From: Simon Pilgrim <llvm-dev@redking.me.uk>
Date: Mon, 3 Apr 2023 11:37:53 +0000 (+0100)
Subject: [X86] Test coverage for icmp_eq(bitcast(vXi1 trunc(Y)),0) style reduction patterns
X-Git-Tag: upstream/17.0.6~12876
X-Git-Url: http://review.tizen.org/git/?a=commitdiff_plain;h=82df745e57be2b5765b0249e6afaff0627c470bd;p=platform%2Fupstream%2Fllvm.git

[X86] Test coverage for icmp_eq(bitcast(vXi1 trunc(Y)),0) style reduction patterns

Add SSE41 test coverage as well
---

diff --git a/llvm/test/CodeGen/X86/bitcast-vector-bool.ll b/llvm/test/CodeGen/X86/bitcast-vector-bool.ll
index 5dbd766..4c87e98 100644
--- a/llvm/test/CodeGen/X86/bitcast-vector-bool.ll
+++ b/llvm/test/CodeGen/X86/bitcast-vector-bool.ll
@@ -1,6 +1,7 @@
 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefix=SSE2-SSSE3
-; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+ssse3 | FileCheck %s --check-prefix=SSE2-SSSE3
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 | FileCheck %s --check-prefixes=SSE,SSE2-SSSE3
+; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+ssse3 | FileCheck %s --check-prefixes=SSE,SSE2-SSSE3
+; 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,AVX12,AVX1
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 | FileCheck %s --check-prefixes=AVX,AVX12,AVX2
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512vl,+avx512bw | FileCheck %s --check-prefixes=AVX,AVX512
@@ -10,13 +11,13 @@
 ;
 
 define i1 @bitcast_v2i64_to_v2i1(<2 x i64> %a0) nounwind {
-; SSE2-SSSE3-LABEL: bitcast_v2i64_to_v2i1:
-; SSE2-SSSE3:       # %bb.0:
-; SSE2-SSSE3-NEXT:    movmskpd %xmm0, %ecx
-; SSE2-SSSE3-NEXT:    movl %ecx, %eax
-; SSE2-SSSE3-NEXT:    shrb %al
-; SSE2-SSSE3-NEXT:    addb %cl, %al
-; SSE2-SSSE3-NEXT:    retq
+; SSE-LABEL: bitcast_v2i64_to_v2i1:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movmskpd %xmm0, %ecx
+; SSE-NEXT:    movl %ecx, %eax
+; SSE-NEXT:    shrb %al
+; SSE-NEXT:    addb %cl, %al
+; SSE-NEXT:    retq
 ;
 ; AVX12-LABEL: bitcast_v2i64_to_v2i1:
 ; AVX12:       # %bb.0:
@@ -44,16 +45,47 @@ define i1 @bitcast_v2i64_to_v2i1(<2 x i64> %a0) nounwind {
   ret i1 %5
 }
 
+define i1 @trunc_v2i64_cmp(<2 x i64> %a0) nounwind {
+; SSE-LABEL: trunc_v2i64_cmp:
+; SSE:       # %bb.0:
+; SSE-NEXT:    psllq $63, %xmm0
+; SSE-NEXT:    movmskpd %xmm0, %eax
+; SSE-NEXT:    testl %eax, %eax
+; SSE-NEXT:    sete %al
+; SSE-NEXT:    retq
+;
+; AVX12-LABEL: trunc_v2i64_cmp:
+; AVX12:       # %bb.0:
+; AVX12-NEXT:    vpsllq $63, %xmm0, %xmm0
+; AVX12-NEXT:    vmovmskpd %xmm0, %eax
+; AVX12-NEXT:    testl %eax, %eax
+; AVX12-NEXT:    sete %al
+; AVX12-NEXT:    retq
+;
+; AVX512-LABEL: trunc_v2i64_cmp:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpsllq $63, %xmm0, %xmm0
+; AVX512-NEXT:    vptestmq %xmm0, %xmm0, %k0
+; AVX512-NEXT:    kmovd %k0, %eax
+; AVX512-NEXT:    testb %al, %al
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    retq
+  %1 = trunc <2 x i64> %a0 to <2 x i1>
+  %2 = bitcast <2 x i1> %1 to i2
+  %3 = icmp eq i2 %2, 0
+  ret i1 %3
+}
+
 define i2 @bitcast_v4i32_to_v2i2(<4 x i32> %a0) nounwind {
-; SSE2-SSSE3-LABEL: bitcast_v4i32_to_v2i2:
-; SSE2-SSSE3:       # %bb.0:
-; SSE2-SSSE3-NEXT:    movmskps %xmm0, %eax
-; SSE2-SSSE3-NEXT:    movl %eax, %ecx
-; SSE2-SSSE3-NEXT:    shrb $2, %cl
-; SSE2-SSSE3-NEXT:    andb $3, %al
-; SSE2-SSSE3-NEXT:    addb %cl, %al
-; SSE2-SSSE3-NEXT:    # kill: def $al killed $al killed $eax
-; SSE2-SSSE3-NEXT:    retq
+; SSE-LABEL: bitcast_v4i32_to_v2i2:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movmskps %xmm0, %eax
+; SSE-NEXT:    movl %eax, %ecx
+; SSE-NEXT:    shrb $2, %cl
+; SSE-NEXT:    andb $3, %al
+; SSE-NEXT:    addb %cl, %al
+; SSE-NEXT:    # kill: def $al killed $al killed $eax
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: bitcast_v4i32_to_v2i2:
 ; AVX:       # %bb.0:
@@ -72,17 +104,48 @@ define i2 @bitcast_v4i32_to_v2i2(<4 x i32> %a0) nounwind {
   ret i2 %5
 }
 
