path: root/llvm/test/CodeGen/X86/vec-strict-inttofp-128.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+sse2 -O3 | FileCheck %s --check-prefixes=SSE,SSE-32
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+sse2 -O3 | FileCheck %s --check-prefixes=SSE,SSE-64
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx -O3 | FileCheck %s --check-prefixes=AVX,AVX1,AVX-32,AVX1-32
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx -O3 | FileCheck %s --check-prefixes=AVX,AVX1,AVX-64,AVX1-64
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx512f -O3 | FileCheck %s --check-prefixes=AVX,AVX512F,AVX-32,AVX512F-32
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512f -O3 | FileCheck %s --check-prefixes=AVX,AVX512F,AVX-64,AVX512F-64
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=avx512vl -O3 | FileCheck %s --check-prefixes=AVX,AVX512VL,AVX-32,AVX512VL-32
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512vl -O3 | FileCheck %s --check-prefixes=AVX,AVX512VL,AVX-64,AVX512VL-64
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=avx512f,avx512dq -O3 | FileCheck %s --check-prefixes=AVX,AVX512DQ,AVX512DQ-32
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512f,avx512dq -O3 | FileCheck %s --check-prefixes=AVX,AVX512DQ,AVX512DQ-64
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=avx512dq,avx512vl -O3 | FileCheck %s --check-prefixes=AVX,AVX512DQVL,AVX512DQVL-32
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=avx512dq,avx512vl -O3 | FileCheck %s --check-prefixes=AVX,AVX512DQVL,AVX512DQVL-64

declare <2 x float> @llvm.experimental.constrained.sitofp.v2f32.v2i32(<2 x i32>, metadata, metadata)
declare <2 x float> @llvm.experimental.constrained.uitofp.v2f32.v2i32(<2 x i32>, metadata, metadata)
declare <2 x float> @llvm.experimental.constrained.sitofp.v2f32.v2i64(<2 x i64>, metadata, metadata)
declare <2 x float> @llvm.experimental.constrained.uitofp.v2f32.v2i64(<2 x i64>, metadata, metadata)
declare <4 x float> @llvm.experimental.constrained.sitofp.v4f32.v4i1(<4 x i1>, metadata, metadata)
declare <4 x float> @llvm.experimental.constrained.uitofp.v4f32.v4i1(<4 x i1>, metadata, metadata)
declare <4 x float> @llvm.experimental.constrained.sitofp.v4f32.v4i8(<4 x i8>, metadata, metadata)
declare <4 x float> @llvm.experimental.constrained.uitofp.v4f32.v4i8(<4 x i8>, metadata, metadata)
declare <4 x float> @llvm.experimental.constrained.sitofp.v4f32.v4i16(<4 x i16>, metadata, metadata)
declare <4 x float> @llvm.experimental.constrained.uitofp.v4f32.v4i16(<4 x i16>, metadata, metadata)
declare <4 x float> @llvm.experimental.constrained.sitofp.v4f32.v4i32(<4 x i32>, metadata, metadata)
declare <4 x float> @llvm.experimental.constrained.uitofp.v4f32.v4i32(<4 x i32>, metadata, metadata)
declare <2 x double> @llvm.experimental.constrained.sitofp.v2f64.v2i1(<2 x i1>, metadata, metadata)
declare <2 x double> @llvm.experimental.constrained.uitofp.v2f64.v2i1(<2 x i1>, metadata, metadata)
declare <2 x double> @llvm.experimental.constrained.sitofp.v2f64.v2i8(<2 x i8>, metadata, metadata)
declare <2 x double> @llvm.experimental.constrained.uitofp.v2f64.v2i8(<2 x i8>, metadata, metadata)
declare <2 x double> @llvm.experimental.constrained.sitofp.v2f64.v2i16(<2 x i16>, metadata, metadata)
declare <2 x double> @llvm.experimental.constrained.uitofp.v2f64.v2i16(<2 x i16>, metadata, metadata)
declare <2 x double> @llvm.experimental.constrained.sitofp.v2f64.v2i32(<2 x i32>, metadata, metadata)
declare <2 x double> @llvm.experimental.constrained.uitofp.v2f64.v2i32(<2 x i32>, metadata, metadata)
declare <2 x double> @llvm.experimental.constrained.sitofp.v2f64.v2i64(<2 x i64>, metadata, metadata)
declare <2 x double> @llvm.experimental.constrained.uitofp.v2f64.v2i64(<2 x i64>, metadata, metadata)

define <2 x float> @sitofp_v2i32_v2f32(<2 x i32> %x) #0 {
; SSE-LABEL: sitofp_v2i32_v2f32:
; SSE:       # %bb.0:
; SSE-NEXT:    movq {{.*#+}} xmm0 = xmm0[0],zero
; SSE-NEXT:    cvtdq2ps %xmm0, %xmm0
; SSE-NEXT:    ret{{[l|q]}}
;
; AVX-LABEL: sitofp_v2i32_v2f32:
; AVX:       # %bb.0:
; AVX-NEXT:    vmovq {{.*#+}} xmm0 = xmm0[0],zero
; AVX-NEXT:    vcvtdq2ps %xmm0, %xmm0
; AVX-NEXT:    ret{{[l|q]}}
 %result = call <2 x float> @llvm.experimental.constrained.sitofp.v2f32.v2i32(<2 x i32> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <2 x float> %result
}

define <2 x float> @uitofp_v2i32_v2f32(<2 x i32> %x) #0 {
; SSE-LABEL: uitofp_v2i32_v2f32:
; SSE:       # %bb.0:
; SSE-NEXT:    xorpd %xmm1, %xmm1
; SSE-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movapd {{.*#+}} xmm1 = [4.503599627370496E+15,4.503599627370496E+15]
; SSE-NEXT:    orpd %xmm1, %xmm0
; SSE-NEXT:    subpd %xmm1, %xmm0
; SSE-NEXT:    cvtpd2ps %xmm0, %xmm0
; SSE-NEXT:    ret{{[l|q]}}
;
; AVX1-LABEL: uitofp_v2i32_v2f32:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [4.503599627370496E+15,4.503599627370496E+15]
; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vsubpd %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vcvtpd2ps %xmm0, %xmm0
; AVX1-NEXT:    ret{{[l|q]}}
;
; AVX512F-LABEL: uitofp_v2i32_v2f32:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vmovq {{.*#+}} xmm0 = xmm0[0],zero
; AVX512F-NEXT:    vcvtudq2ps %zmm0, %zmm0
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    ret{{[l|q]}}
;
; AVX512VL-LABEL: uitofp_v2i32_v2f32:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vmovq {{.*#+}} xmm0 = xmm0[0],zero
; AVX512VL-NEXT:    vcvtudq2ps %xmm0, %xmm0
; AVX512VL-NEXT:    ret{{[l|q]}}
;
; AVX512DQ-LABEL: uitofp_v2i32_v2f32:
; AVX512DQ:       # %bb.0:
; AVX512DQ-NEXT:    vmovq {{.*#+}} xmm0 = xmm0[0],zero
; AVX512DQ-NEXT:    vcvtudq2ps %zmm0, %zmm0
; AVX512DQ-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512DQ-NEXT:    vzeroupper
; AVX512DQ-NEXT:    ret{{[l|q]}}
;
; AVX512DQVL-LABEL: uitofp_v2i32_v2f32:
; AVX512DQVL:       # %bb.0:
; AVX512DQVL-NEXT:    vmovq {{.*#+}} xmm0 = xmm0[0],zero
; AVX512DQVL-NEXT:    vcvtudq2ps %xmm0, %xmm0
; AVX512DQVL-NEXT:    ret{{[l|q]}}
 %result = call <2 x float> @llvm.experimental.constrained.uitofp.v2f32.v2i32(<2 x i32> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <2 x float> %result
}

