; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-- -mattr=+sse2 | FileCheck %s --check-prefix=SSE
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx  | FileCheck %s --check-prefixes=AVX,NO-AVX10_2
; RUN: llc < %s -mtriple=x86_64-- -mattr=+avx10.2-256 | FileCheck %s --check-prefixes=AVX,AVX10_2

;
; SSE
;

define i32 @test_x86_sse_comieq_ss(<4 x float> %a0, <4 x float> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse_comieq_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    comiss %xmm1, %xmm0
; SSE-NEXT:    setnp %cl
; SSE-NEXT:    sete %dl
; SSE-NEXT:    testb %cl, %dl
; SSE-NEXT:    cmovnel %esi, %eax
; SSE-NEXT:    retq
;
; NO-AVX10_2-LABEL: test_x86_sse_comieq_ss:
; NO-AVX10_2:       # %bb.0:
; NO-AVX10_2-NEXT:    movl %edi, %eax
; NO-AVX10_2-NEXT:    vcomiss %xmm1, %xmm0
; NO-AVX10_2-NEXT:    setnp %cl
; NO-AVX10_2-NEXT:    sete %dl
; NO-AVX10_2-NEXT:    testb %cl, %dl
; NO-AVX10_2-NEXT:    cmovnel %esi, %eax
; NO-AVX10_2-NEXT:    retq
;
; AVX10_2-LABEL: test_x86_sse_comieq_ss:
; AVX10_2:       # %bb.0:
; AVX10_2-NEXT:    movl %edi, %eax
; AVX10_2-NEXT:    vcomxss %xmm1, %xmm0
; AVX10_2-NEXT:    cmovel %esi, %eax
; AVX10_2-NEXT:    retq
  %call = call i32 @llvm.x86.sse.comieq.ss(<4 x float> %a0, <4 x float> %a1)
  %cmp = icmp eq i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse.comieq.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_x86_sse_comige_ss(<4 x float> %a0, <4 x float> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse_comige_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    comiss %xmm1, %xmm0
; SSE-NEXT:    cmovbl %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse_comige_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vcomiss %xmm1, %xmm0
; AVX-NEXT:    cmovbl %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse.comige.ss(<4 x float> %a0, <4 x float> %a1)
  %cmp = icmp ne i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse.comige.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_x86_sse_comigt_ss(<4 x float> %a0, <4 x float> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse_comigt_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    comiss %xmm1, %xmm0
; SSE-NEXT:    cmoval %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse_comigt_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vcomiss %xmm1, %xmm0
; AVX-NEXT:    cmoval %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse.comigt.ss(<4 x float> %a0, <4 x float> %a1)
  %cmp = icmp eq i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse.comigt.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_x86_sse_comile_ss(<4 x float> %a0, <4 x float> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse_comile_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    comiss %xmm0, %xmm1
; SSE-NEXT:    cmovbl %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse_comile_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vcomiss %xmm0, %xmm1
; AVX-NEXT:    cmovbl %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse.comile.ss(<4 x float> %a0, <4 x float> %a1)
  %cmp = icmp ne i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse.comile.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_x86_sse_comilt_ss(<4 x float> %a0, <4 x float> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse_comilt_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    comiss %xmm0, %xmm1
; SSE-NEXT:    cmoval %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse_comilt_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vcomiss %xmm0, %xmm1
; AVX-NEXT:    cmoval %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse.comilt.ss(<4 x float> %a0, <4 x float> %a1)
  %cmp = icmp eq i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse.comilt.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_x86_sse_comineq_ss(<4 x float> %a0, <4 x float> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse_comineq_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %esi, %eax
; SSE-NEXT:    comiss %xmm1, %xmm0
; SSE-NEXT:    cmovnel %edi, %eax
; SSE-NEXT:    cmovpl %edi, %eax
; SSE-NEXT:    retq
;
; NO-AVX10_2-LABEL: test_x86_sse_comineq_ss:
; NO-AVX10_2:       # %bb.0:
; NO-AVX10_2-NEXT:    movl %esi, %eax
; NO-AVX10_2-NEXT:    vcomiss %xmm1, %xmm0
; NO-AVX10_2-NEXT:    cmovnel %edi, %eax
; NO-AVX10_2-NEXT:    cmovpl %edi, %eax
; NO-AVX10_2-NEXT:    retq
;
; AVX10_2-LABEL: test_x86_sse_comineq_ss:
; AVX10_2:       # %bb.0:
; AVX10_2-NEXT:    movl %edi, %eax
; AVX10_2-NEXT:    vcomxss %xmm1, %xmm0
; AVX10_2-NEXT:    cmovel %esi, %eax
; AVX10_2-NEXT:    retq
  %call = call i32 @llvm.x86.sse.comineq.ss(<4 x float> %a0, <4 x float> %a1)