+define i1 @trunc_v4i32_cmp(<4 x i32> %a0) nounwind {
+; SSE-LABEL: trunc_v4i32_cmp:
+; 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
+;
+; AVX12-LABEL: trunc_v4i32_cmp:
+; AVX12:       # %bb.0:
+; AVX12-NEXT:    vpslld $31, %xmm0, %xmm0
+; AVX12-NEXT:    vmovmskps %xmm0, %eax
+; AVX12-NEXT:    cmpb $15, %al
+; AVX12-NEXT:    sete %al
+; AVX12-NEXT:    retq
+;
+; AVX512-LABEL: trunc_v4i32_cmp:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpslld $31, %xmm0, %xmm0
+; AVX512-NEXT:    vptestnmd %xmm0, %xmm0, %k0
+; AVX512-NEXT:    kmovd %k0, %eax
+; AVX512-NEXT:    testb $15, %al
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    retq
+  %1 = trunc <4 x i32> %a0 to <4 x i1>
+  %2 = bitcast <4 x i1> %1 to i4
+  %3 = icmp eq i4 %2, -1
+  ret i1 %3
+}
+
 define i4 @bitcast_v8i16_to_v2i4(<8 x i16> %a0) nounwind {
-; SSE2-SSSE3-LABEL: bitcast_v8i16_to_v2i4:
-; SSE2-SSSE3:       # %bb.0:
-; SSE2-SSSE3-NEXT:    packsswb %xmm0, %xmm0
-; SSE2-SSSE3-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-SSSE3-NEXT:    movl %eax, %ecx
-; SSE2-SSSE3-NEXT:    shrb $4, %cl
-; SSE2-SSSE3-NEXT:    andb $15, %al
-; SSE2-SSSE3-NEXT:    addb %cl, %al
-; SSE2-SSSE3-NEXT:    # kill: def $al killed $al killed $eax
-; SSE2-SSSE3-NEXT:    retq
+; SSE-LABEL: bitcast_v8i16_to_v2i4:
+; SSE:       # %bb.0:
+; SSE-NEXT:    packsswb %xmm0, %xmm0
+; SSE-NEXT:    pmovmskb %xmm0, %eax
+; SSE-NEXT:    movl %eax, %ecx
+; SSE-NEXT:    shrb $4, %cl
+; SSE-NEXT:    andb $15, %al
+; SSE-NEXT:    addb %cl, %al
+; SSE-NEXT:    # kill: def $al killed $al killed $eax
+; SSE-NEXT:    retq
 ;
 ; AVX12-LABEL: bitcast_v8i16_to_v2i4:
 ; AVX12:       # %bb.0:
@@ -113,6 +176,37 @@ define i4 @bitcast_v8i16_to_v2i4(<8 x i16> %a0) nounwind {
   ret i4 %5
 }
 
+define i1 @trunc_v8i16_cmp(<8 x i16> %a0) nounwind {
+; SSE-LABEL: trunc_v8i16_cmp:
+; 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
+;
+; AVX12-LABEL: trunc_v8i16_cmp:
+; AVX12:       # %bb.0:
+; AVX12-NEXT:    vpsllw $15, %xmm0, %xmm0
+; AVX12-NEXT:    vpmovmskb %xmm0, %eax
+; AVX12-NEXT:    testl $43690, %eax # imm = 0xAAAA
+; AVX12-NEXT:    setne %al
+; AVX12-NEXT:    retq
+;
+; AVX512-LABEL: trunc_v8i16_cmp:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpsllw $15, %xmm0, %xmm0
+; AVX512-NEXT:    vpmovw2m %xmm0, %k0
+; AVX512-NEXT:    kmovd %k0, %eax
+; AVX512-NEXT:    testb %al, %al
+; AVX512-NEXT:    setne %al
+; AVX512-NEXT:    retq
+  %1 = trunc <8 x i16> %a0 to <8 x i1>
+  %2 = bitcast <8 x i1> %1 to i8
+  %3 = icmp ne i8 %2, 0
+  ret i1 %3
+}
+
 define i8 @bitcast_v16i8_to_v2i8(<16 x i8> %a0) nounwind {
 ; SSE2-SSSE3-LABEL: bitcast_v16i8_to_v2i8:
 ; SSE2-SSSE3:       # %bb.0:
@@ -123,6 +217,15 @@ define i8 @bitcast_v16i8_to_v2i8(<16 x i8> %a0) nounwind {
 ; SSE2-SSSE3-NEXT:    addb -{{[0-9]+}}(%rsp), %al
 ; SSE2-SSSE3-NEXT:    retq
 ;
+; SSE41-LABEL: bitcast_v16i8_to_v2i8:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    pmovmskb %xmm0, %ecx
+; SSE41-NEXT:    movl %ecx, %eax
+; SSE41-NEXT:    shrl $8, %eax
+; SSE41-NEXT:    addb %cl, %al
+; SSE41-NEXT:    # kill: def $al killed $al killed $eax
+; SSE41-NEXT:    retq
+;
 ; AVX12-LABEL: bitcast_v16i8_to_v2i8:
 ; AVX12:       # %bb.0:
 ; AVX12-NEXT:    vpmovmskb %xmm0, %ecx
@@ -149,21 +252,43 @@ define i8 @bitcast_v16i8_to_v2i8(<16 x i8> %a0) nounwind {
   ret i8 %5
 }
 
+define i1 @trunc_v16i8_cmp(<16 x i8> %a0) nounwind {
+; SSE-LABEL: trunc_v16i8_cmp:
+; SSE:       # %bb.0:
+; SSE-NEXT:    psllw $7, %xmm0
+; SSE-NEXT:    pmovmskb %xmm0, %eax
+; SSE-NEXT:    cmpw $-1, %ax
+; SSE-NEXT:    setne %al
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: trunc_v16i8_cmp:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpsllw $7, %xmm0, %xmm0
+; AVX-NEXT:    vpmovmskb %xmm0, %eax
+; AVX-NEXT:    cmpw $-1, %ax
+; AVX-NEXT:    setne %al
+; AVX-NEXT:    retq
+  %1 = trunc <16 x i8> %a0 to <16 x i1>
+  %2 = bitcast <16 x i1> %1 to i16
+  %3 = icmp ne i16 %2, -1
+  ret i1 %3
+}
+
 ;
 ; 256-bit vectors
 ;
 
 define i2 @bitcast_v4i64_to_v2i2(<4 x i64> %a0) nounwind {
-; SSE2-SSSE3-LABEL: bitcast_v4i64_to_v2i2:
-; SSE2-SSSE3:       # %bb.0:
-; SSE2-SSSE3-NEXT:    packssdw %xmm1, %xmm0
-; SSE2-SSSE3-NEXT:    movmskps %xmm0, %eax
-; SSE2-SSSE3-NEXT:    movl %eax, %ecx
-; SSE2-SSSE3-NEXT:    shrb $2, %cl
-; SSE2-SSSE3-NEXT:    andb $3, %al
-; SSE2-SSSE3-NEXT:    addb %cl, %al
-; SSE2-SSSE3-NEXT:    # kill: def $al killed $al killed $eax
-; SSE2-SSSE3-NEXT:    retq
+; SSE-LABEL: bitcast_v4i64_to_v2i2:
+; SSE:       # %bb.0:
+; SSE-NEXT:    packssdw %xmm1, %xmm0
+; SSE-NEXT:    movmskps %xmm0, %eax
+; SSE-NEXT:    movl %eax, %ecx
+; SSE-NEXT:    shrb $2, %cl
+; SSE-NEXT:    andb $3, %al
+; SSE-NEXT:    addb %cl, %al
+; SSE-NEXT:    # kill: def $al killed $al killed $eax
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: bitcast_v4i64_to_v2i2:
 ; AVX:       # %bb.0:
@@ -183,18 +308,63 @@ define i2 @bitcast_v4i64_to_v2i2(<4 x i64> %a0) nounwind {
   ret i2 %5
 }
 