define <2 x float> @sitofp_v2i64_v2f32(<2 x i64> %x) #0 {
; SSE-32-LABEL: sitofp_v2i64_v2f32:
; SSE-32:       # %bb.0:
; SSE-32-NEXT:    pushl %ebp
; SSE-32-NEXT:    .cfi_def_cfa_offset 8
; SSE-32-NEXT:    .cfi_offset %ebp, -8
; SSE-32-NEXT:    movl %esp, %ebp
; SSE-32-NEXT:    .cfi_def_cfa_register %ebp
; SSE-32-NEXT:    andl $-8, %esp
; SSE-32-NEXT:    subl $24, %esp
; SSE-32-NEXT:    movq %xmm0, {{[0-9]+}}(%esp)
; SSE-32-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
; SSE-32-NEXT:    movq %xmm0, {{[0-9]+}}(%esp)
; SSE-32-NEXT:    fildll {{[0-9]+}}(%esp)
; SSE-32-NEXT:    fstps (%esp)
; SSE-32-NEXT:    fildll {{[0-9]+}}(%esp)
; SSE-32-NEXT:    fstps {{[0-9]+}}(%esp)
; SSE-32-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
; SSE-32-NEXT:    movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
; SSE-32-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-32-NEXT:    movl %ebp, %esp
; SSE-32-NEXT:    popl %ebp
; SSE-32-NEXT:    .cfi_def_cfa %esp, 4
; SSE-32-NEXT:    retl
;
; SSE-64-LABEL: sitofp_v2i64_v2f32:
; SSE-64:       # %bb.0:
; SSE-64-NEXT:    movq %xmm0, %rax
; SSE-64-NEXT:    cvtsi2ss %rax, %xmm1
; SSE-64-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
; SSE-64-NEXT:    movq %xmm0, %rax
; SSE-64-NEXT:    xorps %xmm0, %xmm0
; SSE-64-NEXT:    cvtsi2ss %rax, %xmm0
; SSE-64-NEXT:    unpcklps {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; SSE-64-NEXT:    movaps %xmm1, %xmm0
; SSE-64-NEXT:    retq
;
; AVX-32-LABEL: sitofp_v2i64_v2f32:
; AVX-32:       # %bb.0:
; AVX-32-NEXT:    pushl %ebp
; AVX-32-NEXT:    .cfi_def_cfa_offset 8
; AVX-32-NEXT:    .cfi_offset %ebp, -8
; AVX-32-NEXT:    movl %esp, %ebp
; AVX-32-NEXT:    .cfi_def_cfa_register %ebp
; AVX-32-NEXT:    andl $-8, %esp
; AVX-32-NEXT:    subl $24, %esp
; AVX-32-NEXT:    vmovlps %xmm0, {{[0-9]+}}(%esp)
; AVX-32-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[2,3,0,1]
; AVX-32-NEXT:    vmovlps %xmm0, {{[0-9]+}}(%esp)
; AVX-32-NEXT:    fildll {{[0-9]+}}(%esp)
; AVX-32-NEXT:    fstps (%esp)
; AVX-32-NEXT:    fildll {{[0-9]+}}(%esp)
; AVX-32-NEXT:    fstps {{[0-9]+}}(%esp)
; AVX-32-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
; AVX-32-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0],mem[0],xmm0[2,3]
; AVX-32-NEXT:    movl %ebp, %esp
; AVX-32-NEXT:    popl %ebp
; AVX-32-NEXT:    .cfi_def_cfa %esp, 4
; AVX-32-NEXT:    retl
;
; AVX-64-LABEL: sitofp_v2i64_v2f32:
; AVX-64:       # %bb.0:
; AVX-64-NEXT:    vpextrq $1, %xmm0, %rax
; AVX-64-NEXT:    vcvtsi2ss %rax, %xmm1, %xmm1
; AVX-64-NEXT:    vmovq %xmm0, %rax
; AVX-64-NEXT:    vcvtsi2ss %rax, %xmm2, %xmm0
; AVX-64-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[2,3]
; AVX-64-NEXT:    retq
;
; AVX512DQ-32-LABEL: sitofp_v2i64_v2f32:
; AVX512DQ-32:       # %bb.0:
; AVX512DQ-32-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512DQ-32-NEXT:    vcvtqq2ps %zmm0, %ymm1
; AVX512DQ-32-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[2,3,0,1]
; AVX512DQ-32-NEXT:    vcvtqq2ps %zmm0, %ymm0
; AVX512DQ-32-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],zero,zero
; AVX512DQ-32-NEXT:    vzeroupper
; AVX512DQ-32-NEXT:    retl
;
; AVX512DQ-64-LABEL: sitofp_v2i64_v2f32:
; AVX512DQ-64:       # %bb.0:
; AVX512DQ-64-NEXT:    vpextrq $1, %xmm0, %rax
; AVX512DQ-64-NEXT:    vcvtsi2ss %rax, %xmm1, %xmm1
; AVX512DQ-64-NEXT:    vmovq %xmm0, %rax
; AVX512DQ-64-NEXT:    vcvtsi2ss %rax, %xmm2, %xmm0
; AVX512DQ-64-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[2,3]
; AVX512DQ-64-NEXT:    retq
;
; AVX512DQVL-LABEL: sitofp_v2i64_v2f32:
; AVX512DQVL:       # %bb.0:
; AVX512DQVL-NEXT:    vcvtqq2ps %xmm0, %xmm0
; AVX512DQVL-NEXT:    ret{{[l|q]}}
 %result = call <2 x float> @llvm.experimental.constrained.sitofp.v2f32.v2i64(<2 x i64> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <2 x float> %result
}