  %cmp = icmp ne i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse.comineq.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_x86_sse_ucomieq_ss(<4 x float> %a0, <4 x float> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse_ucomieq_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    ucomiss %xmm1, %xmm0
; SSE-NEXT:    setnp %cl
; SSE-NEXT:    sete %dl
; SSE-NEXT:    testb %cl, %dl
; SSE-NEXT:    cmovnel %esi, %eax
; SSE-NEXT:    retq
;
; NO-AVX10_2-LABEL: test_x86_sse_ucomieq_ss:
; NO-AVX10_2:       # %bb.0:
; NO-AVX10_2-NEXT:    movl %edi, %eax
; NO-AVX10_2-NEXT:    vucomiss %xmm1, %xmm0
; NO-AVX10_2-NEXT:    setnp %cl
; NO-AVX10_2-NEXT:    sete %dl
; NO-AVX10_2-NEXT:    testb %cl, %dl
; NO-AVX10_2-NEXT:    cmovnel %esi, %eax
; NO-AVX10_2-NEXT:    retq
;
; AVX10_2-LABEL: test_x86_sse_ucomieq_ss:
; AVX10_2:       # %bb.0:
; AVX10_2-NEXT:    movl %edi, %eax
; AVX10_2-NEXT:    vucomxss %xmm1, %xmm0
; AVX10_2-NEXT:    cmovel %esi, %eax
; AVX10_2-NEXT:    retq
  %call = call i32 @llvm.x86.sse.ucomieq.ss(<4 x float> %a0, <4 x float> %a1)
  %cmp = icmp eq i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse.ucomieq.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_x86_sse_ucomige_ss(<4 x float> %a0, <4 x float> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse_ucomige_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    ucomiss %xmm1, %xmm0
; SSE-NEXT:    cmovbl %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse_ucomige_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vucomiss %xmm1, %xmm0
; AVX-NEXT:    cmovbl %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse.ucomige.ss(<4 x float> %a0, <4 x float> %a1)
  %cmp = icmp ne i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse.ucomige.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_x86_sse_ucomigt_ss(<4 x float> %a0, <4 x float> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse_ucomigt_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    ucomiss %xmm1, %xmm0
; SSE-NEXT:    cmoval %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse_ucomigt_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vucomiss %xmm1, %xmm0
; AVX-NEXT:    cmoval %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse.ucomigt.ss(<4 x float> %a0, <4 x float> %a1)
  %cmp = icmp eq i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse.ucomigt.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_x86_sse_ucomile_ss(<4 x float> %a0, <4 x float> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse_ucomile_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    ucomiss %xmm0, %xmm1
; SSE-NEXT:    cmovbl %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse_ucomile_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vucomiss %xmm0, %xmm1
; AVX-NEXT:    cmovbl %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse.ucomile.ss(<4 x float> %a0, <4 x float> %a1)
  %cmp = icmp ne i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse.ucomile.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_x86_sse_ucomilt_ss(<4 x float> %a0, <4 x float> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse_ucomilt_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    ucomiss %xmm0, %xmm1
; SSE-NEXT:    cmoval %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse_ucomilt_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vucomiss %xmm0, %xmm1
; AVX-NEXT:    cmoval %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse.ucomilt.ss(<4 x float> %a0, <4 x float> %a1)
  %cmp = icmp eq i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse.ucomilt.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_x86_sse_ucomineq_ss(<4 x float> %a0, <4 x float> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse_ucomineq_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %esi, %eax
; SSE-NEXT:    ucomiss %xmm1, %xmm0
; SSE-NEXT:    cmovnel %edi, %eax
; SSE-NEXT:    cmovpl %edi, %eax
; SSE-NEXT:    retq
;
; NO-AVX10_2-LABEL: test_x86_sse_ucomineq_ss:
; NO-AVX10_2:       # %bb.0:
; NO-AVX10_2-NEXT:    movl %esi, %eax
; NO-AVX10_2-NEXT:    vucomiss %xmm1, %xmm0
; NO-AVX10_2-NEXT:    cmovnel %edi, %eax
; NO-AVX10_2-NEXT:    cmovpl %edi, %eax
; NO-AVX10_2-NEXT:    retq
;
; AVX10_2-LABEL: test_x86_sse_ucomineq_ss:
; AVX10_2:       # %bb.0:
; AVX10_2-NEXT:    movl %edi, %eax
; AVX10_2-NEXT:    vucomxss %xmm1, %xmm0
; AVX10_2-NEXT:    cmovel %esi, %eax
; AVX10_2-NEXT:    retq
  %call = call i32 @llvm.x86.sse.ucomineq.ss(<4 x float> %a0, <4 x float> %a1)
  %cmp = icmp ne i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse.ucomineq.ss(<4 x float>, <4 x float>) nounwind readnone

