1(<32 x i1> %a)
   ret i1 %b
 }
 
 define i1 @trunc_v64i8_v64i1(<64 x i8>) {
-; SSE-LABEL: trunc_v64i8_v64i1:
-; SSE:       # %bb.0:
-; SSE-NEXT:    por %xmm3, %xmm1
-; SSE-NEXT:    por %xmm2, %xmm0
-; SSE-NEXT:    por %xmm1, %xmm0
-; SSE-NEXT:    psllw $7, %xmm0
-; SSE-NEXT:    pmovmskb %xmm0, %eax
-; SSE-NEXT:    testl %eax, %eax
-; SSE-NEXT:    setne %al
-; SSE-NEXT:    retq
+; SSE2-LABEL: trunc_v64i8_v64i1:
+; SSE2:       # %bb.0:
+; SSE2-NEXT:    por %xmm3, %xmm1
+; SSE2-NEXT:    por %xmm2, %xmm0
+; SSE2-NEXT:    por %xmm1, %xmm0
+; SSE2-NEXT:    psllw $7, %xmm0
+; SSE2-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-NEXT:    testl %eax, %eax
+; SSE2-NEXT:    setne %al
+; SSE2-NEXT:    retq
+;
+; SSE41-LABEL: trunc_v64i8_v64i1:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    por %xmm3, %xmm1
+; SSE41-NEXT:    por %xmm2, %xmm0
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    ptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE41-NEXT:    setne %al
+; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_v64i8_v64i1:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vorps %ymm1, %ymm0, %ymm0
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vorps %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
-; AVX1-NEXT:    vpmovmskb %xmm0, %eax
-; AVX1-NEXT:    testl %eax, %eax
+; AVX1-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
 ; AVX1-NEXT:    setne %al
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: trunc_v64i8_v64i1:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpor %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpsllw $7, %ymm0, %ymm0
-; AVX2-NEXT:    vpmovmskb %ymm0, %eax
-; AVX2-NEXT:    testl %eax, %eax
+; AVX2-NEXT:    vptest {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0
 ; AVX2-NEXT:    setne %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
-; AVX512F-LABEL: trunc_v64i8_v64i1:
-; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512F-NEXT:    vextracti128 $1, %ymm1, %xmm2
-; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm3
-; AVX512F-NEXT:    vpor %xmm2, %xmm3, %xmm2
-; AVX512F-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpor %xmm2, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxbd %xmm0, %zmm0
-; AVX512F-NEXT:    vpslld $31, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
-; AVX512F-NEXT:    vzeroupper
-; AVX512F-NEXT:    retq
-;
-; AVX512BW-LABEL: trunc_v64i8_v64i1:
-; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpsllw $7, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpmovb2m %zmm0, %k0
-; AVX512BW-NEXT:    kortestq %k0, %k0
-; AVX512BW-NEXT:    setne %al
-; AVX512BW-NEXT:    vzeroupper
-; AVX512BW-NEXT:    retq
-;
-; AVX512VL-LABEL: trunc_v64i8_v64i1:
-; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpsllw $7, %zmm0, %zmm0
-; AVX512VL-NEXT:    vpmovb2m %zmm0, %k0
-; AVX512VL-NEXT:    kortestq %k0, %k0
-; AVX512VL-NEXT:    setne %al
-; AVX512VL-NEXT:    vzeroupper
-; AVX512VL-NEXT:    retq
+; AVX512-LABEL: trunc_v64i8_v64i1:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vptestmd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %k0
+; AVX512-NEXT:    kortestw %k0, %k0
+; AVX512-NEXT:    setne %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
   %a = trunc <64 x i8> %0 to <64 x i1>
   %b = call i1 @llvm.vector.reduce.or.v64i1(<64 x i1> %a)
   ret i1 %b
 }
 
 ;
 ; Comparison With Zero
 ;