define <2 x float> @uitofp_v2i64_v2f32(<2 x i64> %x) #0 {
; SSE-32-LABEL: uitofp_v2i64_v2f32:
; SSE-32:       # %bb.0:
; SSE-32-NEXT:    pushl %ebp
; SSE-32-NEXT:    .cfi_def_cfa_offset 8
; SSE-32-NEXT:    .cfi_offset %ebp, -8
; SSE-32-NEXT:    movl %esp, %ebp
; SSE-32-NEXT:    .cfi_def_cfa_register %ebp
; SSE-32-NEXT:    andl $-8, %esp
; SSE-32-NEXT:    subl $24, %esp
; SSE-32-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
; SSE-32-NEXT:    movq %xmm1, {{[0-9]+}}(%esp)
; SSE-32-NEXT:    movq %xmm0, {{[0-9]+}}(%esp)
; SSE-32-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[3,1,2,3]
; SSE-32-NEXT:    movd %xmm1, %eax
; SSE-32-NEXT:    xorl %ecx, %ecx
; SSE-32-NEXT:    testl %eax, %eax
; SSE-32-NEXT:    setns %cl
; SSE-32-NEXT:    fildll {{[0-9]+}}(%esp)
; SSE-32-NEXT:    fadds {{\.LCPI.*}}(,%ecx,4)
; SSE-32-NEXT:    fstps (%esp)
; SSE-32-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
; SSE-32-NEXT:    movd %xmm0, %eax
; SSE-32-NEXT:    xorl %ecx, %ecx
; SSE-32-NEXT:    testl %eax, %eax
; SSE-32-NEXT:    setns %cl
; SSE-32-NEXT:    fildll {{[0-9]+}}(%esp)
; SSE-32-NEXT:    fadds {{\.LCPI.*}}(,%ecx,4)
; SSE-32-NEXT:    fstps {{[0-9]+}}(%esp)
; SSE-32-NEXT:    movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
; SSE-32-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
; SSE-32-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-32-NEXT:    movl %ebp, %esp
; SSE-32-NEXT:    popl %ebp
; SSE-32-NEXT:    .cfi_def_cfa %esp, 4
; SSE-32-NEXT:    retl
;
; SSE-64-LABEL: uitofp_v2i64_v2f32:
; SSE-64:       # %bb.0:
; SSE-64-NEXT:    movdqa %xmm0, %xmm1
; SSE-64-NEXT:    movq %xmm0, %rax
; SSE-64-NEXT:    movq %rax, %rcx
; SSE-64-NEXT:    shrq %rcx
; SSE-64-NEXT:    movl %eax, %edx
; SSE-64-NEXT:    andl $1, %edx
; SSE-64-NEXT:    orq %rcx, %rdx
; SSE-64-NEXT:    testq %rax, %rax
; SSE-64-NEXT:    cmovnsq %rax, %rdx
; SSE-64-NEXT:    xorps %xmm0, %xmm0
; SSE-64-NEXT:    cvtsi2ss %rdx, %xmm0
; SSE-64-NEXT:    jns .LBB3_2
; SSE-64-NEXT:  # %bb.1:
; SSE-64-NEXT:    addss %xmm0, %xmm0
; SSE-64-NEXT:  .LBB3_2:
; SSE-64-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
; SSE-64-NEXT:    movq %xmm1, %rax
; SSE-64-NEXT:    movq %rax, %rcx
; SSE-64-NEXT:    shrq %rcx
; SSE-64-NEXT:    movl %eax, %edx
; SSE-64-NEXT:    andl $1, %edx
; SSE-64-NEXT:    orq %rcx, %rdx
; SSE-64-NEXT:    testq %rax, %rax
; SSE-64-NEXT:    cmovnsq %rax, %rdx
; SSE-64-NEXT:    xorps %xmm1, %xmm1
; SSE-64-NEXT:    cvtsi2ss %rdx, %xmm1
; SSE-64-NEXT:    jns .LBB3_4
; SSE-64-NEXT:  # %bb.3:
; SSE-64-NEXT:    addss %xmm1, %xmm1
; SSE-64-NEXT:  .LBB3_4:
; SSE-64-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-64-NEXT:    retq
;
; AVX-32-LABEL: uitofp_v2i64_v2f32:
; AVX-32:       # %bb.0:
; AVX-32-NEXT:    pushl %ebp
; AVX-32-NEXT:    .cfi_def_cfa_offset 8
; AVX-32-NEXT:    .cfi_offset %ebp, -8
; AVX-32-NEXT:    movl %esp, %ebp
; AVX-32-NEXT:    .cfi_def_cfa_register %ebp
; AVX-32-NEXT:    andl $-8, %esp
; AVX-32-NEXT:    subl $24, %esp
; AVX-32-NEXT:    vmovlps %xmm0, {{[0-9]+}}(%esp)
; AVX-32-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[2,3,0,1]
; AVX-32-NEXT:    vmovlps %xmm1, {{[0-9]+}}(%esp)
; AVX-32-NEXT:    vextractps $1, %xmm0, %eax
; AVX-32-NEXT:    xorl %ecx, %ecx
; AVX-32-NEXT:    testl %eax, %eax
; AVX-32-NEXT:    setns %cl
; AVX-32-NEXT:    fildll {{[0-9]+}}(%esp)
; AVX-32-NEXT:    fadds {{\.LCPI.*}}(,%ecx,4)
; AVX-32-NEXT:    fstps {{[0-9]+}}(%esp)
; AVX-32-NEXT:    vextractps $3, %xmm0, %eax
; AVX-32-NEXT:    xorl %ecx, %ecx
; AVX-32-NEXT:    testl %eax, %eax
; AVX-32-NEXT:    setns %cl
; AVX-32-NEXT:    fildll {{[0-9]+}}(%esp)
; AVX-32-NEXT:    fadds {{\.LCPI.*}}(,%ecx,4)
; AVX-32-NEXT:    fstps (%esp)
; AVX-32-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
; AVX-32-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0],mem[0],xmm0[2,3]
; AVX-32-NEXT:    movl %ebp, %esp
; AVX-32-NEXT:    popl %ebp
; AVX-32-NEXT:    .cfi_def_cfa %esp, 4
; AVX-32-NEXT:    retl
;
; AVX1-64-LABEL: uitofp_v2i64_v2f32:
; AVX1-64:       # %bb.0:
; AVX1-64-NEXT:    vpextrq $1, %xmm0, %rax
; AVX1-64-NEXT:    movq %rax, %rcx
; AVX1-64-NEXT:    shrq %rcx
; AVX1-64-NEXT:    movl %eax, %edx
; AVX1-64-NEXT:    andl $1, %edx
; AVX1-64-NEXT:    orq %rcx, %rdx
; AVX1-64-NEXT:    testq %rax, %rax
; AVX1-64-NEXT:    cmovnsq %rax, %rdx
; AVX1-64-NEXT:    vcvtsi2ss %rdx, %xmm1, %xmm1
; AVX1-64-NEXT:    jns .LBB3_2
; AVX1-64-NEXT:  # %bb.1:
; AVX1-64-NEXT:    vaddss %xmm1, %xmm1, %xmm1
; AVX1-64-NEXT:  .LBB3_2:
; AVX1-64-NEXT:    vmovq %xmm0, %rax
; AVX1-64-NEXT:    movq %rax, %rcx
; AVX1-64-NEXT:    shrq %rcx
; AVX1-64-NEXT:    movl %eax, %edx
; AVX1-64-NEXT:    andl $1, %edx
; AVX1-64-NEXT:    orq %rcx, %rdx
; AVX1-64-NEXT:    testq %rax, %rax
; AVX1-64-NEXT:    cmovnsq %rax, %rdx
; AVX1-64-NEXT:    vcvtsi2ss %rdx, %xmm2, %xmm0
; AVX1-64-NEXT:    jns .LBB3_4
; AVX1-64-NEXT:  # %bb.3:
; AVX1-64-NEXT:    vaddss %xmm0, %xmm0, %xmm0
; AVX1-64-NEXT:  .LBB3_4:
; AVX1-64-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[2,3]
; AVX1-64-NEXT:    retq
;
; AVX512F-64-LABEL: uitofp_v2i64_v2f32:
; AVX512F-64:       # %bb.0:
; AVX512F-64-NEXT:    vpextrq $1, %xmm0, %rax
; AVX512F-64-NEXT:    vcvtusi2ss %rax, %xmm1, %xmm1
; AVX512F-64-NEXT:    vmovq %xmm0, %rax
; AVX512F-64-NEXT:    vcvtusi2ss %rax, %xmm2, %xmm0
; AVX512F-64-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[2,3]
; AVX512F-64-NEXT:    retq
;
; AVX512VL-64-LABEL: uitofp_v2i64_v2f32:
; AVX512VL-64:       # %bb.0:
; AVX512VL-64-NEXT:    vpextrq $1, %xmm0, %rax
; AVX512VL-64-NEXT:    vcvtusi2ss %rax, %xmm1, %xmm1
; AVX512VL-64-NEXT:    vmovq %xmm0, %rax
; AVX512VL-64-NEXT:    vcvtusi2ss %rax, %xmm2, %xmm0
; AVX512VL-64-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[2,3]
; AVX512VL-64-NEXT:    retq
;
; AVX512DQ-32-LABEL: uitofp_v2i64_v2f32:
; AVX512DQ-32:       # %bb.0:
; AVX512DQ-32-NEXT:    # kill: def $xmm0 killed $xmm0 def $zmm0
; AVX512DQ-32-NEXT:    vcvtuqq2ps %zmm0, %ymm1
; AVX512DQ-32-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[2,3,0,1]
; AVX512DQ-32-NEXT:    vcvtuqq2ps %zmm0, %ymm0
; AVX512DQ-32-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[0],xmm0[0],zero,zero
; AVX512DQ-32-NEXT:    vzeroupper
; AVX512DQ-32-NEXT:    retl
;
; AVX512DQ-64-LABEL: uitofp_v2i64_v2f32:
; AVX512DQ-64:       # %bb.0:
; AVX512DQ-64-NEXT:    vpextrq $1, %xmm0, %rax
; AVX512DQ-64-NEXT:    vcvtusi2ss %rax, %xmm1, %xmm1
; AVX512DQ-64-NEXT:    vmovq %xmm0, %rax
; AVX512DQ-64-NEXT:    vcvtusi2ss %rax, %xmm2, %xmm0
; AVX512DQ-64-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[2,3]
; AVX512DQ-64-NEXT:    retq
;
; AVX512DQVL-LABEL: uitofp_v2i64_v2f32:
; AVX512DQVL:       # %bb.0:
; AVX512DQVL-NEXT:    vcvtuqq2ps %xmm0, %xmm0
; AVX512DQVL-NEXT:    ret{{[l|q]}}
 %result = call <2 x float> @llvm.experimental.constrained.uitofp.v2f32.v2i64(<2 x i64> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <2 x float> %result
}