;
; SSE2
;

define i32 @test_x86_sse2_comieq_sd(<2 x double> %a0, <2 x double> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse2_comieq_sd:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    comisd %xmm1, %xmm0
; SSE-NEXT:    setnp %cl
; SSE-NEXT:    sete %dl
; SSE-NEXT:    testb %cl, %dl
; SSE-NEXT:    cmovnel %esi, %eax
; SSE-NEXT:    retq
;
; NO-AVX10_2-LABEL: test_x86_sse2_comieq_sd:
; NO-AVX10_2:       # %bb.0:
; NO-AVX10_2-NEXT:    movl %edi, %eax
; NO-AVX10_2-NEXT:    vcomisd %xmm1, %xmm0
; NO-AVX10_2-NEXT:    setnp %cl
; NO-AVX10_2-NEXT:    sete %dl
; NO-AVX10_2-NEXT:    testb %cl, %dl
; NO-AVX10_2-NEXT:    cmovnel %esi, %eax
; NO-AVX10_2-NEXT:    retq
;
; AVX10_2-LABEL: test_x86_sse2_comieq_sd:
; AVX10_2:       # %bb.0:
; AVX10_2-NEXT:    movl %edi, %eax
; AVX10_2-NEXT:    vcomxsd %xmm1, %xmm0
; AVX10_2-NEXT:    cmovel %esi, %eax
; AVX10_2-NEXT:    retq
  %call = call i32 @llvm.x86.sse2.comieq.sd(<2 x double> %a0, <2 x double> %a1) ; <i32> [#uses=1]
  %cmp = icmp eq i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse2.comieq.sd(<2 x double>, <2 x double>) nounwind readnone

define i32 @test_x86_sse2_comige_sd(<2 x double> %a0, <2 x double> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse2_comige_sd:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    comisd %xmm1, %xmm0
; SSE-NEXT:    cmovbl %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse2_comige_sd:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vcomisd %xmm1, %xmm0
; AVX-NEXT:    cmovbl %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse2.comige.sd(<2 x double> %a0, <2 x double> %a1) ; <i32> [#uses=1]
  %cmp = icmp ne i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse2.comige.sd(<2 x double>, <2 x double>) nounwind readnone

define i32 @test_x86_sse2_comigt_sd(<2 x double> %a0, <2 x double> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse2_comigt_sd:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    comisd %xmm1, %xmm0
; SSE-NEXT:    cmoval %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse2_comigt_sd:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vcomisd %xmm1, %xmm0
; AVX-NEXT:    cmoval %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse2.comigt.sd(<2 x double> %a0, <2 x double> %a1) ; <i32> [#uses=1]
  %cmp = icmp eq i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse2.comigt.sd(<2 x double>, <2 x double>) nounwind readnone