 ; AVX2-NEXT:    setne %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512F-LABEL: icmp0_v32i8_v32i1:
 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX512F-NEXT:    vpcmpeqb %ymm1, %ymm0, %ymm0
-; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512F-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxbd %xmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
+; AVX512F-NEXT:    vpmovmskb %ymm0, %eax
+; AVX512F-NEXT:    testl %eax, %eax
+; AVX512F-NEXT:    setne %al
 ; AVX512F-NEXT:    vzeroupper
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: icmp0_v32i8_v32i1:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
 ; AVX512BW-NEXT:    vptestnmb %zmm0, %zmm0, %k0
 ; AVX512BW-NEXT:    kortestd %k0, %k0
; AVX512-NEXT:    retq
   %b = call i1 @llvm.vector.reduce.or.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp0_v32i16_v32i1(<32 x i16>) {
 ; SSE-LABEL: icmp0_v32i16_v32i1:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    pxor %xmm4, %xmm4
-; SSE-NEXT:    pcmpeqw %xmm4, %xmm1
-; SSE-NEXT:    pcmpeqw %xmm4, %xmm0
-; SSE-NEXT:    packsswb %xmm1, %xmm0
 ; SSE-NEXT:    pcmpeqw %xmm4, %xmm3
 ; SSE-NEXT:    pcmpeqw %xmm4, %xmm2
 ; SSE-NEXT:    packsswb %xmm3, %xmm2
-; SSE-NEXT:    por %xmm0, %xmm2
-; SSE-NEXT:    pmovmskb %xmm2, %eax
+; SSE-NEXT:    pcmpeqw %xmm4, %xmm1
+; SSE-NEXT:    pcmpeqw %xmm4, %xmm0
+; SSE-NEXT:    packsswb %xmm1, %xmm0
+; SSE-NEXT:    por %xmm2, %xmm0
+; SSE-NEXT:    pmovmskb %xmm0, %eax
 ; SSE-NEXT:    testl %eax, %eax
 ; SSE-NEXT:    setne %al
 ; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp0_v32i16_v32i1:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
 ; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3

 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512F-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; AVX512F-NEXT:    vpcmpeqw %ymm2, %ymm1, %ymm1
 ; AVX512F-NEXT:    vpcmpeqw %ymm2, %ymm0, %ymm0
 ; AVX512F-NEXT:    vpor %ymm1, %ymm0, %ymm0
 ; AVX512F-NEXT:    vpmovsxwd %ymm0, %zmm0
 ; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
+; AVX512F-NEXT:    kortestw %k0, %k0
+; AVX512F-NEXT:    setne %al
 ; AVX512F-NEXT:    vzeroupper
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: icmp0_v32i16_v32i1:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vptestnmw %zmm0, %zmm0, %k0
 ; AVX512BW-NEXT:    kortestd %k0, %k0
 ; AVX512BW-NEXT:    setne %al

 ; AVX2-NEXT:    retq
 ;
 ; AVX512F-LABEL: icmp0_v64i8_v64i1:
 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
 ; AVX512F-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; AVX512F-NEXT:    vpcmpeqb %ymm2, %ymm1, %ymm1
 ; AVX512F-NEXT:    vpcmpeqb %ymm2, %ymm0, %ymm0
-; AVX512F-NEXT:    vextracti128 $1, %ymm1, %xmm2
-; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm3
-; AVX512F-NEXT:    vpor %xmm2, %xmm3, %xmm2
-; AVX512F-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpor %xmm2, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxbd %xmm0, %zmm0
-; AVX512F-NEXT:    vpslld $31, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
+; AVX512F-NEXT:    vpor %ymm1, %ymm0, %ymm0
+; AVX512F-NEXT:    vpmovmskb %ymm0, %eax
+; AVX512F-NEXT:    testl %eax, %eax
+; AVX512F-NEXT:    setne %al
 ; AVX512F-NEXT:    vzeroupper
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: icmp0_v64i8_v64i1:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vptestnmb %zmm0, %zmm0, %k0
 ; AVX512BW-NEXT:    kortestq %k0, %k0
 ; AVX512BW-NEXT:    setne %al

 ; AVX2-NEXT:    testl %eax, %eax
 ; AVX2-NEXT:    setne %al
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512F-LABEL: icmp_v32i8_v32i1:
 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    vpcmpeqb %ymm1, %ymm0, %ymm0
-; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX512F-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxbd %xmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
+; AVX512F-NEXT:    vpmovmskb %ymm0, %eax
+; AVX512F-NEXT:    testl %eax, %eax
+; AVX512F-NEXT:    setne %al
 ; AVX512F-NEXT:    vzeroupper
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: icmp_v32i8_v32i1:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    # kill: def $ymm1 killed $ymm1 def $zmm1
 ; AVX512BW-NEXT:    # kill: def $ymm0 killed $ymm0 def $zmm0
 ; AVX512BW-NEXT:    vpcmpeqb %zmm1, %zmm0, %k0
; AVX512-NEXT:    retq
   %a = icmp eq <16 x i32> %0, %1
   %b = call i1 @llvm.vector.reduce.or.v16i1(<16 x i1> %a)
   ret i1 %b
 }
 