define <4 x float> @sitofp_v4i1_v4f32(<4 x i1> %x) #0 {
; SSE-LABEL: sitofp_v4i1_v4f32:
; SSE:       # %bb.0:
; SSE-NEXT:    pslld $31, %xmm0
; SSE-NEXT:    psrad $31, %xmm0
; SSE-NEXT:    cvtdq2ps %xmm0, %xmm0
; SSE-NEXT:    ret{{[l|q]}}
;
; AVX-LABEL: sitofp_v4i1_v4f32:
; AVX:       # %bb.0:
; AVX-NEXT:    vpslld $31, %xmm0, %xmm0
; AVX-NEXT:    vpsrad $31, %xmm0, %xmm0
; AVX-NEXT:    vcvtdq2ps %xmm0, %xmm0
; AVX-NEXT:    ret{{[l|q]}}
 %result = call <4 x float> @llvm.experimental.constrained.sitofp.v4f32.v4i1(<4 x i1> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <4 x float> %result
}

define <4 x float> @uitofp_v4i1_v4f32(<4 x i1> %x) #0 {
; SSE-32-LABEL: uitofp_v4i1_v4f32:
; SSE-32:       # %bb.0:
; SSE-32-NEXT:    andps {{\.LCPI.*}}, %xmm0
; SSE-32-NEXT:    cvtdq2ps %xmm0, %xmm0
; SSE-32-NEXT:    retl
;
; SSE-64-LABEL: uitofp_v4i1_v4f32:
; SSE-64:       # %bb.0:
; SSE-64-NEXT:    andps {{.*}}(%rip), %xmm0
; SSE-64-NEXT:    cvtdq2ps %xmm0, %xmm0
; SSE-64-NEXT:    retq
;
; AVX1-32-LABEL: uitofp_v4i1_v4f32:
; AVX1-32:       # %bb.0:
; AVX1-32-NEXT:    vandps {{\.LCPI.*}}, %xmm0, %xmm0
; AVX1-32-NEXT:    vcvtdq2ps %xmm0, %xmm0
; AVX1-32-NEXT:    retl
;
; AVX1-64-LABEL: uitofp_v4i1_v4f32:
; AVX1-64:       # %bb.0:
; AVX1-64-NEXT:    vandps {{.*}}(%rip), %xmm0, %xmm0
; AVX1-64-NEXT:    vcvtdq2ps %xmm0, %xmm0
; AVX1-64-NEXT:    retq
;
; AVX512F-LABEL: uitofp_v4i1_v4f32:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vbroadcastss {{.*#+}} xmm1 = [1,1,1,1]
; AVX512F-NEXT:    vandps %xmm1, %xmm0, %xmm0
; AVX512F-NEXT:    vcvtdq2ps %xmm0, %xmm0
; AVX512F-NEXT:    ret{{[l|q]}}
;
; AVX512VL-32-LABEL: uitofp_v4i1_v4f32:
; AVX512VL-32:       # %bb.0:
; AVX512VL-32-NEXT:    vpandd {{\.LCPI.*}}{1to4}, %xmm0, %xmm0
; AVX512VL-32-NEXT:    vcvtdq2ps %xmm0, %xmm0
; AVX512VL-32-NEXT:    retl
;
; AVX512VL-64-LABEL: uitofp_v4i1_v4f32:
; AVX512VL-64:       # %bb.0:
; AVX512VL-64-NEXT:    vpandd {{.*}}(%rip){1to4}, %xmm0, %xmm0
; AVX512VL-64-NEXT:    vcvtdq2ps %xmm0, %xmm0
; AVX512VL-64-NEXT:    retq
;
; AVX512DQ-LABEL: uitofp_v4i1_v4f32:
; AVX512DQ:       # %bb.0:
; AVX512DQ-NEXT:    vbroadcastss {{.*#+}} xmm1 = [1,1,1,1]
; AVX512DQ-NEXT:    vandps %xmm1, %xmm0, %xmm0
; AVX512DQ-NEXT:    vcvtdq2ps %xmm0, %xmm0
; AVX512DQ-NEXT:    ret{{[l|q]}}
;
; AVX512DQVL-32-LABEL: uitofp_v4i1_v4f32:
; AVX512DQVL-32:       # %bb.0:
; AVX512DQVL-32-NEXT:    vandps {{\.LCPI.*}}{1to4}, %xmm0, %xmm0
; AVX512DQVL-32-NEXT:    vcvtdq2ps %xmm0, %xmm0
; AVX512DQVL-32-NEXT:    retl
;
; AVX512DQVL-64-LABEL: uitofp_v4i1_v4f32:
; AVX512DQVL-64:       # %bb.0:
; AVX512DQVL-64-NEXT:    vandps {{.*}}(%rip){1to4}, %xmm0, %xmm0
; AVX512DQVL-64-NEXT:    vcvtdq2ps %xmm0, %xmm0
; AVX512DQVL-64-NEXT:    retq
 %result = call <4 x float> @llvm.experimental.constrained.uitofp.v4f32.v4i1(<4 x i1> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <4 x float> %result
}

define <4 x float> @sitofp_v4i8_v4f32(<4 x i8> %x) #0 {
; SSE-LABEL: sitofp_v4i8_v4f32:
; SSE:       # %bb.0:
; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
; SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
; SSE-NEXT:    psrad $24, %xmm0
; SSE-NEXT:    cvtdq2ps %xmm0, %xmm0
; SSE-NEXT:    ret{{[l|q]}}
;
; AVX-LABEL: sitofp_v4i8_v4f32:
; AVX:       # %bb.0:
; AVX-NEXT:    vpmovsxbd %xmm0, %xmm0
; AVX-NEXT:    vcvtdq2ps %xmm0, %xmm0
; AVX-NEXT:    ret{{[l|q]}}
 %result = call <4 x float> @llvm.experimental.constrained.sitofp.v4f32.v4i8(<4 x i8> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <4 x float> %result
}

define <4 x float> @uitofp_v4i8_v4f32(<4 x i8> %x) #0 {
; SSE-LABEL: uitofp_v4i8_v4f32:
; SSE:       # %bb.0:
; SSE-NEXT:    pxor %xmm1, %xmm1
; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
; SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    cvtdq2ps %xmm0, %xmm0
; SSE-NEXT:    ret{{[l|q]}}
;
; AVX-LABEL: uitofp_v4i8_v4f32:
; AVX:       # %bb.0:
; AVX-NEXT:    vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
; AVX-NEXT:    vcvtdq2ps %xmm0, %xmm0
; AVX-NEXT:    ret{{[l|q]}}
 %result = call <4 x float> @llvm.experimental.constrained.uitofp.v4f32.v4i8(<4 x i8> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <4 x float> %result
}