define i32 @test_x86_sse2_comile_sd(<2 x double> %a0, <2 x double> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse2_comile_sd:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    comisd %xmm0, %xmm1
; SSE-NEXT:    cmovbl %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse2_comile_sd:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vcomisd %xmm0, %xmm1
; AVX-NEXT:    cmovbl %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse2.comile.sd(<2 x double> %a0, <2 x double> %a1) ; <i32> [#uses=1]
  %cmp = icmp ne i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse2.comile.sd(<2 x double>, <2 x double>) nounwind readnone

define i32 @test_x86_sse2_comilt_sd(<2 x double> %a0, <2 x double> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse2_comilt_sd:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    comisd %xmm0, %xmm1
; SSE-NEXT:    cmoval %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse2_comilt_sd:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vcomisd %xmm0, %xmm1
; AVX-NEXT:    cmoval %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse2.comilt.sd(<2 x double> %a0, <2 x double> %a1) ; <i32> [#uses=1]
  %cmp = icmp eq i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse2.comilt.sd(<2 x double>, <2 x double>) nounwind readnone

define i32 @test_x86_sse2_comineq_sd(<2 x double> %a0, <2 x double> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse2_comineq_sd:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %esi, %eax
; SSE-NEXT:    comisd %xmm1, %xmm0
; SSE-NEXT:    cmovnel %edi, %eax
; SSE-NEXT:    cmovpl %edi, %eax
; SSE-NEXT:    retq
;
; NO-AVX10_2-LABEL: test_x86_sse2_comineq_sd:
; NO-AVX10_2:       # %bb.0:
; NO-AVX10_2-NEXT:    movl %esi, %eax
; NO-AVX10_2-NEXT:    vcomisd %xmm1, %xmm0
; NO-AVX10_2-NEXT:    cmovnel %edi, %eax
; NO-AVX10_2-NEXT:    cmovpl %edi, %eax
; NO-AVX10_2-NEXT:    retq
;
; AVX10_2-LABEL: test_x86_sse2_comineq_sd:
; AVX10_2:       # %bb.0:
; AVX10_2-NEXT:    movl %edi, %eax
; AVX10_2-NEXT:    vcomxsd %xmm1, %xmm0
; AVX10_2-NEXT:    cmovel %esi, %eax
; AVX10_2-NEXT:    retq
  %call = call i32 @llvm.x86.sse2.comineq.sd(<2 x double> %a0, <2 x double> %a1) ; <i32> [#uses=1]
  %cmp = icmp ne i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse2.comineq.sd(<2 x double>, <2 x double>) nounwind readnone

define i32 @test_x86_sse2_ucomieq_sd(<2 x double> %a0, <2 x double> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse2_ucomieq_sd:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    ucomisd %xmm1, %xmm0
; SSE-NEXT:    setnp %cl
; SSE-NEXT:    sete %dl
; SSE-NEXT:    testb %cl, %dl
; SSE-NEXT:    cmovnel %esi, %eax
; SSE-NEXT:    retq
;
; NO-AVX10_2-LABEL: test_x86_sse2_ucomieq_sd:
; NO-AVX10_2:       # %bb.0:
; NO-AVX10_2-NEXT:    movl %edi, %eax
; NO-AVX10_2-NEXT:    vucomisd %xmm1, %xmm0
; NO-AVX10_2-NEXT:    setnp %cl
; NO-AVX10_2-NEXT:    sete %dl
; NO-AVX10_2-NEXT:    testb %cl, %dl
; NO-AVX10_2-NEXT:    cmovnel %esi, %eax
; NO-AVX10_2-NEXT:    retq
;
; AVX10_2-LABEL: test_x86_sse2_ucomieq_sd:
; AVX10_2:       # %bb.0:
; AVX10_2-NEXT:    movl %edi, %eax
; AVX10_2-NEXT:    vucomxsd %xmm1, %xmm0
; AVX10_2-NEXT:    cmovel %esi, %eax
; AVX10_2-NEXT:    retq
  %call = call i32 @llvm.x86.sse2.ucomieq.sd(<2 x double> %a0, <2 x double> %a1) ; <i32> [#uses=1]
  %cmp = icmp eq i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse2.ucomieq.sd(<2 x double>, <2 x double>) nounwind readnone