 define i1 @icmp_v32i16_v32i1(<32 x i16>, <32 x i16>) {
 ; SSE-LABEL: icmp_v32i16_v32i1:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    pcmpeqw %xmm5, %xmm1
-; SSE-NEXT:    pcmpeqw %xmm4, %xmm0
-; SSE-NEXT:    packsswb %xmm1, %xmm0
 ; SSE-NEXT:    pcmpeqw %xmm7, %xmm3
 ; SSE-NEXT:    pcmpeqw %xmm6, %xmm2
 ; SSE-NEXT:    packsswb %xmm3, %xmm2
-; SSE-NEXT:    por %xmm0, %xmm2
-; SSE-NEXT:    pmovmskb %xmm2, %eax
+; SSE-NEXT:    pcmpeqw %xmm5, %xmm1
+; SSE-NEXT:    pcmpeqw %xmm4, %xmm0
+; SSE-NEXT:    packsswb %xmm1, %xmm0
+; SSE-NEXT:    por %xmm2, %xmm0
+; SSE-NEXT:    pmovmskb %xmm0, %eax
 ; SSE-NEXT:    testl %eax, %eax
 ; SSE-NEXT:    setne %al
 ; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: icmp_v32i16_v32i1:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm3, %xmm4
 ; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm5

 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm1, %ymm2
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm3
 ; AVX512F-NEXT:    vpcmpeqw %ymm2, %ymm3, %ymm2
 ; AVX512F-NEXT:    vpcmpeqw %ymm1, %ymm0, %ymm0
 ; AVX512F-NEXT:    vpor %ymm2, %ymm0, %ymm0
 ; AVX512F-NEXT:    vpmovsxwd %ymm0, %zmm0
 ; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
+; AVX512F-NEXT:    kortestw %k0, %k0
+; AVX512F-NEXT:    setne %al
 ; AVX512F-NEXT:    vzeroupper
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: icmp_v32i16_v32i1:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vpcmpeqw %zmm1, %zmm0, %k0
 ; AVX512BW-NEXT:    kortestd %k0, %k0
 ; AVX512BW-NEXT:    setne %al

 ; AVX2-NEXT:    retq
 ;
 ; AVX512F-LABEL: icmp_v64i8_v64i1:
 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm1, %ymm2
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm3
 ; AVX512F-NEXT:    vpcmpeqb %ymm2, %ymm3, %ymm2
 ; AVX512F-NEXT:    vpcmpeqb %ymm1, %ymm0, %ymm0
-; AVX512F-NEXT:    vextracti128 $1, %ymm2, %xmm1
-; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm3
-; AVX512F-NEXT:    vpor %xmm1, %xmm3, %xmm1
-; AVX512F-NEXT:    vpor %xmm2, %xmm0, %xmm0
-; AVX512F-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX512F-NEXT:    vpmovsxbd %xmm0, %zmm0
-; AVX512F-NEXT:    vpslld $31, %zmm0, %zmm0
-; AVX512F-NEXT:    vptestmd %zmm0, %zmm0, %k0
-; AVX512F-NEXT:    kshiftrw $8, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $4, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $2, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kshiftrw $1, %k0, %k1
-; AVX512F-NEXT:    korw %k1, %k0, %k0
-; AVX512F-NEXT:    kmovw %k0, %eax
-; AVX512F-NEXT:    # kill: def $al killed $al killed $eax
+; AVX512F-NEXT:    vpor %ymm2, %ymm0, %ymm0
+; AVX512F-NEXT:    vpmovmskb %ymm0, %eax
+; AVX512F-NEXT:    testl %eax, %eax
+; AVX512F-NEXT:    setne %al
 ; AVX512F-NEXT:    vzeroupper
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: icmp_v64i8_v64i1:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vpcmpeqb %zmm1, %zmm0, %k0
 ; AVX512BW-NEXT:    kortestq %k0, %k0
 ; AVX512BW-NEXT:    setne %al