+define i1 @trunc_v4i64_cmp(<4 x i64> %a0) nounwind {
+; SSE-LABEL: trunc_v4i64_cmp:
+; 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
+;
+; AVX1-LABEL: trunc_v4i64_cmp:
+; 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:    setne %al
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: trunc_v4i64_cmp:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpsllq $63, %ymm0, %ymm0
+; AVX2-NEXT:    vmovmskpd %ymm0, %eax
+; AVX2-NEXT:    testl %eax, %eax
+; AVX2-NEXT:    setne %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
+;
+; AVX512-LABEL: trunc_v4i64_cmp:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpsllq $63, %ymm0, %ymm0
+; AVX512-NEXT:    vptestmq %ymm0, %ymm0, %k0
+; AVX512-NEXT:    kmovd %k0, %eax
+; AVX512-NEXT:    testb %al, %al
+; AVX512-NEXT:    setne %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
+  %1 = trunc <4 x i64> %a0 to <4 x i1>
+  %2 = bitcast <4 x i1> %1 to i4
+  %3 = icmp ne i4 %2, 0
+  ret i1 %3
+}
+
 define i4 @bitcast_v8i32_to_v2i4(<8 x i32> %a0) nounwind {
-; SSE2-SSSE3-LABEL: bitcast_v8i32_to_v2i4:
-; SSE2-SSSE3:       # %bb.0:
-; SSE2-SSSE3-NEXT:    packssdw %xmm1, %xmm0
-; SSE2-SSSE3-NEXT:    packsswb %xmm0, %xmm0
-; SSE2-SSSE3-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-SSSE3-NEXT:    movl %eax, %ecx
-; SSE2-SSSE3-NEXT:    shrb $4, %cl
-; SSE2-SSSE3-NEXT:    andb $15, %al
-; SSE2-SSSE3-NEXT:    addb %cl, %al
-; SSE2-SSSE3-NEXT:    # kill: def $al killed $al killed $eax
-; SSE2-SSSE3-NEXT:    retq
+; SSE-LABEL: bitcast_v8i32_to_v2i4:
+; SSE:       # %bb.0:
+; SSE-NEXT:    packssdw %xmm1, %xmm0
+; SSE-NEXT:    packsswb %xmm0, %xmm0
+; SSE-NEXT:    pmovmskb %xmm0, %eax
+; SSE-NEXT:    movl %eax, %ecx
+; SSE-NEXT:    shrb $4, %cl
+; SSE-NEXT:    andb $15, %al
+; SSE-NEXT:    addb %cl, %al
+; SSE-NEXT:    # kill: def $al killed $al killed $eax
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: bitcast_v8i32_to_v2i4:
 ; AVX:       # %bb.0:
@@ -214,6 +384,55 @@ define i4 @bitcast_v8i32_to_v2i4(<8 x i32> %a0) nounwind {
   ret i4 %5
 }
 
+define i1 @trunc_v8i132_cmp(<8 x i32> %a0) nounwind {
+; SSE41-LABEL: trunc_v8i132_cmp:
+; 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:    setne %al
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: trunc_v8i132_cmp:
+; 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:    setne %al
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: trunc_v8i132_cmp:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpslld $31, %ymm0, %ymm0
+; AVX2-NEXT:    vmovmskps %ymm0, %eax
+; AVX2-NEXT:    cmpb $-1, %al
+; AVX2-NEXT:    setne %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
+;
+; AVX512-LABEL: trunc_v8i132_cmp:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpslld $31, %ymm0, %ymm0
+; AVX512-NEXT:    vptestmd %ymm0, %ymm0, %k0
+; AVX512-NEXT:    kmovd %k0, %eax
+; AVX512-NEXT:    cmpb $-1, %al
+; AVX512-NEXT:    setne %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
+  %1 = trunc <8 x i32> %a0 to <8 x i1>
+  %2 = bitcast <8 x i1> %1 to i8
+  %3 = icmp ne i8 %2, -1
+  ret i1 %3
+}
+
 define i8 @bitcast_v16i16_to_v2i8(<16 x i16> %a0) nounwind {
 ; SSE2-SSSE3-LABEL: bitcast_v16i16_to_v2i8:
 ; SSE2-SSSE3:       # %bb.0:
@@ -225,6 +444,16 @@ define i8 @bitcast_v16i16_to_v2i8(<16 x i16> %a0) nounwind {
 ; SSE2-SSSE3-NEXT:    addb -{{[0-9]+}}(%rsp), %al
 ; SSE2-SSSE3-NEXT:    retq
 ;
+; SSE41-LABEL: bitcast_v16i16_to_v2i8:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    packsswb %xmm1, %xmm0
+; SSE41-NEXT:    pmovmskb %xmm0, %ecx
+; SSE41-NEXT:    movl %ecx, %eax
+; SSE41-NEXT:    shrl $8, %eax
+; SSE41-NEXT:    addb %cl, %al
+; SSE41-NEXT:    # kill: def $al killed $al killed $eax
+; SSE41-NEXT:    retq
+;
 ; AVX1-LABEL: bitcast_v16i16_to_v2i8:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
@@ -267,14 +496,65 @@ define i8 @bitcast_v16i16_to_v2i8(<16 x i16> %a0) nounwind {
   ret i8 %5
 }
 