define i32 @test_x86_sse2_ucomige_sd(<2 x double> %a0, <2 x double> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse2_ucomige_sd:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    ucomisd %xmm1, %xmm0
; SSE-NEXT:    cmovbl %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse2_ucomige_sd:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vucomisd %xmm1, %xmm0
; AVX-NEXT:    cmovbl %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse2.ucomige.sd(<2 x double> %a0, <2 x double> %a1) ; <i32> [#uses=1]
  %cmp = icmp ne i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse2.ucomige.sd(<2 x double>, <2 x double>) nounwind readnone

define i32 @test_x86_sse2_ucomigt_sd(<2 x double> %a0, <2 x double> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse2_ucomigt_sd:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    ucomisd %xmm1, %xmm0
; SSE-NEXT:    cmoval %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse2_ucomigt_sd:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vucomisd %xmm1, %xmm0
; AVX-NEXT:    cmoval %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse2.ucomigt.sd(<2 x double> %a0, <2 x double> %a1) ; <i32> [#uses=1]
  %cmp = icmp eq i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse2.ucomigt.sd(<2 x double>, <2 x double>) nounwind readnone

define i32 @test_x86_sse2_ucomile_sd(<2 x double> %a0, <2 x double> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse2_ucomile_sd:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    ucomisd %xmm0, %xmm1
; SSE-NEXT:    cmovbl %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse2_ucomile_sd:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vucomisd %xmm0, %xmm1
; AVX-NEXT:    cmovbl %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse2.ucomile.sd(<2 x double> %a0, <2 x double> %a1) ; <i32> [#uses=1]
  %cmp = icmp ne i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse2.ucomile.sd(<2 x double>, <2 x double>) nounwind readnone

define i32 @test_x86_sse2_ucomilt_sd(<2 x double> %a0, <2 x double> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse2_ucomilt_sd:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %edi, %eax
; SSE-NEXT:    ucomisd %xmm0, %xmm1
; SSE-NEXT:    cmoval %esi, %eax
; SSE-NEXT:    retq
;
; AVX-LABEL: test_x86_sse2_ucomilt_sd:
; AVX:       # %bb.0:
; AVX-NEXT:    movl %edi, %eax
; AVX-NEXT:    vucomisd %xmm0, %xmm1
; AVX-NEXT:    cmoval %esi, %eax
; AVX-NEXT:    retq
  %call = call i32 @llvm.x86.sse2.ucomilt.sd(<2 x double> %a0, <2 x double> %a1) ; <i32> [#uses=1]
  %cmp = icmp eq i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse2.ucomilt.sd(<2 x double>, <2 x double>) nounwind readnone

define i32 @test_x86_sse2_ucomineq_sd(<2 x double> %a0, <2 x double> %a1, i32 %a2, i32 %a3) {
; SSE-LABEL: test_x86_sse2_ucomineq_sd:
; SSE:       # %bb.0:
; SSE-NEXT:    movl %esi, %eax
; SSE-NEXT:    ucomisd %xmm1, %xmm0
; SSE-NEXT:    cmovnel %edi, %eax
; SSE-NEXT:    cmovpl %edi, %eax
; SSE-NEXT:    retq
;
; NO-AVX10_2-LABEL: test_x86_sse2_ucomineq_sd:
; NO-AVX10_2:       # %bb.0:
; NO-AVX10_2-NEXT:    movl %esi, %eax
; NO-AVX10_2-NEXT:    vucomisd %xmm1, %xmm0
; NO-AVX10_2-NEXT:    cmovnel %edi, %eax
; NO-AVX10_2-NEXT:    cmovpl %edi, %eax
; NO-AVX10_2-NEXT:    retq
;
; AVX10_2-LABEL: test_x86_sse2_ucomineq_sd:
; AVX10_2:       # %bb.0:
; AVX10_2-NEXT:    movl %edi, %eax
; AVX10_2-NEXT:    vucomxsd %xmm1, %xmm0
; AVX10_2-NEXT:    cmovel %esi, %eax
; AVX10_2-NEXT:    retq
  %call = call i32 @llvm.x86.sse2.ucomineq.sd(<2 x double> %a0, <2 x double> %a1) ; <i32> [#uses=1]
  %cmp = icmp ne i32 %call, 0
  %res = select i1 %cmp, i32 %a2, i32 %a3
  ret i32 %res
}
declare i32 @llvm.x86.sse2.ucomineq.sd(<2 x double>, <2 x double>) nounwind readnone