define <4 x float> @sitofp_v4i16_v4f32(<4 x i16> %x) #0 {
; SSE-LABEL: sitofp_v4i16_v4f32:
; SSE:       # %bb.0:
; SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
; SSE-NEXT:    psrad $16, %xmm0
; SSE-NEXT:    cvtdq2ps %xmm0, %xmm0
; SSE-NEXT:    ret{{[l|q]}}
;
; AVX-LABEL: sitofp_v4i16_v4f32:
; AVX:       # %bb.0:
; AVX-NEXT:    vpmovsxwd %xmm0, %xmm0
; AVX-NEXT:    vcvtdq2ps %xmm0, %xmm0
; AVX-NEXT:    ret{{[l|q]}}
 %result = call <4 x float> @llvm.experimental.constrained.sitofp.v4f32.v4i16(<4 x i16> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <4 x float> %result
}

define <4 x float> @uitofp_v4i16_v4f32(<4 x i16> %x) #0 {
; SSE-LABEL: uitofp_v4i16_v4f32:
; SSE:       # %bb.0:
; SSE-NEXT:    pxor %xmm1, %xmm1
; SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    cvtdq2ps %xmm0, %xmm0
; SSE-NEXT:    ret{{[l|q]}}
;
; AVX-LABEL: uitofp_v4i16_v4f32:
; AVX:       # %bb.0:
; AVX-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
; AVX-NEXT:    vcvtdq2ps %xmm0, %xmm0
; AVX-NEXT:    ret{{[l|q]}}
 %result = call <4 x float> @llvm.experimental.constrained.uitofp.v4f32.v4i16(<4 x i16> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <4 x float> %result
}

define <4 x float> @sitofp_v4i32_v4f32(<4 x i32> %x) #0 {
; SSE-LABEL: sitofp_v4i32_v4f32:
; SSE:       # %bb.0:
; SSE-NEXT:    cvtdq2ps %xmm0, %xmm0
; SSE-NEXT:    ret{{[l|q]}}
;
; AVX-LABEL: sitofp_v4i32_v4f32:
; AVX:       # %bb.0:
; AVX-NEXT:    vcvtdq2ps %xmm0, %xmm0
; AVX-NEXT:    ret{{[l|q]}}
 %result = call <4 x float> @llvm.experimental.constrained.sitofp.v4f32.v4i32(<4 x i32> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <4 x float> %result
}

define <4 x float> @uitofp_v4i32_v4f32(<4 x i32> %x) #0 {
; SSE-32-LABEL: uitofp_v4i32_v4f32:
; SSE-32:       # %bb.0:
; SSE-32-NEXT:    movdqa {{.*#+}} xmm1 = [65535,65535,65535,65535]
; SSE-32-NEXT:    pand %xmm0, %xmm1
; SSE-32-NEXT:    por {{\.LCPI.*}}, %xmm1
; SSE-32-NEXT:    psrld $16, %xmm0
; SSE-32-NEXT:    por {{\.LCPI.*}}, %xmm0
; SSE-32-NEXT:    subps {{\.LCPI.*}}, %xmm0
; SSE-32-NEXT:    addps %xmm1, %xmm0
; SSE-32-NEXT:    retl
;
; SSE-64-LABEL: uitofp_v4i32_v4f32:
; SSE-64:       # %bb.0:
; SSE-64-NEXT:    movdqa {{.*#+}} xmm1 = [65535,65535,65535,65535]
; SSE-64-NEXT:    pand %xmm0, %xmm1
; SSE-64-NEXT:    por {{.*}}(%rip), %xmm1
; SSE-64-NEXT:    psrld $16, %xmm0
; SSE-64-NEXT:    por {{.*}}(%rip), %xmm0
; SSE-64-NEXT:    subps {{.*}}(%rip), %xmm0
; SSE-64-NEXT:    addps %xmm1, %xmm0
; SSE-64-NEXT:    retq
;
; AVX1-32-LABEL: uitofp_v4i32_v4f32:
; AVX1-32:       # %bb.0:
; AVX1-32-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0],mem[1],xmm0[2],mem[3],xmm0[4],mem[5],xmm0[6],mem[7]
; AVX1-32-NEXT:    vpsrld $16, %xmm0, %xmm0
; AVX1-32-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0],mem[1],xmm0[2],mem[3],xmm0[4],mem[5],xmm0[6],mem[7]
; AVX1-32-NEXT:    vsubps {{\.LCPI.*}}, %xmm0, %xmm0
; AVX1-32-NEXT:    vaddps %xmm0, %xmm1, %xmm0
; AVX1-32-NEXT:    retl
;
; AVX1-64-LABEL: uitofp_v4i32_v4f32:
; AVX1-64:       # %bb.0:
; AVX1-64-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0],mem[1],xmm0[2],mem[3],xmm0[4],mem[5],xmm0[6],mem[7]
; AVX1-64-NEXT:    vpsrld $16, %xmm0, %xmm0
; AVX1-64-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0],mem[1],xmm0[2],mem[3],xmm0[4],mem[5],xmm0[6],mem[7]
; AVX1-64-NEXT:    vsubps {{.*}}(%rip), %xmm0, %xmm0
; AVX1-64-NEXT:    vaddps %xmm0, %xmm1, %xmm0
; AVX1-64-NEXT:    retq
;
; AVX512F-LABEL: uitofp_v4i32_v4f32:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vmovaps %xmm0, %xmm0
; AVX512F-NEXT:    vcvtudq2ps %zmm0, %zmm0
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    ret{{[l|q]}}
;
; AVX512VL-LABEL: uitofp_v4i32_v4f32:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vcvtudq2ps %xmm0, %xmm0
; AVX512VL-NEXT:    ret{{[l|q]}}
;
; AVX512DQ-LABEL: uitofp_v4i32_v4f32:
; AVX512DQ:       # %bb.0:
; AVX512DQ-NEXT:    vmovaps %xmm0, %xmm0
; AVX512DQ-NEXT:    vcvtudq2ps %zmm0, %zmm0
; AVX512DQ-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512DQ-NEXT:    vzeroupper
; AVX512DQ-NEXT:    ret{{[l|q]}}
;
; AVX512DQVL-LABEL: uitofp_v4i32_v4f32:
; AVX512DQVL:       # %bb.0:
; AVX512DQVL-NEXT:    vcvtudq2ps %xmm0, %xmm0
; AVX512DQVL-NEXT:    ret{{[l|q]}}
 %result = call <4 x float> @llvm.experimental.constrained.uitofp.v4f32.v4i32(<4 x i32> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <4 x float> %result
}

define <2 x double> @sitofp_v2i1_v2f64(<2 x i1> %x) #0 {
; SSE-LABEL: sitofp_v2i1_v2f64:
; SSE:       # %bb.0:
; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE-NEXT:    pslld $31, %xmm0
; SSE-NEXT:    psrad $31, %xmm0
; SSE-NEXT:    cvtdq2pd %xmm0, %xmm0
; SSE-NEXT:    ret{{[l|q]}}
;
; AVX-LABEL: sitofp_v2i1_v2f64:
; AVX:       # %bb.0:
; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; AVX-NEXT:    vpslld $31, %xmm0, %xmm0
; AVX-NEXT:    vpsrad $31, %xmm0, %xmm0
; AVX-NEXT:    vcvtdq2pd %xmm0, %xmm0
; AVX-NEXT:    ret{{[l|q]}}
 %result = call <2 x double> @llvm.experimental.constrained.sitofp.v2f64.v2i1(<2 x i1> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <2 x double> %result
}