+define i1 @trunc_v16i16_cmp(<16 x i16> %a0) nounwind {
+; SSE-LABEL: trunc_v16i16_cmp:
+; 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:    sete %al
+; SSE-NEXT:    retq
+;
+; AVX1-LABEL: trunc_v16i16_cmp:
+; 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:    sete %al
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: trunc_v16i16_cmp:
+; 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:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
+;
+; AVX512-LABEL: trunc_v16i16_cmp:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpsllw $15, %ymm0, %ymm0
+; AVX512-NEXT:    vpmovw2m %ymm0, %k0
+; AVX512-NEXT:    kortestw %k0, %k0
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
+  %1 = trunc <16 x i16> %a0 to <16 x i1>
+  %2 = bitcast <16 x i1> %1 to i16
+  %3 = icmp eq i16 %2, 0
+  ret i1 %3
+}
+
 define i16 @bitcast_v32i8_to_v2i16(<32 x i8> %a0) nounwind {
-; SSE2-SSSE3-LABEL: bitcast_v32i8_to_v2i16:
-; SSE2-SSSE3:       # %bb.0:
-; SSE2-SSSE3-NEXT:    pmovmskb %xmm1, %ecx
-; SSE2-SSSE3-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-SSSE3-NEXT:    addl %ecx, %eax
-; SSE2-SSSE3-NEXT:    # kill: def $ax killed $ax killed $eax
-; SSE2-SSSE3-NEXT:    retq
+; SSE-LABEL: bitcast_v32i8_to_v2i16:
+; SSE:       # %bb.0:
+; SSE-NEXT:    pmovmskb %xmm1, %ecx
+; SSE-NEXT:    pmovmskb %xmm0, %eax
+; SSE-NEXT:    addl %ecx, %eax
+; SSE-NEXT:    # kill: def $ax killed $ax killed $eax
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: bitcast_v32i8_to_v2i16:
 ; AVX1:       # %bb.0:
@@ -314,24 +594,68 @@ define i16 @bitcast_v32i8_to_v2i16(<32 x i8> %a0) nounwind {
   ret i16 %5
 }
 
+define i1 @trunc_v32i8_cmp(<32 x i8> %a0) nounwind {
+; SSE-LABEL: trunc_v32i8_cmp:
+; 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_cmp:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpand %xmm0, %xmm1, %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_cmp:
+; 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
+;
+; AVX512-LABEL: trunc_v32i8_cmp:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpsllw $7, %ymm0, %ymm0
+; AVX512-NEXT:    vpmovmskb %ymm0, %eax
+; AVX512-NEXT:    cmpl $-1, %eax
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
+  %1 = trunc <32 x i8> %a0 to <32 x i1>
+  %2 = bitcast <32 x i1> %1 to i32
+  %3 = icmp eq i32 %2, -1
+  ret i1 %3
+}
+
 ;
 ; 512-bit vectors
 ;
 
 define i4 @bitcast_v8i64_to_v2i4(<8 x i64> %a0) nounwind {
-; SSE2-SSSE3-LABEL: bitcast_v8i64_to_v2i4:
-; SSE2-SSSE3:       # %bb.0:
-; SSE2-SSSE3-NEXT:    packssdw %xmm3, %xmm2
-; SSE2-SSSE3-NEXT:    packssdw %xmm1, %xmm0
-; SSE2-SSSE3-NEXT:    packssdw %xmm2, %xmm0
-; SSE2-SSSE3-NEXT:    packsswb %xmm0, %xmm0
-; SSE2-SSSE3-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-SSSE3-NEXT:    movl %eax, %ecx
-; SSE2-SSSE3-NEXT:    shrb $4, %cl
-; SSE2-SSSE3-NEXT:    andb $15, %al
-; SSE2-SSSE3-NEXT:    addb %cl, %al
-; SSE2-SSSE3-NEXT:    # kill: def $al killed $al killed $eax
-; SSE2-SSSE3-NEXT:    retq
+; SSE-LABEL: bitcast_v8i64_to_v2i4:
+; SSE:       # %bb.0:
+; SSE-NEXT:    packssdw %xmm3, %xmm2
+; SSE-NEXT:    packssdw %xmm1, %xmm0
+; SSE-NEXT:    packssdw %xmm2, %xmm0
+; SSE-NEXT:    packsswb %xmm0, %xmm0
+; SSE-NEXT:    pmovmskb %xmm0, %eax
+; SSE-NEXT:    movl %eax, %ecx
+; SSE-NEXT:    shrb $4, %cl
+; SSE-NEXT:    andb $15, %al
+; SSE-NEXT:    addb %cl, %al
+; SSE-NEXT:    # kill: def $al killed $al killed $eax
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: bitcast_v8i64_to_v2i4:
 ; AVX1:       # %bb.0:
@@ -385,6 +709,85 @@ define i4 @bitcast_v8i64_to_v2i4(<8 x i64> %a0) nounwind {
   ret i4 %5
 }
 
+define i1 @trunc_v8i64_cmp(<8 x i64> %a0) nounwind {
+; SSE2-SSSE3-LABEL: trunc_v8i64_cmp:
+; SSE2-SSSE3:       # %bb.0:
+; SSE2-SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; SSE2-SSSE3-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
+; SSE2-SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SSE2-SSSE3-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSE2-SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE2-SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[0,2,2,3]
+; SSE2-SSSE3-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,1,0,2,4,5,6,7]
+; SSE2-SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
+; SSE2-SSSE3-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,0,2,4,5,6,7]
+; SSE2-SSSE3-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
+; SSE2-SSSE3-NEXT:    movsd {{.*#+}} xmm2 = xmm0[0],xmm2[1]
+; SSE2-SSSE3-NEXT:    psllw $15, %xmm2
+; SSE2-SSSE3-NEXT:    packsswb %xmm2, %xmm2
+; SSE2-SSSE3-NEXT:    pmovmskb %xmm2, %eax
+; SSE2-SSSE3-NEXT:    cmpb $-1, %al
+; SSE2-SSSE3-NEXT:    sete %al
+; SSE2-SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc_v8i64_cmp:
+; 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:    retq
+;
+; AVX1-LABEL: trunc_v8i64_cmp:
+; 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:    vzeroupper
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: trunc_v8i64_cmp:
+; 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:    vzeroupper
+; AVX2-NEXT:    retq
+;
+; AVX512-LABEL: trunc_v8i64_cmp:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpsllq $63, %zmm0, %zmm0
+; AVX512-NEXT:    vptestmq %zmm0, %zmm0, %k0
+; AVX512-NEXT:    kmovd %k0, %eax
+; AVX512-NEXT:    cmpb $-1, %al
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
+  %1 = trunc <8 x i64> %a0 to <8 x i1>
+  %2 = bitcast <8 x i1> %1 to i8
+  %3 = icmp eq i8 %2, -1
+  ret i1 %3
+}
+
 define i8 @bitcast_v16i32_to_v2i8(<16 x i32> %a0) nounwind {
 ; SSE2-SSSE3-LABEL: bitcast_v16i32_to_v2i8:
 ; SSE2-SSSE3:       # %bb.0:
@@ -398,6 +801,18 @@ define i8 @bitcast_v16i32_to_v2i8(<16 x i32> %a0) nounwind {
 ; SSE2-SSSE3-NEXT:    addb -{{[0-9]+}}(%rsp), %al
 ; SSE2-SSSE3-NEXT:    retq
 ;
+; SSE41-LABEL: bitcast_v16i32_to_v2i8:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    packssdw %xmm3, %xmm2
+; SSE41-NEXT:    packssdw %xmm1, %xmm0
+; SSE41-NEXT:    packsswb %xmm2, %xmm0
+; SSE41-NEXT:    pmovmskb %xmm0, %ecx
+; SSE41-NEXT:    movl %ecx, %eax
+; SSE41-NEXT:    shrl $8, %eax
+; SSE41-NEXT:    addb %cl, %al
+; SSE41-NEXT:    # kill: def $al killed $al killed $eax
+; SSE41-NEXT:    retq
+;
 ; AVX1-LABEL: bitcast_v16i32_to_v2i8:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
@@ -449,17 +864,97 @@ define i8 @bitcast_v16i32_to_v2i8(<16 x i32> %a0) nounwind {
   ret i8 %5
 }
 