define void @PR38960_eq(<4 x float> %A, <4 x float> %B) {
; SSE-LABEL: PR38960_eq:
; SSE:       # %bb.0: # %entry
; SSE-NEXT:    comiss %xmm1, %xmm0
; SSE-NEXT:    setnp %al
; SSE-NEXT:    sete %cl
; SSE-NEXT:    testb %al, %cl
; SSE-NEXT:    jne foo@PLT # TAILCALL
; SSE-NEXT:  # %bb.1: # %if.end
; SSE-NEXT:    retq
;
; NO-AVX10_2-LABEL: PR38960_eq:
; NO-AVX10_2:       # %bb.0: # %entry
; NO-AVX10_2-NEXT:    vcomiss %xmm1, %xmm0
; NO-AVX10_2-NEXT:    setnp %al
; NO-AVX10_2-NEXT:    sete %cl
; NO-AVX10_2-NEXT:    testb %al, %cl
; NO-AVX10_2-NEXT:    jne foo@PLT # TAILCALL
; NO-AVX10_2-NEXT:  # %bb.1: # %if.end
; NO-AVX10_2-NEXT:    retq
;
; AVX10_2-LABEL: PR38960_eq:
; AVX10_2:       # %bb.0: # %entry
; AVX10_2-NEXT:    vcomxss %xmm1, %xmm0
; AVX10_2-NEXT:    je foo@PLT # TAILCALL
; AVX10_2-NEXT:  # %bb.1: # %if.end
; AVX10_2-NEXT:    retq
entry:
  %call = tail call i32 @llvm.x86.sse.comieq.ss(<4 x float> %A, <4 x float> %B) #3
  %cmp = icmp eq i32 %call, 0
  br i1 %cmp, label %if.end, label %if.then

if.then:
  tail call void @foo()
  br label %if.end

if.end:
  ret void
}

define void @PR38960_neq(<4 x float> %A, <4 x float> %B) {
; SSE-LABEL: PR38960_neq:
; SSE:       # %bb.0: # %entry
; SSE-NEXT:    comiss %xmm1, %xmm0
; SSE-NEXT:    setp %al
; SSE-NEXT:    setne %cl
; SSE-NEXT:    orb %al, %cl
; SSE-NEXT:    jne foo@PLT # TAILCALL
; SSE-NEXT:  # %bb.1: # %if.end
; SSE-NEXT:    retq
;
; NO-AVX10_2-LABEL: PR38960_neq:
; NO-AVX10_2:       # %bb.0: # %entry
; NO-AVX10_2-NEXT:    vcomiss %xmm1, %xmm0
; NO-AVX10_2-NEXT:    setp %al
; NO-AVX10_2-NEXT:    setne %cl
; NO-AVX10_2-NEXT:    orb %al, %cl
; NO-AVX10_2-NEXT:    jne foo@PLT # TAILCALL
; NO-AVX10_2-NEXT:  # %bb.1: # %if.end
; NO-AVX10_2-NEXT:    retq
;
; AVX10_2-LABEL: PR38960_neq:
; AVX10_2:       # %bb.0: # %entry
; AVX10_2-NEXT:    vcomxss %xmm1, %xmm0
; AVX10_2-NEXT:    jne foo@PLT # TAILCALL
; AVX10_2-NEXT:  # %bb.1: # %if.end
; AVX10_2-NEXT:    retq
entry:
  %call = tail call i32 @llvm.x86.sse.comineq.ss(<4 x float> %A, <4 x float> %B) #3
  %cmp = icmp eq i32 %call, 0
  br i1 %cmp, label %if.end, label %if.then

if.then:
  tail call void @foo()
  br label %if.end

if.end:
  ret void
}
declare void @foo()