define <2 x double> @uitofp_v2i1_v2f64(<2 x i1> %x) #0 {
; SSE-32-LABEL: uitofp_v2i1_v2f64:
; SSE-32:       # %bb.0:
; SSE-32-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE-32-NEXT:    pand {{\.LCPI.*}}, %xmm0
; SSE-32-NEXT:    cvtdq2pd %xmm0, %xmm0
; SSE-32-NEXT:    retl
;
; SSE-64-LABEL: uitofp_v2i1_v2f64:
; SSE-64:       # %bb.0:
; SSE-64-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; SSE-64-NEXT:    pand {{.*}}(%rip), %xmm0
; SSE-64-NEXT:    cvtdq2pd %xmm0, %xmm0
; SSE-64-NEXT:    retq
;
; AVX1-32-LABEL: uitofp_v2i1_v2f64:
; AVX1-32:       # %bb.0:
; AVX1-32-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
; AVX1-32-NEXT:    vandps {{\.LCPI.*}}, %xmm0, %xmm0
; AVX1-32-NEXT:    vcvtdq2pd %xmm0, %xmm0
; AVX1-32-NEXT:    retl
;
; AVX1-64-LABEL: uitofp_v2i1_v2f64:
; AVX1-64:       # %bb.0:
; AVX1-64-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
; AVX1-64-NEXT:    vandps {{.*}}(%rip), %xmm0, %xmm0
; AVX1-64-NEXT:    vcvtdq2pd %xmm0, %xmm0
; AVX1-64-NEXT:    retq
;
; AVX512F-LABEL: uitofp_v2i1_v2f64:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
; AVX512F-NEXT:    vbroadcastss {{.*#+}} xmm1 = [1,1,1,1]
; AVX512F-NEXT:    vandps %xmm1, %xmm0, %xmm0
; AVX512F-NEXT:    vcvtdq2pd %xmm0, %xmm0
; AVX512F-NEXT:    ret{{[l|q]}}
;
; AVX512VL-32-LABEL: uitofp_v2i1_v2f64:
; AVX512VL-32:       # %bb.0:
; AVX512VL-32-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; AVX512VL-32-NEXT:    vpandd {{\.LCPI.*}}{1to4}, %xmm0, %xmm0
; AVX512VL-32-NEXT:    vcvtdq2pd %xmm0, %xmm0
; AVX512VL-32-NEXT:    retl
;
; AVX512VL-64-LABEL: uitofp_v2i1_v2f64:
; AVX512VL-64:       # %bb.0:
; AVX512VL-64-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
; AVX512VL-64-NEXT:    vpandd {{.*}}(%rip){1to4}, %xmm0, %xmm0
; AVX512VL-64-NEXT:    vcvtdq2pd %xmm0, %xmm0
; AVX512VL-64-NEXT:    retq
;
; AVX512DQ-LABEL: uitofp_v2i1_v2f64:
; AVX512DQ:       # %bb.0:
; AVX512DQ-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
; AVX512DQ-NEXT:    vbroadcastss {{.*#+}} xmm1 = [1,1,1,1]
; AVX512DQ-NEXT:    vandps %xmm1, %xmm0, %xmm0
; AVX512DQ-NEXT:    vcvtdq2pd %xmm0, %xmm0
; AVX512DQ-NEXT:    ret{{[l|q]}}
;
; AVX512DQVL-32-LABEL: uitofp_v2i1_v2f64:
; AVX512DQVL-32:       # %bb.0:
; AVX512DQVL-32-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
; AVX512DQVL-32-NEXT:    vandps {{\.LCPI.*}}{1to4}, %xmm0, %xmm0
; AVX512DQVL-32-NEXT:    vcvtdq2pd %xmm0, %xmm0
; AVX512DQVL-32-NEXT:    retl
;
; AVX512DQVL-64-LABEL: uitofp_v2i1_v2f64:
; AVX512DQVL-64:       # %bb.0:
; AVX512DQVL-64-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
; AVX512DQVL-64-NEXT:    vandps {{.*}}(%rip){1to4}, %xmm0, %xmm0
; AVX512DQVL-64-NEXT:    vcvtdq2pd %xmm0, %xmm0
; AVX512DQVL-64-NEXT:    retq
 %result = call <2 x double> @llvm.experimental.constrained.uitofp.v2f64.v2i1(<2 x i1> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <2 x double> %result
}

define <2 x double> @sitofp_v2i8_v2f64(<2 x i8> %x) #0 {
; SSE-LABEL: sitofp_v2i8_v2f64:
; SSE:       # %bb.0:
; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
; SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
; SSE-NEXT:    psrad $24, %xmm0
; SSE-NEXT:    cvtdq2pd %xmm0, %xmm0
; SSE-NEXT:    ret{{[l|q]}}
;
; AVX-LABEL: sitofp_v2i8_v2f64:
; AVX:       # %bb.0:
; AVX-NEXT:    vpmovsxbd %xmm0, %xmm0
; AVX-NEXT:    vcvtdq2pd %xmm0, %xmm0
; AVX-NEXT:    ret{{[l|q]}}
 %result = call <2 x double> @llvm.experimental.constrained.sitofp.v2f64.v2i8(<2 x i8> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <2 x double> %result
}

define <2 x double> @uitofp_v2i8_v2f64(<2 x i8> %x) #0 {
; SSE-LABEL: uitofp_v2i8_v2f64:
; SSE:       # %bb.0:
; SSE-NEXT:    pxor %xmm1, %xmm1
; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
; SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    cvtdq2pd %xmm0, %xmm0
; SSE-NEXT:    ret{{[l|q]}}
;
; AVX-LABEL: uitofp_v2i8_v2f64:
; AVX:       # %bb.0:
; AVX-NEXT:    vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
; AVX-NEXT:    vcvtdq2pd %xmm0, %xmm0
; AVX-NEXT:    ret{{[l|q]}}
 %result = call <2 x double> @llvm.experimental.constrained.uitofp.v2f64.v2i8(<2 x i8> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <2 x double> %result
}

define <2 x double> @sitofp_v2i16_v2f64(<2 x i16> %x) #0 {
; SSE-LABEL: sitofp_v2i16_v2f64:
; SSE:       # %bb.0:
; SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
; SSE-NEXT:    psrad $16, %xmm0
; SSE-NEXT:    cvtdq2pd %xmm0, %xmm0
; SSE-NEXT:    ret{{[l|q]}}
;
; AVX-LABEL: sitofp_v2i16_v2f64:
; AVX:       # %bb.0:
; AVX-NEXT:    vpmovsxwd %xmm0, %xmm0
; AVX-NEXT:    vcvtdq2pd %xmm0, %xmm0
; AVX-NEXT:    ret{{[l|q]}}
 %result = call <2 x double> @llvm.experimental.constrained.sitofp.v2f64.v2i16(<2 x i16> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <2 x double> %result
}

define <2 x double> @uitofp_v2i16_v2f64(<2 x i16> %x) #0 {
; SSE-LABEL: uitofp_v2i16_v2f64:
; SSE:       # %bb.0:
; SSE-NEXT:    pxor %xmm1, %xmm1
; SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    cvtdq2pd %xmm0, %xmm0
; SSE-NEXT:    ret{{[l|q]}}
;
; AVX-LABEL: uitofp_v2i16_v2f64:
; AVX:       # %bb.0:
; AVX-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
; AVX-NEXT:    vcvtdq2pd %xmm0, %xmm0
; AVX-NEXT:    ret{{[l|q]}}
 %result = call <2 x double> @llvm.experimental.constrained.uitofp.v2f64.v2i16(<2 x i16> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <2 x double> %result
}