-define i16 @bitcast_v32i16_to_v2i16(<32 x i16> %a0) nounwind {
-; SSE2-SSSE3-LABEL: bitcast_v32i16_to_v2i16:
+define i1 @trunc_v16i32_cmp(<16 x i32> %a0) nounwind {
+; SSE2-SSSE3-LABEL: trunc_v16i32_cmp:
 ; SSE2-SSSE3:       # %bb.0:
-; SSE2-SSSE3-NEXT:    packsswb %xmm3, %xmm2
-; SSE2-SSSE3-NEXT:    pmovmskb %xmm2, %ecx
-; SSE2-SSSE3-NEXT:    packsswb %xmm1, %xmm0
+; SSE2-SSSE3-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
+; SSE2-SSSE3-NEXT:    pand %xmm4, %xmm3
+; SSE2-SSSE3-NEXT:    pand %xmm4, %xmm2
+; SSE2-SSSE3-NEXT:    packuswb %xmm3, %xmm2
+; SSE2-SSSE3-NEXT:    pand %xmm4, %xmm1
+; SSE2-SSSE3-NEXT:    pand %xmm4, %xmm0
+; SSE2-SSSE3-NEXT:    packuswb %xmm1, %xmm0
+; SSE2-SSSE3-NEXT:    packuswb %xmm2, %xmm0
+; SSE2-SSSE3-NEXT:    psllw $7, %xmm0
 ; SSE2-SSSE3-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-SSSE3-NEXT:    addl %ecx, %eax
-; SSE2-SSSE3-NEXT:    # kill: def $ax killed $ax killed $eax
+; SSE2-SSSE3-NEXT:    testl %eax, %eax
+; SSE2-SSSE3-NEXT:    sete %al
 ; SSE2-SSSE3-NEXT:    retq
 ;
+; SSE41-LABEL: trunc_v16i32_cmp:
+; 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:    sete %al
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: trunc_v16i32_cmp:
+; 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:    sete %al
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: trunc_v16i32_cmp:
+; 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:    sete %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
+;
+; AVX512-LABEL: trunc_v16i32_cmp:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpslld $31, %zmm0, %zmm0
+; AVX512-NEXT:    vptestmd %zmm0, %zmm0, %k0
+; AVX512-NEXT:    kortestw %k0, %k0
+; AVX512-NEXT:    sete %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
+  %1 = trunc <16 x i32> %a0 to <16 x i1>
+  %2 = bitcast <16 x i1> %1 to i16
+  %3 = icmp eq i16 %2, 0
+  ret i1 %3
+}
+
+define i16 @bitcast_v32i16_to_v2i16(<32 x i16> %a0) nounwind {
+; SSE-LABEL: bitcast_v32i16_to_v2i16:
+; SSE:       # %bb.0:
+; SSE-NEXT:    packsswb %xmm3, %xmm2
+; SSE-NEXT:    pmovmskb %xmm2, %ecx
+; SSE-NEXT:    packsswb %xmm1, %xmm0
+; SSE-NEXT:    pmovmskb %xmm0, %eax
+; SSE-NEXT:    addl %ecx, %eax
+; SSE-NEXT:    # kill: def $ax killed $ax killed $eax
+; SSE-NEXT:    retq
+;
 ; AVX1-LABEL: bitcast_v32i16_to_v2i16:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
@@ -503,6 +998,68 @@ define i16 @bitcast_v32i16_to_v2i16(<32 x i16> %a0) nounwind {
   ret i16 %5
 }
 
+define i1 @trunc_v32i16_cmp(<32 x i16> %a0) nounwind {
+; SSE-LABEL: trunc_v32i16_cmp:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
+; SSE-NEXT:    pand %xmm4, %xmm1
+; SSE-NEXT:    pand %xmm4, %xmm0
+; SSE-NEXT:    packuswb %xmm1, %xmm0
+; SSE-NEXT:    pand %xmm4, %xmm3
+; SSE-NEXT:    pand %xmm4, %xmm2
+; SSE-NEXT:    packuswb %xmm3, %xmm2
+; SSE-NEXT:    pand %xmm0, %xmm2
+; SSE-NEXT:    psllw $7, %xmm2
+; SSE-NEXT:    pmovmskb %xmm2, %eax
+; SSE-NEXT:    cmpw $-1, %ax
+; SSE-NEXT:    setne %al
+; SSE-NEXT:    retq
+;
+; AVX1-LABEL: trunc_v32i16_cmp:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vmovaps {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
+; AVX1-NEXT:    vandps %ymm2, %ymm0, %ymm0
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm3
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vandps %ymm2, %ymm1, %ymm1
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpand %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    cmpw $-1, %ax
+; AVX1-NEXT:    setne %al
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: trunc_v32i16_cmp:
+; 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:    setne %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
+;
+; AVX512-LABEL: trunc_v32i16_cmp:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpsllw $15, %zmm0, %zmm0
+; AVX512-NEXT:    vpmovw2m %zmm0, %k0
+; AVX512-NEXT:    kortestd %k0, %k0
+; AVX512-NEXT:    setae %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
+  %1 = trunc <32 x i16> %a0 to <32 x i1>
+  %2 = bitcast <32 x i1> %1 to i32
+  %3 = icmp ne i32 %2, -1
+  ret i1 %3
+}
+
 define i32 @bitcast_v64i8_to_v2i32(<64 x i8> %a0) nounwind {
 ; SSE2-SSSE3-LABEL: bitcast_v64i8_to_v2i32:
 ; SSE2-SSSE3:       # %bb.0:
@@ -522,6 +1079,19 @@ define i32 @bitcast_v64i8_to_v2i32(<64 x i8> %a0) nounwind {
 ; SSE2-SSSE3-NEXT:    addl %ecx, %eax
 ; SSE2-SSSE3-NEXT:    retq
 ;
+; SSE41-LABEL: bitcast_v64i8_to_v2i32:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    pmovmskb %xmm2, %eax
+; SSE41-NEXT:    pmovmskb %xmm3, %ecx
+; SSE41-NEXT:    shll $16, %ecx
+; SSE41-NEXT:    orl %eax, %ecx
+; SSE41-NEXT:    pmovmskb %xmm0, %edx
+; SSE41-NEXT:    pmovmskb %xmm1, %eax
+; SSE41-NEXT:    shll $16, %eax
+; SSE41-NEXT:    orl %edx, %eax
+; SSE41-NEXT:    addl %ecx, %eax
+; SSE41-NEXT:    retq
+;
 ; AVX1-LABEL: bitcast_v64i8_to_v2i32:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vpmovmskb %xmm1, %eax
@@ -563,31 +1133,81 @@ define i32 @bitcast_v64i8_to_v2i32(<64 x i8> %a0) nounwind {
   ret i32 %5
 }
 
+define i1 @trunc_v64i8_cmp(<64 x i8> %a0) nounwind {
+; SSE-LABEL: trunc_v64i8_cmp:
+; 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
+;
+; AVX1-LABEL: trunc_v64i8_cmp:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm3
+; AVX1-NEXT:    vpor %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpor %xmm2, %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_v64i8_cmp:
+; 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:    setne %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
+;
+; AVX512-LABEL: trunc_v64i8_cmp:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpsllw $7, %zmm0, %zmm0
+; AVX512-NEXT:    vpmovb2m %zmm0, %k0
+; AVX512-NEXT:    kortestq %k0, %k0
+; AVX512-NEXT:    setne %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
+  %1 = trunc <64 x i8> %a0 to <64 x i1>
+  %2 = bitcast <64 x i1> %1 to i64
+  %3 = icmp ne i64 %2, 0
+  ret i1 %3
+}
+
 define i64 @bitcast_v128i8_to_v2i64(<128 x i8> %a0) nounwind {
-; SSE2-SSSE3-LABEL: bitcast_v128i8_to_v2i64:
-; SSE2-SSSE3:       # %bb.0:
-; SSE2-SSSE3-NEXT:    pmovmskb %xmm4, %eax
-; SSE2-SSSE3-NEXT:    pmovmskb %xmm5, %ecx
-; SSE2-SSSE3-NEXT:    shll $16, %ecx
-; SSE2-SSSE3-NEXT:    orl %eax, %ecx
-; SSE2-SSSE3-NEXT:    pmovmskb %xmm6, %eax
-; SSE2-SSSE3-NEXT:    pmovmskb %xmm7, %edx
-; SSE2-SSSE3-NEXT:    shll $16, %edx
-; SSE2-SSSE3-NEXT:    orl %eax, %edx
-; SSE2-SSSE3-NEXT:    shlq $32, %rdx
-; SSE2-SSSE3-NEXT:    orq %rcx, %rdx
-; SSE2-SSSE3-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-SSSE3-NEXT:    pmovmskb %xmm1, %ecx
-; SSE2-SSSE3-NEXT:    shll $16, %ecx
-; SSE2-SSSE3-NEXT:    orl %eax, %ecx
-; SSE2-SSSE3-NEXT:    pmovmskb %xmm2, %esi
-; SSE2-SSSE3-NEXT:    pmovmskb %xmm3, %eax
-; SSE2-SSSE3-NEXT:    shll $16, %eax
-; SSE2-SSSE3-NEXT:    orl %esi, %eax
-; SSE2-SSSE3-NEXT:    shlq $32, %rax
-; SSE2-SSSE3-NEXT:    orq %rcx, %rax
-; SSE2-SSSE3-NEXT:    addq %rdx, %rax
-; SSE2-SSSE3-NEXT:    retq
+; SSE-LABEL: bitcast_v128i8_to_v2i64:
+; SSE:       # %bb.0:
+; SSE-NEXT:    pmovmskb %xmm4, %eax
+; SSE-NEXT:    pmovmskb %xmm5, %ecx
+; SSE-NEXT:    shll $16, %ecx
+; SSE-NEXT:    orl %eax, %ecx
+; SSE-NEXT:    pmovmskb %xmm6, %eax
+; SSE-NEXT:    pmovmskb %xmm7, %edx
+; SSE-NEXT:    shll $16, %edx
+; SSE-NEXT:    orl %eax, %edx
+; SSE-NEXT:    shlq $32, %rdx
+; SSE-NEXT:    orq %rcx, %rdx
+; SSE-NEXT:    pmovmskb %xmm0, %eax
+; SSE-NEXT:    pmovmskb %xmm1, %ecx
+; SSE-NEXT:    shll $16, %ecx
+; SSE-NEXT:    orl %eax, %ecx
+; SSE-NEXT:    pmovmskb %xmm2, %esi
+; SSE-NEXT:    pmovmskb %xmm3, %eax
+; SSE-NEXT:    shll $16, %eax
+; SSE-NEXT:    orl %esi, %eax
+; SSE-NEXT:    shlq $32, %rax
+; SSE-NEXT:    orq %rcx, %rax
+; SSE-NEXT:    addq %rdx, %rax
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: bitcast_v128i8_to_v2i64:
 ; AVX1:       # %bb.0:
@@ -650,24 +1270,191 @@ define i64 @bitcast_v128i8_to_v2i64(<128 x i8> %a0) nounwind {
   ret i64 %5
 }
 