define <2 x double> @sitofp_v2i32_v2f64(<2 x i32> %x) #0 {
; SSE-LABEL: sitofp_v2i32_v2f64:
; SSE:       # %bb.0:
; SSE-NEXT:    cvtdq2pd %xmm0, %xmm0
; SSE-NEXT:    ret{{[l|q]}}
;
; AVX-LABEL: sitofp_v2i32_v2f64:
; AVX:       # %bb.0:
; AVX-NEXT:    vcvtdq2pd %xmm0, %xmm0
; AVX-NEXT:    ret{{[l|q]}}
 %result = call <2 x double> @llvm.experimental.constrained.sitofp.v2f64.v2i32(<2 x i32> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <2 x double> %result
}

define <2 x double> @uitofp_v2i32_v2f64(<2 x i32> %x) #0 {
; SSE-LABEL: uitofp_v2i32_v2f64:
; SSE:       # %bb.0:
; SSE-NEXT:    xorpd %xmm1, %xmm1
; SSE-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    movapd {{.*#+}} xmm1 = [4.503599627370496E+15,4.503599627370496E+15]
; SSE-NEXT:    orpd %xmm1, %xmm0
; SSE-NEXT:    subpd %xmm1, %xmm0
; SSE-NEXT:    ret{{[l|q]}}
;
; AVX1-LABEL: uitofp_v2i32_v2f64:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [4.503599627370496E+15,4.503599627370496E+15]
; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    vsubpd %xmm1, %xmm0, %xmm0
; AVX1-NEXT:    ret{{[l|q]}}
;
; AVX512F-LABEL: uitofp_v2i32_v2f64:
; AVX512F:       # %bb.0:
; AVX512F-NEXT:    vmovq {{.*#+}} xmm0 = xmm0[0],zero
; AVX512F-NEXT:    vcvtudq2pd %ymm0, %zmm0
; AVX512F-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512F-NEXT:    vzeroupper
; AVX512F-NEXT:    ret{{[l|q]}}
;
; AVX512VL-LABEL: uitofp_v2i32_v2f64:
; AVX512VL:       # %bb.0:
; AVX512VL-NEXT:    vcvtudq2pd %xmm0, %xmm0
; AVX512VL-NEXT:    ret{{[l|q]}}
;
; AVX512DQ-LABEL: uitofp_v2i32_v2f64:
; AVX512DQ:       # %bb.0:
; AVX512DQ-NEXT:    vmovq {{.*#+}} xmm0 = xmm0[0],zero
; AVX512DQ-NEXT:    vcvtudq2pd %ymm0, %zmm0
; AVX512DQ-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512DQ-NEXT:    vzeroupper
; AVX512DQ-NEXT:    ret{{[l|q]}}
;
; AVX512DQVL-LABEL: uitofp_v2i32_v2f64:
; AVX512DQVL:       # %bb.0:
; AVX512DQVL-NEXT:    vcvtudq2pd %xmm0, %xmm0
; AVX512DQVL-NEXT:    ret{{[l|q]}}
 %result = call <2 x double> @llvm.experimental.constrained.uitofp.v2f64.v2i32(<2 x i32> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <2 x double> %result
}

define <2 x double> @sitofp_v2i64_v2f64(<2 x i64> %x) #0 {
; SSE-32-LABEL: sitofp_v2i64_v2f64:
; SSE-32:       # %bb.0:
; SSE-32-NEXT:    pushl %ebp
; SSE-32-NEXT:    .cfi_def_cfa_offset 8
; SSE-32-NEXT:    .cfi_offset %ebp, -8
; SSE-32-NEXT:    movl %esp, %ebp
; SSE-32-NEXT:    .cfi_def_cfa_register %ebp
; SSE-32-NEXT:    andl $-8, %esp
; SSE-32-NEXT:    subl $32, %esp
; SSE-32-NEXT:    movq %xmm0, {{[0-9]+}}(%esp)
; SSE-32-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
; SSE-32-NEXT:    movq %xmm0, {{[0-9]+}}(%esp)
; SSE-32-NEXT:    fildll {{[0-9]+}}(%esp)
; SSE-32-NEXT:    fstpl {{[0-9]+}}(%esp)
; SSE-32-NEXT:    fildll {{[0-9]+}}(%esp)
; SSE-32-NEXT:    fstpl (%esp)
; SSE-32-NEXT:    movsd {{.*#+}} xmm0 = mem[0],zero
; SSE-32-NEXT:    movhps {{.*#+}} xmm0 = xmm0[0,1],mem[0,1]
; SSE-32-NEXT:    movl %ebp, %esp
; SSE-32-NEXT:    popl %ebp
; SSE-32-NEXT:    .cfi_def_cfa %esp, 4
; SSE-32-NEXT:    retl
;
; SSE-64-LABEL: sitofp_v2i64_v2f64:
; SSE-64:       # %bb.0:
; SSE-64-NEXT:    movq %xmm0, %rax
; SSE-64-NEXT:    cvtsi2sd %rax, %xmm1
; SSE-64-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
; SSE-64-NEXT:    movq %xmm0, %rax
; SSE-64-NEXT:    xorps %xmm0, %xmm0
; SSE-64-NEXT:    cvtsi2sd %rax, %xmm0
; SSE-64-NEXT:    unpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
; SSE-64-NEXT:    movapd %xmm1, %xmm0
; SSE-64-NEXT:    retq
;
; AVX-32-LABEL: sitofp_v2i64_v2f64:
; AVX-32:       # %bb.0:
; AVX-32-NEXT:    pushl %ebp
; AVX-32-NEXT:    .cfi_def_cfa_offset 8
; AVX-32-NEXT:    .cfi_offset %ebp, -8
; AVX-32-NEXT:    movl %esp, %ebp
; AVX-32-NEXT:    .cfi_def_cfa_register %ebp
; AVX-32-NEXT:    andl $-8, %esp
; AVX-32-NEXT:    subl $32, %esp
; AVX-32-NEXT:    vmovlps %xmm0, {{[0-9]+}}(%esp)
; AVX-32-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[2,3,0,1]
; AVX-32-NEXT:    vmovlps %xmm0, {{[0-9]+}}(%esp)
; AVX-32-NEXT:    fildll {{[0-9]+}}(%esp)
; AVX-32-NEXT:    fstpl {{[0-9]+}}(%esp)
; AVX-32-NEXT:    fildll {{[0-9]+}}(%esp)
; AVX-32-NEXT:    fstpl (%esp)
; AVX-32-NEXT:    vmovsd {{.*#+}} xmm0 = mem[0],zero
; AVX-32-NEXT:    vmovhps {{.*#+}} xmm0 = xmm0[0,1],mem[0,1]
; AVX-32-NEXT:    movl %ebp, %esp
; AVX-32-NEXT:    popl %ebp
; AVX-32-NEXT:    .cfi_def_cfa %esp, 4
; AVX-32-NEXT:    retl
;
; AVX-64-LABEL: sitofp_v2i64_v2f64:
; AVX-64:       # %bb.0:
; AVX-64-NEXT:    vpextrq $1, %xmm0, %rax
; AVX-64-NEXT:    vcvtsi2sd %rax, %xmm1, %xmm1
; AVX-64-NEXT:    vmovq %xmm0, %rax
; AVX-64-NEXT:    vcvtsi2sd %rax, %xmm2, %xmm0
; AVX-64-NEXT:    vunpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
; AVX-64-NEXT:    retq
;
; AVX512DQ-LABEL: sitofp_v2i64_v2f64:
; AVX512DQ:       # %bb.0:
; AVX512DQ-NEXT:    vmovaps %xmm0, %xmm0
; AVX512DQ-NEXT:    vcvtqq2pd %zmm0, %zmm0
; AVX512DQ-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512DQ-NEXT:    vzeroupper
; AVX512DQ-NEXT:    ret{{[l|q]}}
;
; AVX512DQVL-LABEL: sitofp_v2i64_v2f64:
; AVX512DQVL:       # %bb.0:
; AVX512DQVL-NEXT:    vcvtqq2pd %xmm0, %xmm0
; AVX512DQVL-NEXT:    ret{{[l|q]}}
 %result = call <2 x double> @llvm.experimental.constrained.sitofp.v2f64.v2i64(<2 x i64> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <2 x double> %result
}

define <2 x double> @uitofp_v2i64_v2f64(<2 x i64> %x) #0 {
; SSE-32-LABEL: uitofp_v2i64_v2f64:
; SSE-32:       # %bb.0:
; SSE-32-NEXT:    movdqa {{.*#+}} xmm1 = [4294967295,0,4294967295,0]
; SSE-32-NEXT:    pand %xmm0, %xmm1
; SSE-32-NEXT:    por {{\.LCPI.*}}, %xmm1
; SSE-32-NEXT:    psrlq $32, %xmm0
; SSE-32-NEXT:    por {{\.LCPI.*}}, %xmm0
; SSE-32-NEXT:    subpd {{\.LCPI.*}}, %xmm0
; SSE-32-NEXT:    addpd %xmm1, %xmm0
; SSE-32-NEXT:    retl
;
; SSE-64-LABEL: uitofp_v2i64_v2f64:
; SSE-64:       # %bb.0:
; SSE-64-NEXT:    movdqa {{.*#+}} xmm1 = [4294967295,4294967295]
; SSE-64-NEXT:    pand %xmm0, %xmm1
; SSE-64-NEXT:    por {{.*}}(%rip), %xmm1
; SSE-64-NEXT:    psrlq $32, %xmm0
; SSE-64-NEXT:    por {{.*}}(%rip), %xmm0
; SSE-64-NEXT:    subpd {{.*}}(%rip), %xmm0
; SSE-64-NEXT:    addpd %xmm1, %xmm0
; SSE-64-NEXT:    retq
;
; AVX1-32-LABEL: uitofp_v2i64_v2f64:
; AVX1-32:       # %bb.0:
; AVX1-32-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; AVX1-32-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5],xmm1[6,7]
; AVX1-32-NEXT:    vpor {{\.LCPI.*}}, %xmm1, %xmm1
; AVX1-32-NEXT:    vpsrlq $32, %xmm0, %xmm0
; AVX1-32-NEXT:    vpor {{\.LCPI.*}}, %xmm0, %xmm0
; AVX1-32-NEXT:    vsubpd {{\.LCPI.*}}, %xmm0, %xmm0
; AVX1-32-NEXT:    vaddpd %xmm0, %xmm1, %xmm0
; AVX1-32-NEXT:    retl
;
; AVX1-64-LABEL: uitofp_v2i64_v2f64:
; AVX1-64:       # %bb.0:
; AVX1-64-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; AVX1-64-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5],xmm1[6,7]
; AVX1-64-NEXT:    vpor {{.*}}(%rip), %xmm1, %xmm1
; AVX1-64-NEXT:    vpsrlq $32, %xmm0, %xmm0
; AVX1-64-NEXT:    vpor {{.*}}(%rip), %xmm0, %xmm0
; AVX1-64-NEXT:    vsubpd {{.*}}(%rip), %xmm0, %xmm0
; AVX1-64-NEXT:    vaddpd %xmm0, %xmm1, %xmm0
; AVX1-64-NEXT:    retq
;
; AVX512F-32-LABEL: uitofp_v2i64_v2f64:
; AVX512F-32:       # %bb.0:
; AVX512F-32-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; AVX512F-32-NEXT:    vpblendd {{.*#+}} xmm1 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
; AVX512F-32-NEXT:    vpor {{\.LCPI.*}}, %xmm1, %xmm1
; AVX512F-32-NEXT:    vpsrlq $32, %xmm0, %xmm0
; AVX512F-32-NEXT:    vpor {{\.LCPI.*}}, %xmm0, %xmm0
; AVX512F-32-NEXT:    vsubpd {{\.LCPI.*}}, %xmm0, %xmm0
; AVX512F-32-NEXT:    vaddpd %xmm0, %xmm1, %xmm0
; AVX512F-32-NEXT:    retl
;
; AVX512F-64-LABEL: uitofp_v2i64_v2f64:
; AVX512F-64:       # %bb.0:
; AVX512F-64-NEXT:    vpxor %xmm1, %xmm1, %xmm1
; AVX512F-64-NEXT:    vpblendd {{.*#+}} xmm1 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
; AVX512F-64-NEXT:    vpor {{.*}}(%rip), %xmm1, %xmm1
; AVX512F-64-NEXT:    vpsrlq $32, %xmm0, %xmm0
; AVX512F-64-NEXT:    vpor {{.*}}(%rip), %xmm0, %xmm0
; AVX512F-64-NEXT:    vsubpd {{.*}}(%rip), %xmm0, %xmm0
; AVX512F-64-NEXT:    vaddpd %xmm0, %xmm1, %xmm0
; AVX512F-64-NEXT:    retq
;
; AVX512VL-32-LABEL: uitofp_v2i64_v2f64:
; AVX512VL-32:       # %bb.0:
; AVX512VL-32-NEXT:    vpand {{\.LCPI.*}}, %xmm0, %xmm1
; AVX512VL-32-NEXT:    vpor {{\.LCPI.*}}, %xmm1, %xmm1
; AVX512VL-32-NEXT:    vpsrlq $32, %xmm0, %xmm0
; AVX512VL-32-NEXT:    vpor {{\.LCPI.*}}, %xmm0, %xmm0
; AVX512VL-32-NEXT:    vsubpd {{\.LCPI.*}}, %xmm0, %xmm0
; AVX512VL-32-NEXT:    vaddpd %xmm0, %xmm1, %xmm0
; AVX512VL-32-NEXT:    retl
;
; AVX512VL-64-LABEL: uitofp_v2i64_v2f64:
; AVX512VL-64:       # %bb.0:
; AVX512VL-64-NEXT:    vpand {{.*}}(%rip), %xmm0, %xmm1
; AVX512VL-64-NEXT:    vpor {{.*}}(%rip), %xmm1, %xmm1
; AVX512VL-64-NEXT:    vpsrlq $32, %xmm0, %xmm0
; AVX512VL-64-NEXT:    vpor {{.*}}(%rip), %xmm0, %xmm0
; AVX512VL-64-NEXT:    vsubpd {{.*}}(%rip), %xmm0, %xmm0
; AVX512VL-64-NEXT:    vaddpd %xmm0, %xmm1, %xmm0
; AVX512VL-64-NEXT:    retq
;
; AVX512DQ-LABEL: uitofp_v2i64_v2f64:
; AVX512DQ:       # %bb.0:
; AVX512DQ-NEXT:    vmovaps %xmm0, %xmm0
; AVX512DQ-NEXT:    vcvtuqq2pd %zmm0, %zmm0
; AVX512DQ-NEXT:    # kill: def $xmm0 killed $xmm0 killed $zmm0
; AVX512DQ-NEXT:    vzeroupper
; AVX512DQ-NEXT:    ret{{[l|q]}}
;
; AVX512DQVL-LABEL: uitofp_v2i64_v2f64:
; AVX512DQVL:       # %bb.0:
; AVX512DQVL-NEXT:    vcvtuqq2pd %xmm0, %xmm0
; AVX512DQVL-NEXT:    ret{{[l|q]}}
 %result = call <2 x double> @llvm.experimental.constrained.uitofp.v2f64.v2i64(<2 x i64> %x,
                                                              metadata !"round.dynamic",
                                                              metadata !"fpexcept.strict") #0
  ret <2 x double> %result
}

attributes #0 = { strictfp }