-define [2 x i8] @PR58546(<16 x float> %a0) {
-; SSE2-SSSE3-LABEL: PR58546:
+define i1 @trunc_v128i8_cmp(<128 x i8> %a0) nounwind {
+; SSE2-SSSE3-LABEL: trunc_v128i8_cmp:
 ; SSE2-SSSE3:       # %bb.0:
-; SSE2-SSSE3-NEXT:    xorps %xmm4, %xmm4
-; SSE2-SSSE3-NEXT:    cmpunordps %xmm4, %xmm3
-; SSE2-SSSE3-NEXT:    cmpunordps %xmm4, %xmm2
-; SSE2-SSSE3-NEXT:    packssdw %xmm3, %xmm2
-; SSE2-SSSE3-NEXT:    cmpunordps %xmm4, %xmm1
-; SSE2-SSSE3-NEXT:    cmpunordps %xmm4, %xmm0
-; SSE2-SSSE3-NEXT:    packssdw %xmm1, %xmm0
-; SSE2-SSSE3-NEXT:    packsswb %xmm2, %xmm0
+; SSE2-SSSE3-NEXT:    psllw $7, %xmm0
+; SSE2-SSSE3-NEXT:    pmovmskb %xmm0, %eax
+; SSE2-SSSE3-NEXT:    psllw $7, %xmm1
+; SSE2-SSSE3-NEXT:    pmovmskb %xmm1, %ecx
+; SSE2-SSSE3-NEXT:    shll $16, %ecx
+; SSE2-SSSE3-NEXT:    orl %eax, %ecx
+; SSE2-SSSE3-NEXT:    psllw $7, %xmm2
+; SSE2-SSSE3-NEXT:    pmovmskb %xmm2, %edx
+; SSE2-SSSE3-NEXT:    psllw $7, %xmm3
+; SSE2-SSSE3-NEXT:    pmovmskb %xmm3, %eax
+; SSE2-SSSE3-NEXT:    shll $16, %eax
+; SSE2-SSSE3-NEXT:    orl %edx, %eax
+; SSE2-SSSE3-NEXT:    shlq $32, %rax
+; SSE2-SSSE3-NEXT:    orq %rcx, %rax
+; SSE2-SSSE3-NEXT:    psllw $7, %xmm4
+; SSE2-SSSE3-NEXT:    pmovmskb %xmm4, %ecx
+; SSE2-SSSE3-NEXT:    psllw $7, %xmm5
+; SSE2-SSSE3-NEXT:    pmovmskb %xmm5, %edx
+; SSE2-SSSE3-NEXT:    shll $16, %edx
+; SSE2-SSSE3-NEXT:    orl %ecx, %edx
+; SSE2-SSSE3-NEXT:    psllw $7, %xmm6
+; SSE2-SSSE3-NEXT:    pmovmskb %xmm6, %ecx
+; SSE2-SSSE3-NEXT:    psllw $7, %xmm7
+; SSE2-SSSE3-NEXT:    pmovmskb %xmm7, %esi
+; SSE2-SSSE3-NEXT:    shll $16, %esi
+; SSE2-SSSE3-NEXT:    orl %ecx, %esi
+; SSE2-SSSE3-NEXT:    shlq $32, %rsi
+; SSE2-SSSE3-NEXT:    orq %rdx, %rsi
+; SSE2-SSSE3-NEXT:    movq %rsi, %xmm0
+; SSE2-SSSE3-NEXT:    movq %rax, %xmm1
+; SSE2-SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm0[0]
+; SSE2-SSSE3-NEXT:    pcmpeqd %xmm0, %xmm0
+; SSE2-SSSE3-NEXT:    pcmpeqb %xmm1, %xmm0
 ; SSE2-SSSE3-NEXT:    pmovmskb %xmm0, %eax
-; SSE2-SSSE3-NEXT:    movl %eax, %edx
-; SSE2-SSSE3-NEXT:    shrl $8, %edx
-; SSE2-SSSE3-NEXT:    # kill: def $al killed $al killed $eax
-; SSE2-SSSE3-NEXT:    # kill: def $dl killed $dl killed $edx
+; SSE2-SSSE3-NEXT:    cmpl $65535, %eax # imm = 0xFFFF
+; SSE2-SSSE3-NEXT:    setne %al
 ; SSE2-SSSE3-NEXT:    retq
 ;
+; SSE41-LABEL: trunc_v128i8_cmp:
+; SSE41:       # %bb.0:
+; SSE41-NEXT:    psllw $7, %xmm0
+; SSE41-NEXT:    pmovmskb %xmm0, %eax
+; SSE41-NEXT:    psllw $7, %xmm1
+; SSE41-NEXT:    pmovmskb %xmm1, %ecx
+; SSE41-NEXT:    shll $16, %ecx
+; SSE41-NEXT:    orl %eax, %ecx
+; SSE41-NEXT:    psllw $7, %xmm2
+; SSE41-NEXT:    pmovmskb %xmm2, %edx
+; SSE41-NEXT:    psllw $7, %xmm3
+; SSE41-NEXT:    pmovmskb %xmm3, %eax
+; SSE41-NEXT:    shll $16, %eax
+; SSE41-NEXT:    orl %edx, %eax
+; SSE41-NEXT:    shlq $32, %rax
+; SSE41-NEXT:    orq %rcx, %rax
+; SSE41-NEXT:    psllw $7, %xmm4
+; SSE41-NEXT:    pmovmskb %xmm4, %ecx
+; SSE41-NEXT:    psllw $7, %xmm5
+; SSE41-NEXT:    pmovmskb %xmm5, %edx
+; SSE41-NEXT:    shll $16, %edx
+; SSE41-NEXT:    orl %ecx, %edx
+; SSE41-NEXT:    psllw $7, %xmm6
+; SSE41-NEXT:    pmovmskb %xmm6, %ecx
+; SSE41-NEXT:    psllw $7, %xmm7
+; SSE41-NEXT:    pmovmskb %xmm7, %esi
+; SSE41-NEXT:    shll $16, %esi
+; SSE41-NEXT:    orl %ecx, %esi
+; SSE41-NEXT:    shlq $32, %rsi
+; SSE41-NEXT:    orq %rdx, %rsi
+; SSE41-NEXT:    movq %rsi, %xmm0
+; SSE41-NEXT:    movq %rax, %xmm1
+; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm0[0]
+; SSE41-NEXT:    pcmpeqd %xmm0, %xmm0
+; SSE41-NEXT:    ptest %xmm0, %xmm1
+; SSE41-NEXT:    setae %al
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: trunc_v128i8_cmp:
+; AVX1:       # %bb.0:
+; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm4
+; AVX1-NEXT:    vpmovmskb %xmm4, %eax
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
+; AVX1-NEXT:    shll $16, %ecx
+; AVX1-NEXT:    orl %eax, %ecx
+; AVX1-NEXT:    vpsllw $7, %xmm1, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %edx
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm0
+; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %eax
+; AVX1-NEXT:    shll $16, %eax
+; AVX1-NEXT:    orl %edx, %eax
+; AVX1-NEXT:    shlq $32, %rax
+; AVX1-NEXT:    orq %rcx, %rax
+; AVX1-NEXT:    vpsllw $7, %xmm2, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
+; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm0
+; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %edx
+; AVX1-NEXT:    shll $16, %edx
+; AVX1-NEXT:    orl %ecx, %edx
+; AVX1-NEXT:    vpsllw $7, %xmm3, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %ecx
+; AVX1-NEXT:    vextractf128 $1, %ymm3, %xmm0
+; AVX1-NEXT:    vpsllw $7, %xmm0, %xmm0
+; AVX1-NEXT:    vpmovmskb %xmm0, %esi
+; AVX1-NEXT:    shll $16, %esi
+; AVX1-NEXT:    orl %ecx, %esi
+; AVX1-NEXT:    shlq $32, %rsi
+; AVX1-NEXT:    orq %rdx, %rsi
+; AVX1-NEXT:    vmovq %rsi, %xmm0
+; AVX1-NEXT:    vmovq %rax, %xmm1
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
+; AVX1-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vptest %xmm1, %xmm0
+; AVX1-NEXT:    setae %al
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: trunc_v128i8_cmp:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vpsllw $7, %ymm1, %ymm1
+; AVX2-NEXT:    vpmovmskb %ymm1, %eax
+; AVX2-NEXT:    shlq $32, %rax
+; AVX2-NEXT:    vpsllw $7, %ymm0, %ymm0
+; AVX2-NEXT:    vpmovmskb %ymm0, %ecx
+; AVX2-NEXT:    orq %rax, %rcx
+; AVX2-NEXT:    vpsllw $7, %ymm3, %ymm0
+; AVX2-NEXT:    vpmovmskb %ymm0, %eax
+; AVX2-NEXT:    shlq $32, %rax
+; AVX2-NEXT:    vpsllw $7, %ymm2, %ymm0
+; AVX2-NEXT:    vpmovmskb %ymm0, %edx
+; AVX2-NEXT:    orq %rax, %rdx
+; AVX2-NEXT:    vmovq %rdx, %xmm0
+; AVX2-NEXT:    vmovq %rcx, %xmm1
+; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
+; AVX2-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
+; AVX2-NEXT:    vptest %xmm1, %xmm0
+; AVX2-NEXT:    setae %al
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
+;
+; AVX512-LABEL: trunc_v128i8_cmp:
+; AVX512:       # %bb.0:
+; AVX512-NEXT:    vpsllw $7, %zmm0, %zmm0
+; AVX512-NEXT:    vpmovb2m %zmm0, %k0
+; AVX512-NEXT:    kmovq %k0, %rax
+; AVX512-NEXT:    vpsllw $7, %zmm1, %zmm0
+; AVX512-NEXT:    vpmovb2m %zmm0, %k0
+; AVX512-NEXT:    kmovq %k0, %rcx
+; AVX512-NEXT:    vmovq %rcx, %xmm0
+; AVX512-NEXT:    vmovq %rax, %xmm1
+; AVX512-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
+; AVX512-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
+; AVX512-NEXT:    vptest %xmm1, %xmm0
+; AVX512-NEXT:    setae %al
+; AVX512-NEXT:    vzeroupper
+; AVX512-NEXT:    retq
+  %1 = trunc <128 x i8> %a0 to <128 x i1>
+  %2 = bitcast <128 x i1> %1 to i128
+  %3 = icmp ne i128 %2, -1
+  ret i1 %3
+}
+
+define [2 x i8] @PR58546(<16 x float> %a0) {
+; SSE-LABEL: PR58546:
+; SSE:       # %bb.0:
+; SSE-NEXT:    xorps %xmm4, %xmm4
+; SSE-NEXT:    cmpunordps %xmm4, %xmm3
+; SSE-NEXT:    cmpunordps %xmm4, %xmm2
+; SSE-NEXT:    packssdw %xmm3, %xmm2
+; SSE-NEXT:    cmpunordps %xmm4, %xmm1
+; SSE-NEXT:    cmpunordps %xmm4, %xmm0
+; SSE-NEXT:    packssdw %xmm1, %xmm0
+; SSE-NEXT:    packsswb %xmm2, %xmm0
+; SSE-NEXT:    pmovmskb %xmm0, %eax
+; SSE-NEXT:    movl %eax, %edx
+; SSE-NEXT:    shrl $8, %edx
+; SSE-NEXT:    # kill: def $al killed $al killed $eax
+; SSE-NEXT:    # kill: def $dl killed $dl killed $edx
+; SSE-NEXT:    retq
+;
 ; AVX1-LABEL: PR58546:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vxorps %xmm2, %xmm2, %xmm2
@@ -724,23 +1511,23 @@ define [2 x i8] @PR58546(<16 x float> %a0) {
 }
 
 define i8 @PR59526(<8 x i32> %a, <8 x i32> %b, ptr %mask) {
-; SSE2-SSSE3-LABEL: PR59526:
-; SSE2-SSSE3:       # %bb.0:
-; SSE2-SSSE3-NEXT:    pcmpeqd %xmm3, %xmm1
-; SSE2-SSSE3-NEXT:    pcmpeqd %xmm2, %xmm0
-; SSE2-SSSE3-NEXT:    packssdw %xmm1, %xmm0
-; SSE2-SSSE3-NEXT:    movdqu (%rdi), %xmm1
-; SSE2-SSSE3-NEXT:    movdqu 16(%rdi), %xmm2
-; SSE2-SSSE3-NEXT:    pxor %xmm3, %xmm3
-; SSE2-SSSE3-NEXT:    pxor %xmm4, %xmm4
-; SSE2-SSSE3-NEXT:    pcmpgtd %xmm2, %xmm4
-; SSE2-SSSE3-NEXT:    pcmpgtd %xmm1, %xmm3
-; SSE2-SSSE3-NEXT:    packssdw %xmm4, %xmm3
-; SSE2-SSSE3-NEXT:    pand %xmm0, %xmm3
-; SSE2-SSSE3-NEXT:    pmovmskb %xmm3, %eax
-; SSE2-SSSE3-NEXT:    testl %eax, %eax
-; SSE2-SSSE3-NEXT:    setne %al
-; SSE2-SSSE3-NEXT:    retq
+; SSE-LABEL: PR59526:
+; SSE:       # %bb.0:
+; SSE-NEXT:    pcmpeqd %xmm3, %xmm1
+; SSE-NEXT:    pcmpeqd %xmm2, %xmm0
+; SSE-NEXT:    packssdw %xmm1, %xmm0
+; SSE-NEXT:    movdqu (%rdi), %xmm1
+; SSE-NEXT:    movdqu 16(%rdi), %xmm2
+; SSE-NEXT:    pxor %xmm3, %xmm3
+; SSE-NEXT:    pxor %xmm4, %xmm4
+; SSE-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE-NEXT:    pcmpgtd %xmm1, %xmm3
+; SSE-NEXT:    packssdw %xmm4, %xmm3
+; SSE-NEXT:    pand %xmm0, %xmm3
+; SSE-NEXT:    pmovmskb %xmm3, %eax
+; SSE-NEXT:    testl %eax, %eax
+; SSE-NEXT:    setne %al
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: PR59526:
 ; AVX1:       # %bb.0: