test/CodeGen/X86/sse-intrinsics-fast-isel.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -show-mc-encoding -fast-isel-sink-local-values < %s -fast-isel -mtriple=i386-unknown-unknown -mattr=+sse | FileCheck %s --check-prefixes=CHECK,X86,SSE,X86-SSE
; RUN: llc -show-mc-encoding -fast-isel-sink-local-values < %s -fast-isel -mtriple=i386-unknown-unknown -mattr=+avx | FileCheck %s --check-prefixes=CHECK,X86,AVX,X86-AVX,AVX1,X86-AVX1
; RUN: llc -show-mc-encoding -fast-isel-sink-local-values < %s -fast-isel -mtriple=i386-unknown-unknown -mattr=+avx512f,+avx512bw,+avx512dq,+avx512vl | FileCheck %s --check-prefixes=CHECK,X86,AVX,X86-AVX,AVX512,X86-AVX512
; RUN: llc -show-mc-encoding -fast-isel-sink-local-values < %s -fast-isel -mtriple=x86_64-unknown-unknown -mattr=+sse,-sse2 | FileCheck %s --check-prefixes=CHECK,X64,SSE,X64-SSE
; RUN: llc -show-mc-encoding -fast-isel-sink-local-values < %s -fast-isel -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefixes=CHECK,X64,AVX,X64-AVX,AVX1,X64-AVX1
; RUN: llc -show-mc-encoding -fast-isel-sink-local-values < %s -fast-isel -mtriple=x86_64-unknown-unknown -mattr=+avx512f,+avx512bw,+avx512dq,+avx512vl | FileCheck %s --check-prefixes=CHECK,X64,AVX,X64-AVX,AVX512,X64-AVX512

; NOTE: This should use IR equivalent to what is generated by clang/test/CodeGen/sse-builtins.c

define <4 x float> @test_mm_add_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_add_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    addps %xmm1, %xmm0 # encoding: [0x0f,0x58,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_add_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vaddps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x58,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_add_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vaddps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x58,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = fadd <4 x float> %a0, %a1
  ret <4 x float> %res
}

define <4 x float> @test_mm_add_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_add_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    addss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0x58,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_add_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vaddss %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0x58,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_add_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vaddss %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x58,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %ext0 = extractelement <4 x float> %a0, i32 0
  %ext1 = extractelement <4 x float> %a1, i32 0
  %fadd = fadd float %ext0, %ext1
  %res = insertelement <4 x float> %a0, float %fadd, i32 0
  ret <4 x float> %res
}

define <4 x float> @test_mm_and_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_and_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    andps %xmm1, %xmm0 # encoding: [0x0f,0x54,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_and_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vandps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x54,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_and_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vandps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x54,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %arg0 = bitcast <4 x float> %a0 to <4 x i32>
  %arg1 = bitcast <4 x float> %a1 to <4 x i32>
  %res = and <4 x i32> %arg0, %arg1
  %bc = bitcast <4 x i32> %res to <4 x float>
  ret <4 x float> %bc
}

define <4 x float> @test_mm_andnot_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_andnot_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    andnps %xmm1, %xmm0 # encoding: [0x0f,0x55,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_andnot_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm2, %xmm2 # encoding: [0xc5,0xe9,0x76,0xd2]
; AVX1-NEXT:    vpxor %xmm2, %xmm0, %xmm0 # encoding: [0xc5,0xf9,0xef,0xc2]
; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf9,0xdb,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_andnot_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vpternlogq $15, %xmm0, %xmm0, %xmm0 # encoding: [0x62,0xf3,0xfd,0x08,0x25,0xc0,0x0f]
; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0xdb,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %arg0 = bitcast <4 x float> %a0 to <4 x i32>
  %arg1 = bitcast <4 x float> %a1 to <4 x i32>
  %not = xor <4 x i32> %arg0, <i32 -1, i32 -1, i32 -1, i32 -1>
  %res = and <4 x i32> %not, %arg1
  %bc = bitcast <4 x i32> %res to <4 x float>
  ret <4 x float> %bc
}

define <4 x float> @test_mm_cmpeq_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpeq_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpeqps %xmm1, %xmm0 # encoding: [0x0f,0xc2,0xc1,0x00]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cmpeq_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcmpeqps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0xc2,0xc1,0x00]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cmpeq_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcmpeqps %xmm1, %xmm0, %k0 # encoding: [0x62,0xf1,0x7c,0x08,0xc2,0xc1,0x00]
; AVX512-NEXT:    vpmovm2d %k0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = fcmp oeq <4 x float> %a0, %a1
  %sext = sext <4 x i1> %cmp to <4 x i32>
  %res = bitcast <4 x i32> %sext to <4 x float>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpeq_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpeq_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpeqss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0xc2,0xc1,0x00]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_cmpeq_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    vcmpeqss %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0xc2,0xc1,0x00]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.sse.cmp.ss(<4 x float> %a0, <4 x float> %a1, i8 0)
  ret <4 x float> %res
}
declare <4 x float> @llvm.x86.sse.cmp.ss(<4 x float>, <4 x float>, i8) nounwind readnone

define <4 x float> @test_mm_cmpge_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpge_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpleps %xmm0, %xmm1 # encoding: [0x0f,0xc2,0xc8,0x02]
; SSE-NEXT:    movaps %xmm1, %xmm0 # encoding: [0x0f,0x28,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cmpge_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcmpleps %xmm0, %xmm1, %xmm0 # encoding: [0xc5,0xf0,0xc2,0xc0,0x02]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cmpge_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcmpleps %xmm0, %xmm1, %k0 # encoding: [0x62,0xf1,0x74,0x08,0xc2,0xc0,0x02]
; AVX512-NEXT:    vpmovm2d %k0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = fcmp ole <4 x float> %a1, %a0
  %sext = sext <4 x i1> %cmp to <4 x i32>
  %res = bitcast <4 x i32> %sext to <4 x float>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpge_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpge_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpless %xmm0, %xmm1 # encoding: [0xf3,0x0f,0xc2,0xc8,0x02]
; SSE-NEXT:    movss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0x10,0xc1]
; SSE-NEXT:    # xmm0 = xmm1[0],xmm0[1,2,3]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_cmpge_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    vcmpless %xmm0, %xmm1, %xmm1 # encoding: [0xc5,0xf2,0xc2,0xc8,0x02]
; AVX-NEXT:    vblendps $1, %xmm1, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x0c,0xc1,0x01]
; AVX-NEXT:    # xmm0 = xmm1[0],xmm0[1,2,3]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = call <4 x float> @llvm.x86.sse.cmp.ss(<4 x float> %a1, <4 x float> %a0, i8 2)
  %res = shufflevector <4 x float> %a0, <4 x float> %cmp, <4 x i32> <i32 4, i32 1, i32 2, i32 3>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpgt_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpgt_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpltps %xmm0, %xmm1 # encoding: [0x0f,0xc2,0xc8,0x01]
; SSE-NEXT:    movaps %xmm1, %xmm0 # encoding: [0x0f,0x28,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cmpgt_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcmpltps %xmm0, %xmm1, %xmm0 # encoding: [0xc5,0xf0,0xc2,0xc0,0x01]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cmpgt_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcmpltps %xmm0, %xmm1, %k0 # encoding: [0x62,0xf1,0x74,0x08,0xc2,0xc0,0x01]
; AVX512-NEXT:    vpmovm2d %k0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = fcmp olt <4 x float> %a1, %a0
  %sext = sext <4 x i1> %cmp to <4 x i32>
  %res = bitcast <4 x i32> %sext to <4 x float>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpgt_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpgt_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpltss %xmm0, %xmm1 # encoding: [0xf3,0x0f,0xc2,0xc8,0x01]
; SSE-NEXT:    movss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0x10,0xc1]
; SSE-NEXT:    # xmm0 = xmm1[0],xmm0[1,2,3]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_cmpgt_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    vcmpltss %xmm0, %xmm1, %xmm1 # encoding: [0xc5,0xf2,0xc2,0xc8,0x01]
; AVX-NEXT:    vblendps $1, %xmm1, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x0c,0xc1,0x01]
; AVX-NEXT:    # xmm0 = xmm1[0],xmm0[1,2,3]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = call <4 x float> @llvm.x86.sse.cmp.ss(<4 x float> %a1, <4 x float> %a0, i8 1)
  %res = shufflevector <4 x float> %a0, <4 x float> %cmp, <4 x i32> <i32 4, i32 1, i32 2, i32 3>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmple_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmple_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpleps %xmm1, %xmm0 # encoding: [0x0f,0xc2,0xc1,0x02]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cmple_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcmpleps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0xc2,0xc1,0x02]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cmple_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcmpleps %xmm1, %xmm0, %k0 # encoding: [0x62,0xf1,0x7c,0x08,0xc2,0xc1,0x02]
; AVX512-NEXT:    vpmovm2d %k0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = fcmp ole <4 x float> %a0, %a1
  %sext = sext <4 x i1> %cmp to <4 x i32>
  %res = bitcast <4 x i32> %sext to <4 x float>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmple_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmple_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpless %xmm1, %xmm0 # encoding: [0xf3,0x0f,0xc2,0xc1,0x02]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_cmple_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    vcmpless %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0xc2,0xc1,0x02]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.sse.cmp.ss(<4 x float> %a0, <4 x float> %a1, i8 2)
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmplt_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmplt_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpltps %xmm1, %xmm0 # encoding: [0x0f,0xc2,0xc1,0x01]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cmplt_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcmpltps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0xc2,0xc1,0x01]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cmplt_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcmpltps %xmm1, %xmm0, %k0 # encoding: [0x62,0xf1,0x7c,0x08,0xc2,0xc1,0x01]
; AVX512-NEXT:    vpmovm2d %k0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = fcmp olt <4 x float> %a0, %a1
  %sext = sext <4 x i1> %cmp to <4 x i32>
  %res = bitcast <4 x i32> %sext to <4 x float>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmplt_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmplt_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpltss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0xc2,0xc1,0x01]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_cmplt_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    vcmpltss %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0xc2,0xc1,0x01]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.sse.cmp.ss(<4 x float> %a0, <4 x float> %a1, i8 1)
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpneq_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpneq_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpneqps %xmm1, %xmm0 # encoding: [0x0f,0xc2,0xc1,0x04]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cmpneq_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcmpneqps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0xc2,0xc1,0x04]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cmpneq_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcmpneqps %xmm1, %xmm0, %k0 # encoding: [0x62,0xf1,0x7c,0x08,0xc2,0xc1,0x04]
; AVX512-NEXT:    vpmovm2d %k0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = fcmp une <4 x float> %a0, %a1
  %sext = sext <4 x i1> %cmp to <4 x i32>
  %res = bitcast <4 x i32> %sext to <4 x float>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpneq_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpneq_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpneqss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0xc2,0xc1,0x04]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_cmpneq_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    vcmpneqss %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0xc2,0xc1,0x04]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.sse.cmp.ss(<4 x float> %a0, <4 x float> %a1, i8 4)
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpnge_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpnge_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpnleps %xmm0, %xmm1 # encoding: [0x0f,0xc2,0xc8,0x06]
; SSE-NEXT:    movaps %xmm1, %xmm0 # encoding: [0x0f,0x28,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cmpnge_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcmpnleps %xmm0, %xmm1, %xmm0 # encoding: [0xc5,0xf0,0xc2,0xc0,0x06]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cmpnge_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcmpnleps %xmm0, %xmm1, %k0 # encoding: [0x62,0xf1,0x74,0x08,0xc2,0xc0,0x06]
; AVX512-NEXT:    vpmovm2d %k0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = fcmp ugt <4 x float> %a1, %a0
  %sext = sext <4 x i1> %cmp to <4 x i32>
  %res = bitcast <4 x i32> %sext to <4 x float>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpnge_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpnge_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpnless %xmm0, %xmm1 # encoding: [0xf3,0x0f,0xc2,0xc8,0x06]
; SSE-NEXT:    movss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0x10,0xc1]
; SSE-NEXT:    # xmm0 = xmm1[0],xmm0[1,2,3]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_cmpnge_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    vcmpnless %xmm0, %xmm1, %xmm1 # encoding: [0xc5,0xf2,0xc2,0xc8,0x06]
; AVX-NEXT:    vblendps $1, %xmm1, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x0c,0xc1,0x01]
; AVX-NEXT:    # xmm0 = xmm1[0],xmm0[1,2,3]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = call <4 x float> @llvm.x86.sse.cmp.ss(<4 x float> %a1, <4 x float> %a0, i8 6)
  %res = shufflevector <4 x float> %a0, <4 x float> %cmp, <4 x i32> <i32 4, i32 1, i32 2, i32 3>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpngt_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpngt_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpnltps %xmm0, %xmm1 # encoding: [0x0f,0xc2,0xc8,0x05]
; SSE-NEXT:    movaps %xmm1, %xmm0 # encoding: [0x0f,0x28,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cmpngt_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcmpnltps %xmm0, %xmm1, %xmm0 # encoding: [0xc5,0xf0,0xc2,0xc0,0x05]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cmpngt_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcmpnltps %xmm0, %xmm1, %k0 # encoding: [0x62,0xf1,0x74,0x08,0xc2,0xc0,0x05]
; AVX512-NEXT:    vpmovm2d %k0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = fcmp uge <4 x float> %a1, %a0
  %sext = sext <4 x i1> %cmp to <4 x i32>
  %res = bitcast <4 x i32> %sext to <4 x float>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpngt_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpngt_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpnltss %xmm0, %xmm1 # encoding: [0xf3,0x0f,0xc2,0xc8,0x05]
; SSE-NEXT:    movss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0x10,0xc1]
; SSE-NEXT:    # xmm0 = xmm1[0],xmm0[1,2,3]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_cmpngt_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    vcmpnltss %xmm0, %xmm1, %xmm1 # encoding: [0xc5,0xf2,0xc2,0xc8,0x05]
; AVX-NEXT:    vblendps $1, %xmm1, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x0c,0xc1,0x01]
; AVX-NEXT:    # xmm0 = xmm1[0],xmm0[1,2,3]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = call <4 x float> @llvm.x86.sse.cmp.ss(<4 x float> %a1, <4 x float> %a0, i8 5)
  %res = shufflevector <4 x float> %a0, <4 x float> %cmp, <4 x i32> <i32 4, i32 1, i32 2, i32 3>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpnle_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpnle_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpnleps %xmm1, %xmm0 # encoding: [0x0f,0xc2,0xc1,0x06]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cmpnle_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcmpnleps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0xc2,0xc1,0x06]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cmpnle_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcmpnleps %xmm1, %xmm0, %k0 # encoding: [0x62,0xf1,0x7c,0x08,0xc2,0xc1,0x06]
; AVX512-NEXT:    vpmovm2d %k0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = fcmp ugt <4 x float> %a0, %a1
  %sext = sext <4 x i1> %cmp to <4 x i32>
  %res = bitcast <4 x i32> %sext to <4 x float>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpnle_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpnle_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpnless %xmm1, %xmm0 # encoding: [0xf3,0x0f,0xc2,0xc1,0x06]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_cmpnle_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    vcmpnless %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0xc2,0xc1,0x06]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.sse.cmp.ss(<4 x float> %a0, <4 x float> %a1, i8 6)
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpnlt_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpnlt_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpnltps %xmm1, %xmm0 # encoding: [0x0f,0xc2,0xc1,0x05]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cmpnlt_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcmpnltps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0xc2,0xc1,0x05]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cmpnlt_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcmpnltps %xmm1, %xmm0, %k0 # encoding: [0x62,0xf1,0x7c,0x08,0xc2,0xc1,0x05]
; AVX512-NEXT:    vpmovm2d %k0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = fcmp uge <4 x float> %a0, %a1
  %sext = sext <4 x i1> %cmp to <4 x i32>
  %res = bitcast <4 x i32> %sext to <4 x float>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpnlt_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpnlt_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpnltss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0xc2,0xc1,0x05]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_cmpnlt_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    vcmpnltss %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0xc2,0xc1,0x05]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.sse.cmp.ss(<4 x float> %a0, <4 x float> %a1, i8 5)
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpord_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpord_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpordps %xmm1, %xmm0 # encoding: [0x0f,0xc2,0xc1,0x07]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cmpord_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcmpordps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0xc2,0xc1,0x07]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cmpord_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcmpordps %xmm1, %xmm0, %k0 # encoding: [0x62,0xf1,0x7c,0x08,0xc2,0xc1,0x07]
; AVX512-NEXT:    vpmovm2d %k0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = fcmp ord <4 x float> %a0, %a1
  %sext = sext <4 x i1> %cmp to <4 x i32>
  %res = bitcast <4 x i32> %sext to <4 x float>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpord_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpord_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpordss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0xc2,0xc1,0x07]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_cmpord_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    vcmpordss %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0xc2,0xc1,0x07]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.sse.cmp.ss(<4 x float> %a0, <4 x float> %a1, i8 7)
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpunord_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpunord_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpunordps %xmm1, %xmm0 # encoding: [0x0f,0xc2,0xc1,0x03]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cmpunord_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcmpunordps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0xc2,0xc1,0x03]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cmpunord_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcmpunordps %xmm1, %xmm0, %k0 # encoding: [0x62,0xf1,0x7c,0x08,0xc2,0xc1,0x03]
; AVX512-NEXT:    vpmovm2d %k0, %xmm0 # encoding: [0x62,0xf2,0x7e,0x08,0x38,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %cmp = fcmp uno <4 x float> %a0, %a1
  %sext = sext <4 x i1> %cmp to <4 x i32>
  %res = bitcast <4 x i32> %sext to <4 x float>
  ret <4 x float> %res
}

define <4 x float> @test_mm_cmpunord_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_cmpunord_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    cmpunordss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0xc2,0xc1,0x03]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_cmpunord_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    vcmpunordss %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0xc2,0xc1,0x03]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.sse.cmp.ss(<4 x float> %a0, <4 x float> %a1, i8 3)
  ret <4 x float> %res
}

define i32 @test_mm_comieq_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_comieq_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    comiss %xmm1, %xmm0 # encoding: [0x0f,0x2f,0xc1]
; SSE-NEXT:    setnp %al # encoding: [0x0f,0x9b,0xc0]
; SSE-NEXT:    sete %cl # encoding: [0x0f,0x94,0xc1]
; SSE-NEXT:    andb %al, %cl # encoding: [0x20,0xc1]
; SSE-NEXT:    movzbl %cl, %eax # encoding: [0x0f,0xb6,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_comieq_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcomiss %xmm1, %xmm0 # encoding: [0xc5,0xf8,0x2f,0xc1]
; AVX1-NEXT:    setnp %al # encoding: [0x0f,0x9b,0xc0]
; AVX1-NEXT:    sete %cl # encoding: [0x0f,0x94,0xc1]
; AVX1-NEXT:    andb %al, %cl # encoding: [0x20,0xc1]
; AVX1-NEXT:    movzbl %cl, %eax # encoding: [0x0f,0xb6,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_comieq_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcomiss %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x2f,0xc1]
; AVX512-NEXT:    setnp %al # encoding: [0x0f,0x9b,0xc0]
; AVX512-NEXT:    sete %cl # encoding: [0x0f,0x94,0xc1]
; AVX512-NEXT:    andb %al, %cl # encoding: [0x20,0xc1]
; AVX512-NEXT:    movzbl %cl, %eax # encoding: [0x0f,0xb6,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.comieq.ss(<4 x float> %a0, <4 x float> %a1)
  ret i32 %res
}
declare i32 @llvm.x86.sse.comieq.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_mm_comige_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_comige_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; SSE-NEXT:    comiss %xmm1, %xmm0 # encoding: [0x0f,0x2f,0xc1]
; SSE-NEXT:    setae %al # encoding: [0x0f,0x93,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_comige_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX1-NEXT:    vcomiss %xmm1, %xmm0 # encoding: [0xc5,0xf8,0x2f,0xc1]
; AVX1-NEXT:    setae %al # encoding: [0x0f,0x93,0xc0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_comige_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX512-NEXT:    vcomiss %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x2f,0xc1]
; AVX512-NEXT:    setae %al # encoding: [0x0f,0x93,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.comige.ss(<4 x float> %a0, <4 x float> %a1)
  ret i32 %res
}
declare i32 @llvm.x86.sse.comige.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_mm_comigt_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_comigt_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; SSE-NEXT:    comiss %xmm1, %xmm0 # encoding: [0x0f,0x2f,0xc1]
; SSE-NEXT:    seta %al # encoding: [0x0f,0x97,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_comigt_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX1-NEXT:    vcomiss %xmm1, %xmm0 # encoding: [0xc5,0xf8,0x2f,0xc1]
; AVX1-NEXT:    seta %al # encoding: [0x0f,0x97,0xc0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_comigt_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX512-NEXT:    vcomiss %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x2f,0xc1]
; AVX512-NEXT:    seta %al # encoding: [0x0f,0x97,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.comigt.ss(<4 x float> %a0, <4 x float> %a1)
  ret i32 %res
}
declare i32 @llvm.x86.sse.comigt.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_mm_comile_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_comile_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; SSE-NEXT:    comiss %xmm0, %xmm1 # encoding: [0x0f,0x2f,0xc8]
; SSE-NEXT:    setae %al # encoding: [0x0f,0x93,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_comile_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX1-NEXT:    vcomiss %xmm0, %xmm1 # encoding: [0xc5,0xf8,0x2f,0xc8]
; AVX1-NEXT:    setae %al # encoding: [0x0f,0x93,0xc0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_comile_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX512-NEXT:    vcomiss %xmm0, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x2f,0xc8]
; AVX512-NEXT:    setae %al # encoding: [0x0f,0x93,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.comile.ss(<4 x float> %a0, <4 x float> %a1)
  ret i32 %res
}
declare i32 @llvm.x86.sse.comile.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_mm_comilt_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_comilt_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; SSE-NEXT:    comiss %xmm0, %xmm1 # encoding: [0x0f,0x2f,0xc8]
; SSE-NEXT:    seta %al # encoding: [0x0f,0x97,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_comilt_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX1-NEXT:    vcomiss %xmm0, %xmm1 # encoding: [0xc5,0xf8,0x2f,0xc8]
; AVX1-NEXT:    seta %al # encoding: [0x0f,0x97,0xc0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_comilt_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX512-NEXT:    vcomiss %xmm0, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x2f,0xc8]
; AVX512-NEXT:    seta %al # encoding: [0x0f,0x97,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.comilt.ss(<4 x float> %a0, <4 x float> %a1)
  ret i32 %res
}
declare i32 @llvm.x86.sse.comilt.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_mm_comineq_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_comineq_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    comiss %xmm1, %xmm0 # encoding: [0x0f,0x2f,0xc1]
; SSE-NEXT:    setp %al # encoding: [0x0f,0x9a,0xc0]
; SSE-NEXT:    setne %cl # encoding: [0x0f,0x95,0xc1]
; SSE-NEXT:    orb %al, %cl # encoding: [0x08,0xc1]
; SSE-NEXT:    movzbl %cl, %eax # encoding: [0x0f,0xb6,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_comineq_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcomiss %xmm1, %xmm0 # encoding: [0xc5,0xf8,0x2f,0xc1]
; AVX1-NEXT:    setp %al # encoding: [0x0f,0x9a,0xc0]
; AVX1-NEXT:    setne %cl # encoding: [0x0f,0x95,0xc1]
; AVX1-NEXT:    orb %al, %cl # encoding: [0x08,0xc1]
; AVX1-NEXT:    movzbl %cl, %eax # encoding: [0x0f,0xb6,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_comineq_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcomiss %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x2f,0xc1]
; AVX512-NEXT:    setp %al # encoding: [0x0f,0x9a,0xc0]
; AVX512-NEXT:    setne %cl # encoding: [0x0f,0x95,0xc1]
; AVX512-NEXT:    orb %al, %cl # encoding: [0x08,0xc1]
; AVX512-NEXT:    movzbl %cl, %eax # encoding: [0x0f,0xb6,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.comineq.ss(<4 x float> %a0, <4 x float> %a1)
  ret i32 %res
}
declare i32 @llvm.x86.sse.comineq.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_mm_cvt_ss2si(<4 x float> %a0) nounwind {
; SSE-LABEL: test_mm_cvt_ss2si:
; SSE:       # %bb.0:
; SSE-NEXT:    cvtss2si %xmm0, %eax # encoding: [0xf3,0x0f,0x2d,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cvt_ss2si:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcvtss2si %xmm0, %eax # encoding: [0xc5,0xfa,0x2d,0xc0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cvt_ss2si:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcvtss2si %xmm0, %eax # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x2d,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.cvtss2si(<4 x float> %a0)
  ret i32 %res
}
declare i32 @llvm.x86.sse.cvtss2si(<4 x float>) nounwind readnone

define <4 x float> @test_mm_cvtsi32_ss(<4 x float> %a0, i32 %a1) nounwind {
; X86-SSE-LABEL: test_mm_cvtsi32_ss:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    cvtsi2ssl {{[0-9]+}}(%esp), %xmm0 # encoding: [0xf3,0x0f,0x2a,0x44,0x24,0x04]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_cvtsi32_ss:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    vcvtsi2ssl {{[0-9]+}}(%esp), %xmm0, %xmm0 # encoding: [0xc5,0xfa,0x2a,0x44,0x24,0x04]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_cvtsi32_ss:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    vcvtsi2ssl {{[0-9]+}}(%esp), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x2a,0x44,0x24,0x04]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_cvtsi32_ss:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    cvtsi2ssl %edi, %xmm0 # encoding: [0xf3,0x0f,0x2a,0xc7]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_cvtsi32_ss:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vcvtsi2ssl %edi, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0x2a,0xc7]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_cvtsi32_ss:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vcvtsi2ssl %edi, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x2a,0xc7]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.sse.cvtsi2ss(<4 x float> %a0, i32 %a1)
  ret <4 x float> %res
}
declare <4 x float> @llvm.x86.sse.cvtsi2ss(<4 x float>, i32) nounwind readnone

define float @test_mm_cvtss_f32(<4 x float> %a0) nounwind {
; X86-SSE-LABEL: test_mm_cvtss_f32:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    pushl %eax # encoding: [0x50]
; X86-SSE-NEXT:    movss %xmm0, (%esp) # encoding: [0xf3,0x0f,0x11,0x04,0x24]
; X86-SSE-NEXT:    flds (%esp) # encoding: [0xd9,0x04,0x24]
; X86-SSE-NEXT:    popl %eax # encoding: [0x58]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_cvtss_f32:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    pushl %eax # encoding: [0x50]
; X86-AVX1-NEXT:    vmovss %xmm0, (%esp) # encoding: [0xc5,0xfa,0x11,0x04,0x24]
; X86-AVX1-NEXT:    flds (%esp) # encoding: [0xd9,0x04,0x24]
; X86-AVX1-NEXT:    popl %eax # encoding: [0x58]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_cvtss_f32:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    pushl %eax # encoding: [0x50]
; X86-AVX512-NEXT:    vmovss %xmm0, (%esp) # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x11,0x04,0x24]
; X86-AVX512-NEXT:    flds (%esp) # encoding: [0xd9,0x04,0x24]
; X86-AVX512-NEXT:    popl %eax # encoding: [0x58]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mm_cvtss_f32:
; X64:       # %bb.0:
; X64-NEXT:    retq # encoding: [0xc3]
  %res = extractelement <4 x float> %a0, i32 0
  ret float %res
}

define i32 @test_mm_cvtss_si32(<4 x float> %a0) nounwind {
; SSE-LABEL: test_mm_cvtss_si32:
; SSE:       # %bb.0:
; SSE-NEXT:    cvtss2si %xmm0, %eax # encoding: [0xf3,0x0f,0x2d,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cvtss_si32:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcvtss2si %xmm0, %eax # encoding: [0xc5,0xfa,0x2d,0xc0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cvtss_si32:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcvtss2si %xmm0, %eax # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x2d,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.cvtss2si(<4 x float> %a0)
  ret i32 %res
}

define i32 @test_mm_cvttss_si(<4 x float> %a0) nounwind {
; SSE-LABEL: test_mm_cvttss_si:
; SSE:       # %bb.0:
; SSE-NEXT:    cvttss2si %xmm0, %eax # encoding: [0xf3,0x0f,0x2c,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cvttss_si:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcvttss2si %xmm0, %eax # encoding: [0xc5,0xfa,0x2c,0xc0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cvttss_si:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcvttss2si %xmm0, %eax # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x2c,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.cvttss2si(<4 x float> %a0)
  ret i32 %res
}
declare i32 @llvm.x86.sse.cvttss2si(<4 x float>) nounwind readnone

define i32 @test_mm_cvttss_si32(<4 x float> %a0) nounwind {
; SSE-LABEL: test_mm_cvttss_si32:
; SSE:       # %bb.0:
; SSE-NEXT:    cvttss2si %xmm0, %eax # encoding: [0xf3,0x0f,0x2c,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_cvttss_si32:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vcvttss2si %xmm0, %eax # encoding: [0xc5,0xfa,0x2c,0xc0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_cvttss_si32:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vcvttss2si %xmm0, %eax # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x2c,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.cvttss2si(<4 x float> %a0)
  ret i32 %res
}

define <4 x float> @test_mm_div_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_div_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    divps %xmm1, %xmm0 # encoding: [0x0f,0x5e,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_div_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vdivps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x5e,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_div_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vdivps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x5e,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = fdiv <4 x float> %a0, %a1
  ret <4 x float> %res
}

define <4 x float> @test_mm_div_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_div_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    divss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0x5e,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_div_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vdivss %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0x5e,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_div_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vdivss %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x5e,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %ext0 = extractelement <4 x float> %a0, i32 0
  %ext1 = extractelement <4 x float> %a1, i32 0
  %fdiv = fdiv float %ext0, %ext1
  %res = insertelement <4 x float> %a0, float %fdiv, i32 0
  ret <4 x float> %res
}

define i32 @test_MM_GET_EXCEPTION_MASK() nounwind {
; X86-SSE-LABEL: test_MM_GET_EXCEPTION_MASK:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    pushl %eax # encoding: [0x50]
; X86-SSE-NEXT:    movl %esp, %eax # encoding: [0x89,0xe0]
; X86-SSE-NEXT:    stmxcsr (%eax) # encoding: [0x0f,0xae,0x18]
; X86-SSE-NEXT:    movl (%esp), %eax # encoding: [0x8b,0x04,0x24]
; X86-SSE-NEXT:    andl $8064, %eax # encoding: [0x25,0x80,0x1f,0x00,0x00]
; X86-SSE-NEXT:    # imm = 0x1F80
; X86-SSE-NEXT:    popl %ecx # encoding: [0x59]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX-LABEL: test_MM_GET_EXCEPTION_MASK:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    pushl %eax # encoding: [0x50]
; X86-AVX-NEXT:    movl %esp, %eax # encoding: [0x89,0xe0]
; X86-AVX-NEXT:    vstmxcsr (%eax) # encoding: [0xc5,0xf8,0xae,0x18]
; X86-AVX-NEXT:    movl (%esp), %eax # encoding: [0x8b,0x04,0x24]
; X86-AVX-NEXT:    andl $8064, %eax # encoding: [0x25,0x80,0x1f,0x00,0x00]
; X86-AVX-NEXT:    # imm = 0x1F80
; X86-AVX-NEXT:    popl %ecx # encoding: [0x59]
; X86-AVX-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_MM_GET_EXCEPTION_MASK:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-SSE-NEXT:    stmxcsr (%rax) # encoding: [0x0f,0xae,0x18]
; X64-SSE-NEXT:    movl -{{[0-9]+}}(%rsp), %eax # encoding: [0x8b,0x44,0x24,0xfc]
; X64-SSE-NEXT:    andl $8064, %eax # encoding: [0x25,0x80,0x1f,0x00,0x00]
; X64-SSE-NEXT:    # imm = 0x1F80
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX-LABEL: test_MM_GET_EXCEPTION_MASK:
; X64-AVX:       # %bb.0:
; X64-AVX-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-AVX-NEXT:    vstmxcsr (%rax) # encoding: [0xc5,0xf8,0xae,0x18]
; X64-AVX-NEXT:    movl -{{[0-9]+}}(%rsp), %eax # encoding: [0x8b,0x44,0x24,0xfc]
; X64-AVX-NEXT:    andl $8064, %eax # encoding: [0x25,0x80,0x1f,0x00,0x00]
; X64-AVX-NEXT:    # imm = 0x1F80
; X64-AVX-NEXT:    retq # encoding: [0xc3]
  %1 = alloca i32, align 4
  %2 = bitcast i32* %1 to i8*
  call void @llvm.x86.sse.stmxcsr(i8* %2)
  %3 = load i32, i32* %1, align 4
  %4 = and i32 %3, 8064
  ret i32 %4
}
declare void @llvm.x86.sse.stmxcsr(i8*) nounwind readnone

define i32 @test_MM_GET_EXCEPTION_STATE() nounwind {
; X86-SSE-LABEL: test_MM_GET_EXCEPTION_STATE:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    pushl %eax # encoding: [0x50]
; X86-SSE-NEXT:    movl %esp, %eax # encoding: [0x89,0xe0]
; X86-SSE-NEXT:    stmxcsr (%eax) # encoding: [0x0f,0xae,0x18]
; X86-SSE-NEXT:    movl (%esp), %eax # encoding: [0x8b,0x04,0x24]
; X86-SSE-NEXT:    andl $63, %eax # encoding: [0x83,0xe0,0x3f]
; X86-SSE-NEXT:    popl %ecx # encoding: [0x59]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX-LABEL: test_MM_GET_EXCEPTION_STATE:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    pushl %eax # encoding: [0x50]
; X86-AVX-NEXT:    movl %esp, %eax # encoding: [0x89,0xe0]
; X86-AVX-NEXT:    vstmxcsr (%eax) # encoding: [0xc5,0xf8,0xae,0x18]
; X86-AVX-NEXT:    movl (%esp), %eax # encoding: [0x8b,0x04,0x24]
; X86-AVX-NEXT:    andl $63, %eax # encoding: [0x83,0xe0,0x3f]
; X86-AVX-NEXT:    popl %ecx # encoding: [0x59]
; X86-AVX-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_MM_GET_EXCEPTION_STATE:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-SSE-NEXT:    stmxcsr (%rax) # encoding: [0x0f,0xae,0x18]
; X64-SSE-NEXT:    movl -{{[0-9]+}}(%rsp), %eax # encoding: [0x8b,0x44,0x24,0xfc]
; X64-SSE-NEXT:    andl $63, %eax # encoding: [0x83,0xe0,0x3f]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX-LABEL: test_MM_GET_EXCEPTION_STATE:
; X64-AVX:       # %bb.0:
; X64-AVX-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-AVX-NEXT:    vstmxcsr (%rax) # encoding: [0xc5,0xf8,0xae,0x18]
; X64-AVX-NEXT:    movl -{{[0-9]+}}(%rsp), %eax # encoding: [0x8b,0x44,0x24,0xfc]
; X64-AVX-NEXT:    andl $63, %eax # encoding: [0x83,0xe0,0x3f]
; X64-AVX-NEXT:    retq # encoding: [0xc3]
  %1 = alloca i32, align 4
  %2 = bitcast i32* %1 to i8*
  call void @llvm.x86.sse.stmxcsr(i8* %2)
  %3 = load i32, i32* %1, align 4
  %4 = and i32 %3, 63
  ret i32 %4
}

define i32 @test_MM_GET_FLUSH_ZERO_MODE() nounwind {
; X86-SSE-LABEL: test_MM_GET_FLUSH_ZERO_MODE:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    pushl %eax # encoding: [0x50]
; X86-SSE-NEXT:    movl %esp, %eax # encoding: [0x89,0xe0]
; X86-SSE-NEXT:    stmxcsr (%eax) # encoding: [0x0f,0xae,0x18]
; X86-SSE-NEXT:    movl (%esp), %eax # encoding: [0x8b,0x04,0x24]
; X86-SSE-NEXT:    andl $32768, %eax # encoding: [0x25,0x00,0x80,0x00,0x00]
; X86-SSE-NEXT:    # imm = 0x8000
; X86-SSE-NEXT:    popl %ecx # encoding: [0x59]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX-LABEL: test_MM_GET_FLUSH_ZERO_MODE:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    pushl %eax # encoding: [0x50]
; X86-AVX-NEXT:    movl %esp, %eax # encoding: [0x89,0xe0]
; X86-AVX-NEXT:    vstmxcsr (%eax) # encoding: [0xc5,0xf8,0xae,0x18]
; X86-AVX-NEXT:    movl (%esp), %eax # encoding: [0x8b,0x04,0x24]
; X86-AVX-NEXT:    andl $32768, %eax # encoding: [0x25,0x00,0x80,0x00,0x00]
; X86-AVX-NEXT:    # imm = 0x8000
; X86-AVX-NEXT:    popl %ecx # encoding: [0x59]
; X86-AVX-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_MM_GET_FLUSH_ZERO_MODE:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-SSE-NEXT:    stmxcsr (%rax) # encoding: [0x0f,0xae,0x18]
; X64-SSE-NEXT:    movl -{{[0-9]+}}(%rsp), %eax # encoding: [0x8b,0x44,0x24,0xfc]
; X64-SSE-NEXT:    andl $32768, %eax # encoding: [0x25,0x00,0x80,0x00,0x00]
; X64-SSE-NEXT:    # imm = 0x8000
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX-LABEL: test_MM_GET_FLUSH_ZERO_MODE:
; X64-AVX:       # %bb.0:
; X64-AVX-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-AVX-NEXT:    vstmxcsr (%rax) # encoding: [0xc5,0xf8,0xae,0x18]
; X64-AVX-NEXT:    movl -{{[0-9]+}}(%rsp), %eax # encoding: [0x8b,0x44,0x24,0xfc]
; X64-AVX-NEXT:    andl $32768, %eax # encoding: [0x25,0x00,0x80,0x00,0x00]
; X64-AVX-NEXT:    # imm = 0x8000
; X64-AVX-NEXT:    retq # encoding: [0xc3]
  %1 = alloca i32, align 4
  %2 = bitcast i32* %1 to i8*
  call void @llvm.x86.sse.stmxcsr(i8* %2)
  %3 = load i32, i32* %1, align 4
  %4 = and i32 %3, 32768
  ret i32 %4
}

define i32 @test_MM_GET_ROUNDING_MODE() nounwind {
; X86-SSE-LABEL: test_MM_GET_ROUNDING_MODE:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    pushl %eax # encoding: [0x50]
; X86-SSE-NEXT:    movl %esp, %eax # encoding: [0x89,0xe0]
; X86-SSE-NEXT:    stmxcsr (%eax) # encoding: [0x0f,0xae,0x18]
; X86-SSE-NEXT:    movl (%esp), %eax # encoding: [0x8b,0x04,0x24]
; X86-SSE-NEXT:    andl $24576, %eax # encoding: [0x25,0x00,0x60,0x00,0x00]
; X86-SSE-NEXT:    # imm = 0x6000
; X86-SSE-NEXT:    popl %ecx # encoding: [0x59]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX-LABEL: test_MM_GET_ROUNDING_MODE:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    pushl %eax # encoding: [0x50]
; X86-AVX-NEXT:    movl %esp, %eax # encoding: [0x89,0xe0]
; X86-AVX-NEXT:    vstmxcsr (%eax) # encoding: [0xc5,0xf8,0xae,0x18]
; X86-AVX-NEXT:    movl (%esp), %eax # encoding: [0x8b,0x04,0x24]
; X86-AVX-NEXT:    andl $24576, %eax # encoding: [0x25,0x00,0x60,0x00,0x00]
; X86-AVX-NEXT:    # imm = 0x6000
; X86-AVX-NEXT:    popl %ecx # encoding: [0x59]
; X86-AVX-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_MM_GET_ROUNDING_MODE:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-SSE-NEXT:    stmxcsr (%rax) # encoding: [0x0f,0xae,0x18]
; X64-SSE-NEXT:    movl -{{[0-9]+}}(%rsp), %eax # encoding: [0x8b,0x44,0x24,0xfc]
; X64-SSE-NEXT:    andl $24576, %eax # encoding: [0x25,0x00,0x60,0x00,0x00]
; X64-SSE-NEXT:    # imm = 0x6000
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX-LABEL: test_MM_GET_ROUNDING_MODE:
; X64-AVX:       # %bb.0:
; X64-AVX-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-AVX-NEXT:    vstmxcsr (%rax) # encoding: [0xc5,0xf8,0xae,0x18]
; X64-AVX-NEXT:    movl -{{[0-9]+}}(%rsp), %eax # encoding: [0x8b,0x44,0x24,0xfc]
; X64-AVX-NEXT:    andl $24576, %eax # encoding: [0x25,0x00,0x60,0x00,0x00]
; X64-AVX-NEXT:    # imm = 0x6000
; X64-AVX-NEXT:    retq # encoding: [0xc3]
  %1 = alloca i32, align 4
  %2 = bitcast i32* %1 to i8*
  call void @llvm.x86.sse.stmxcsr(i8* %2)
  %3 = load i32, i32* %1, align 4
  %4 = and i32 %3, 24576
  ret i32 %4
}

define i32 @test_mm_getcsr() nounwind {
; X86-SSE-LABEL: test_mm_getcsr:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    pushl %eax # encoding: [0x50]
; X86-SSE-NEXT:    movl %esp, %eax # encoding: [0x89,0xe0]
; X86-SSE-NEXT:    stmxcsr (%eax) # encoding: [0x0f,0xae,0x18]
; X86-SSE-NEXT:    movl (%esp), %eax # encoding: [0x8b,0x04,0x24]
; X86-SSE-NEXT:    popl %ecx # encoding: [0x59]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX-LABEL: test_mm_getcsr:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    pushl %eax # encoding: [0x50]
; X86-AVX-NEXT:    movl %esp, %eax # encoding: [0x89,0xe0]
; X86-AVX-NEXT:    vstmxcsr (%eax) # encoding: [0xc5,0xf8,0xae,0x18]
; X86-AVX-NEXT:    movl (%esp), %eax # encoding: [0x8b,0x04,0x24]
; X86-AVX-NEXT:    popl %ecx # encoding: [0x59]
; X86-AVX-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_getcsr:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-SSE-NEXT:    stmxcsr (%rax) # encoding: [0x0f,0xae,0x18]
; X64-SSE-NEXT:    movl -{{[0-9]+}}(%rsp), %eax # encoding: [0x8b,0x44,0x24,0xfc]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX-LABEL: test_mm_getcsr:
; X64-AVX:       # %bb.0:
; X64-AVX-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-AVX-NEXT:    vstmxcsr (%rax) # encoding: [0xc5,0xf8,0xae,0x18]
; X64-AVX-NEXT:    movl -{{[0-9]+}}(%rsp), %eax # encoding: [0x8b,0x44,0x24,0xfc]
; X64-AVX-NEXT:    retq # encoding: [0xc3]
  %1 = alloca i32, align 4
  %2 = bitcast i32* %1 to i8*
  call void @llvm.x86.sse.stmxcsr(i8* %2)
  %3 = load i32, i32* %1, align 4
  ret i32 %3
}

define <4 x float> @test_mm_load_ps(float* %a0) nounwind {
; X86-SSE-LABEL: test_mm_load_ps:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-SSE-NEXT:    movaps (%eax), %xmm0 # encoding: [0x0f,0x28,0x00]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_load_ps:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vmovaps (%eax), %xmm0 # encoding: [0xc5,0xf8,0x28,0x00]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_load_ps:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vmovaps (%eax), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0x00]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_load_ps:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movaps (%rdi), %xmm0 # encoding: [0x0f,0x28,0x07]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_load_ps:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vmovaps (%rdi), %xmm0 # encoding: [0xc5,0xf8,0x28,0x07]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_load_ps:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vmovaps (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0x07]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %arg0 = bitcast float* %a0 to <4 x float>*
  %res = load <4 x float>, <4 x float>* %arg0, align 16
  ret <4 x float> %res
}

define <4 x float> @test_mm_load_ps1(float* %a0) nounwind {
; X86-SSE-LABEL: test_mm_load_ps1:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-SSE-NEXT:    movss (%eax), %xmm0 # encoding: [0xf3,0x0f,0x10,0x00]
; X86-SSE-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    shufps $0, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x00]
; X86-SSE-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_load_ps1:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vbroadcastss (%eax), %xmm0 # encoding: [0xc4,0xe2,0x79,0x18,0x00]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_load_ps1:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vbroadcastss (%eax), %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x18,0x00]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_load_ps1:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movss (%rdi), %xmm0 # encoding: [0xf3,0x0f,0x10,0x07]
; X64-SSE-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X64-SSE-NEXT:    shufps $0, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x00]
; X64-SSE-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_load_ps1:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vbroadcastss (%rdi), %xmm0 # encoding: [0xc4,0xe2,0x79,0x18,0x07]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_load_ps1:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vbroadcastss (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x18,0x07]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %ld = load float, float* %a0, align 4
  %res0 = insertelement <4 x float> undef, float %ld, i32 0
  %res1 = insertelement <4 x float> %res0, float %ld, i32 1
  %res2 = insertelement <4 x float> %res1, float %ld, i32 2
  %res3 = insertelement <4 x float> %res2, float %ld, i32 3
  ret <4 x float> %res3
}

define <4 x float> @test_mm_load_ss(float* %a0) nounwind {
; X86-SSE-LABEL: test_mm_load_ss:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-SSE-NEXT:    movss (%eax), %xmm0 # encoding: [0xf3,0x0f,0x10,0x00]
; X86-SSE-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_load_ss:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vmovss (%eax), %xmm0 # encoding: [0xc5,0xfa,0x10,0x00]
; X86-AVX1-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_load_ss:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vmovss (%eax), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x00]
; X86-AVX512-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_load_ss:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movss (%rdi), %xmm0 # encoding: [0xf3,0x0f,0x10,0x07]
; X64-SSE-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_load_ss:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vmovss (%rdi), %xmm0 # encoding: [0xc5,0xfa,0x10,0x07]
; X64-AVX1-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_load_ss:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vmovss (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x07]
; X64-AVX512-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %ld = load float, float* %a0, align 1
  %res0 = insertelement <4 x float> undef, float %ld, i32 0
  %res1 = insertelement <4 x float> %res0, float 0.0, i32 1
  %res2 = insertelement <4 x float> %res1, float 0.0, i32 2
  %res3 = insertelement <4 x float> %res2, float 0.0, i32 3
  ret <4 x float> %res3
}

define <4 x float> @test_mm_load1_ps(float* %a0) nounwind {
; X86-SSE-LABEL: test_mm_load1_ps:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-SSE-NEXT:    movss (%eax), %xmm0 # encoding: [0xf3,0x0f,0x10,0x00]
; X86-SSE-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    shufps $0, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x00]
; X86-SSE-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_load1_ps:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vbroadcastss (%eax), %xmm0 # encoding: [0xc4,0xe2,0x79,0x18,0x00]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_load1_ps:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vbroadcastss (%eax), %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x18,0x00]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_load1_ps:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movss (%rdi), %xmm0 # encoding: [0xf3,0x0f,0x10,0x07]
; X64-SSE-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X64-SSE-NEXT:    shufps $0, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x00]
; X64-SSE-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_load1_ps:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vbroadcastss (%rdi), %xmm0 # encoding: [0xc4,0xe2,0x79,0x18,0x07]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_load1_ps:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vbroadcastss (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x18,0x07]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %ld = load float, float* %a0, align 4
  %res0 = insertelement <4 x float> undef, float %ld, i32 0
  %res1 = insertelement <4 x float> %res0, float %ld, i32 1
  %res2 = insertelement <4 x float> %res1, float %ld, i32 2
  %res3 = insertelement <4 x float> %res2, float %ld, i32 3
  ret <4 x float> %res3
}

define <4 x float> @test_mm_loadh_pi(<4 x float> %a0, x86_mmx* %a1) {
; X86-SSE-LABEL: test_mm_loadh_pi:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-SSE-NEXT:    movss (%eax), %xmm1 # encoding: [0xf3,0x0f,0x10,0x08]
; X86-SSE-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    movss 4(%eax), %xmm2 # encoding: [0xf3,0x0f,0x10,0x50,0x04]
; X86-SSE-NEXT:    # xmm2 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    shufps $0, %xmm1, %xmm2 # encoding: [0x0f,0xc6,0xd1,0x00]
; X86-SSE-NEXT:    # xmm2 = xmm2[0,0],xmm1[0,0]
; X86-SSE-NEXT:    shufps $36, %xmm2, %xmm0 # encoding: [0x0f,0xc6,0xc2,0x24]
; X86-SSE-NEXT:    # xmm0 = xmm0[0,1],xmm2[2,0]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_loadh_pi:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vmovhpd (%eax), %xmm0, %xmm0 # encoding: [0xc5,0xf9,0x16,0x00]
; X86-AVX1-NEXT:    # xmm0 = xmm0[0],mem[0]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_loadh_pi:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vmovhpd (%eax), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x16,0x00]
; X86-AVX512-NEXT:    # xmm0 = xmm0[0],mem[0]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_loadh_pi:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
; X64-SSE-NEXT:    movl %eax, -{{[0-9]+}}(%rsp) # encoding: [0x89,0x44,0x24,0xf8]
; X64-SSE-NEXT:    shrq $32, %rax # encoding: [0x48,0xc1,0xe8,0x20]
; X64-SSE-NEXT:    movl %eax, -{{[0-9]+}}(%rsp) # encoding: [0x89,0x44,0x24,0xfc]
; X64-SSE-NEXT:    movss -{{[0-9]+}}(%rsp), %xmm1 # encoding: [0xf3,0x0f,0x10,0x4c,0x24,0xf8]
; X64-SSE-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X64-SSE-NEXT:    movss -{{[0-9]+}}(%rsp), %xmm2 # encoding: [0xf3,0x0f,0x10,0x54,0x24,0xfc]
; X64-SSE-NEXT:    # xmm2 = mem[0],zero,zero,zero
; X64-SSE-NEXT:    unpcklps %xmm2, %xmm1 # encoding: [0x0f,0x14,0xca]
; X64-SSE-NEXT:    # xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
; X64-SSE-NEXT:    movlhps %xmm1, %xmm0 # encoding: [0x0f,0x16,0xc1]
; X64-SSE-NEXT:    # xmm0 = xmm0[0],xmm1[0]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_loadh_pi:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vmovhpd (%rdi), %xmm0, %xmm0 # encoding: [0xc5,0xf9,0x16,0x07]
; X64-AVX1-NEXT:    # xmm0 = xmm0[0],mem[0]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_loadh_pi:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vmovhpd (%rdi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x16,0x07]
; X64-AVX512-NEXT:    # xmm0 = xmm0[0],mem[0]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %ptr = bitcast x86_mmx* %a1 to <2 x float>*
  %ld  = load <2 x float>, <2 x float>* %ptr
  %ext = shufflevector <2 x float> %ld, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
  %res = shufflevector <4 x float> %a0, <4 x float> %ext, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
  ret <4 x float> %res
}

define <4 x float> @test_mm_loadl_pi(<4 x float> %a0, x86_mmx* %a1) {
; X86-SSE-LABEL: test_mm_loadl_pi:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-SSE-NEXT:    movss (%eax), %xmm2 # encoding: [0xf3,0x0f,0x10,0x10]
; X86-SSE-NEXT:    # xmm2 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    movss 4(%eax), %xmm1 # encoding: [0xf3,0x0f,0x10,0x48,0x04]
; X86-SSE-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    shufps $0, %xmm2, %xmm1 # encoding: [0x0f,0xc6,0xca,0x00]
; X86-SSE-NEXT:    # xmm1 = xmm1[0,0],xmm2[0,0]
; X86-SSE-NEXT:    shufps $226, %xmm0, %xmm1 # encoding: [0x0f,0xc6,0xc8,0xe2]
; X86-SSE-NEXT:    # xmm1 = xmm1[2,0],xmm0[2,3]
; X86-SSE-NEXT:    movaps %xmm1, %xmm0 # encoding: [0x0f,0x28,0xc1]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_loadl_pi:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vmovlpd (%eax), %xmm0, %xmm0 # encoding: [0xc5,0xf9,0x12,0x00]
; X86-AVX1-NEXT:    # xmm0 = mem[0],xmm0[1]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_loadl_pi:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vmovlpd (%eax), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x12,0x00]
; X86-AVX512-NEXT:    # xmm0 = mem[0],xmm0[1]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_loadl_pi:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movq (%rdi), %rax # encoding: [0x48,0x8b,0x07]
; X64-SSE-NEXT:    movl %eax, -{{[0-9]+}}(%rsp) # encoding: [0x89,0x44,0x24,0xf8]
; X64-SSE-NEXT:    shrq $32, %rax # encoding: [0x48,0xc1,0xe8,0x20]
; X64-SSE-NEXT:    movl %eax, -{{[0-9]+}}(%rsp) # encoding: [0x89,0x44,0x24,0xfc]
; X64-SSE-NEXT:    movss -{{[0-9]+}}(%rsp), %xmm1 # encoding: [0xf3,0x0f,0x10,0x4c,0x24,0xf8]
; X64-SSE-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X64-SSE-NEXT:    movss -{{[0-9]+}}(%rsp), %xmm2 # encoding: [0xf3,0x0f,0x10,0x54,0x24,0xfc]
; X64-SSE-NEXT:    # xmm2 = mem[0],zero,zero,zero
; X64-SSE-NEXT:    unpcklps %xmm2, %xmm1 # encoding: [0x0f,0x14,0xca]
; X64-SSE-NEXT:    # xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
; X64-SSE-NEXT:    shufps $228, %xmm0, %xmm1 # encoding: [0x0f,0xc6,0xc8,0xe4]
; X64-SSE-NEXT:    # xmm1 = xmm1[0,1],xmm0[2,3]
; X64-SSE-NEXT:    movaps %xmm1, %xmm0 # encoding: [0x0f,0x28,0xc1]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_loadl_pi:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vmovlpd (%rdi), %xmm0, %xmm0 # encoding: [0xc5,0xf9,0x12,0x07]
; X64-AVX1-NEXT:    # xmm0 = mem[0],xmm0[1]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_loadl_pi:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vmovlpd (%rdi), %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x12,0x07]
; X64-AVX512-NEXT:    # xmm0 = mem[0],xmm0[1]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %ptr = bitcast x86_mmx* %a1 to <2 x float>*
  %ld  = load <2 x float>, <2 x float>* %ptr
  %ext = shufflevector <2 x float> %ld, <2 x float> undef, <4 x i32> <i32 0, i32 1, i32 undef, i32 undef>
  %res = shufflevector <4 x float> %a0, <4 x float> %ext, <4 x i32> <i32 4, i32 5, i32 2, i32 3>
  ret <4 x float> %res
}

define <4 x float> @test_mm_loadr_ps(float* %a0) nounwind {
; X86-SSE-LABEL: test_mm_loadr_ps:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-SSE-NEXT:    movaps (%eax), %xmm0 # encoding: [0x0f,0x28,0x00]
; X86-SSE-NEXT:    shufps $27, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x1b]
; X86-SSE-NEXT:    # xmm0 = xmm0[3,2,1,0]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_loadr_ps:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vpermilps $27, (%eax), %xmm0 # encoding: [0xc4,0xe3,0x79,0x04,0x00,0x1b]
; X86-AVX1-NEXT:    # xmm0 = mem[3,2,1,0]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_loadr_ps:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vpermilps $27, (%eax), %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x79,0x04,0x00,0x1b]
; X86-AVX512-NEXT:    # xmm0 = mem[3,2,1,0]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_loadr_ps:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movaps (%rdi), %xmm0 # encoding: [0x0f,0x28,0x07]
; X64-SSE-NEXT:    shufps $27, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x1b]
; X64-SSE-NEXT:    # xmm0 = xmm0[3,2,1,0]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_loadr_ps:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vpermilps $27, (%rdi), %xmm0 # encoding: [0xc4,0xe3,0x79,0x04,0x07,0x1b]
; X64-AVX1-NEXT:    # xmm0 = mem[3,2,1,0]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_loadr_ps:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vpermilps $27, (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x79,0x04,0x07,0x1b]
; X64-AVX512-NEXT:    # xmm0 = mem[3,2,1,0]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %arg0 = bitcast float* %a0 to <4 x float>*
  %ld = load <4 x float>, <4 x float>* %arg0, align 16
  %res = shufflevector <4 x float> %ld, <4 x float> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
  ret <4 x float> %res
}

define <4 x float> @test_mm_loadu_ps(float* %a0) nounwind {
; X86-SSE-LABEL: test_mm_loadu_ps:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-SSE-NEXT:    movups (%eax), %xmm0 # encoding: [0x0f,0x10,0x00]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_loadu_ps:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vmovups (%eax), %xmm0 # encoding: [0xc5,0xf8,0x10,0x00]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_loadu_ps:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vmovups (%eax), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x10,0x00]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_loadu_ps:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movups (%rdi), %xmm0 # encoding: [0x0f,0x10,0x07]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_loadu_ps:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vmovups (%rdi), %xmm0 # encoding: [0xc5,0xf8,0x10,0x07]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_loadu_ps:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vmovups (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x10,0x07]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %arg0 = bitcast float* %a0 to <4 x float>*
  %res = load <4 x float>, <4 x float>* %arg0, align 1
  ret <4 x float> %res
}

define <4 x float> @test_mm_max_ps(<4 x float> %a0, <4 x float> %a1) {
; SSE-LABEL: test_mm_max_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    maxps %xmm1, %xmm0 # encoding: [0x0f,0x5f,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_max_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vmaxps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x5f,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_max_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vmaxps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x5f,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.sse.max.ps(<4 x float> %a0, <4 x float> %a1)
  ret <4 x float> %res
}
declare <4 x float> @llvm.x86.sse.max.ps(<4 x float>, <4 x float>) nounwind readnone

define <4 x float> @test_mm_max_ss(<4 x float> %a0, <4 x float> %a1) {
; SSE-LABEL: test_mm_max_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    maxss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0x5f,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_max_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vmaxss %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0x5f,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_max_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vmaxss %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x5f,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.sse.max.ss(<4 x float> %a0, <4 x float> %a1)
  ret <4 x float> %res
}
declare <4 x float> @llvm.x86.sse.max.ss(<4 x float>, <4 x float>) nounwind readnone

define <4 x float> @test_mm_min_ps(<4 x float> %a0, <4 x float> %a1) {
; SSE-LABEL: test_mm_min_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    minps %xmm1, %xmm0 # encoding: [0x0f,0x5d,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_min_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vminps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x5d,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_min_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vminps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x5d,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.sse.min.ps(<4 x float> %a0, <4 x float> %a1)
  ret <4 x float> %res
}
declare <4 x float> @llvm.x86.sse.min.ps(<4 x float>, <4 x float>) nounwind readnone

define <4 x float> @test_mm_min_ss(<4 x float> %a0, <4 x float> %a1) {
; SSE-LABEL: test_mm_min_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    minss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0x5d,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_min_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vminss %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0x5d,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_min_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vminss %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x5d,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.sse.min.ss(<4 x float> %a0, <4 x float> %a1)
  ret <4 x float> %res
}
declare <4 x float> @llvm.x86.sse.min.ss(<4 x float>, <4 x float>) nounwind readnone

define <4 x float> @test_mm_move_ss(<4 x float> %a0, <4 x float> %a1) {
; SSE-LABEL: test_mm_move_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    movss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0x10,0xc1]
; SSE-NEXT:    # xmm0 = xmm1[0],xmm0[1,2,3]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_move_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    vblendps $1, %xmm1, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x0c,0xc1,0x01]
; AVX-NEXT:    # xmm0 = xmm1[0],xmm0[1,2,3]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = shufflevector <4 x float> %a0, <4 x float> %a1, <4 x i32> <i32 4, i32 1, i32 2, i32 3>
  ret <4 x float> %res
}

define <4 x float> @test_mm_movehl_ps(<4 x float> %a0, <4 x float> %a1) {
; SSE-LABEL: test_mm_movehl_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    movhlps %xmm1, %xmm0 # encoding: [0x0f,0x12,0xc1]
; SSE-NEXT:    # xmm0 = xmm1[1],xmm0[1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_movehl_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vunpckhpd %xmm0, %xmm1, %xmm0 # encoding: [0xc5,0xf1,0x15,0xc0]
; AVX1-NEXT:    # xmm0 = xmm1[1],xmm0[1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_movehl_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vunpckhpd %xmm0, %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf1,0x15,0xc0]
; AVX512-NEXT:    # xmm0 = xmm1[1],xmm0[1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = shufflevector <4 x float> %a0, <4 x float> %a1, <4 x i32> <i32 6, i32 7, i32 2, i32 3>
  ret <4 x float> %res
}

define <4 x float> @test_mm_movelh_ps(<4 x float> %a0, <4 x float> %a1) {
; SSE-LABEL: test_mm_movelh_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    movlhps %xmm1, %xmm0 # encoding: [0x0f,0x16,0xc1]
; SSE-NEXT:    # xmm0 = xmm0[0],xmm1[0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_movelh_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vmovlhps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x16,0xc1]
; AVX1-NEXT:    # xmm0 = xmm0[0],xmm1[0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_movelh_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vmovlhps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x16,0xc1]
; AVX512-NEXT:    # xmm0 = xmm0[0],xmm1[0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = shufflevector <4 x float> %a0, <4 x float> %a1, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
  ret <4 x float> %res
}

define i32 @test_mm_movemask_ps(<4 x float> %a0) nounwind {
; SSE-LABEL: test_mm_movemask_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    movmskps %xmm0, %eax # encoding: [0x0f,0x50,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_movemask_ps:
; AVX:       # %bb.0:
; AVX-NEXT:    vmovmskps %xmm0, %eax # encoding: [0xc5,0xf8,0x50,0xc0]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.movmsk.ps(<4 x float> %a0)
  ret i32 %res
}
declare i32 @llvm.x86.sse.movmsk.ps(<4 x float>) nounwind readnone

define <4 x float> @test_mm_mul_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_mul_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    mulps %xmm1, %xmm0 # encoding: [0x0f,0x59,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_mul_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vmulps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x59,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_mul_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vmulps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x59,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = fmul <4 x float> %a0, %a1
  ret <4 x float> %res
}

define <4 x float> @test_mm_mul_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_mul_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    mulss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0x59,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_mul_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vmulss %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0x59,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_mul_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vmulss %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x59,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %ext0 = extractelement <4 x float> %a0, i32 0
  %ext1 = extractelement <4 x float> %a1, i32 0
  %fmul = fmul float %ext0, %ext1
  %res = insertelement <4 x float> %a0, float %fmul, i32 0
  ret <4 x float> %res
}

define <4 x float> @test_mm_or_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_or_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    orps %xmm1, %xmm0 # encoding: [0x0f,0x56,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_or_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vorps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x56,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_or_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vorps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x56,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %arg0 = bitcast <4 x float> %a0 to <4 x i32>
  %arg1 = bitcast <4 x float> %a1 to <4 x i32>
  %res = or <4 x i32> %arg0, %arg1
  %bc = bitcast <4 x i32> %res to <4 x float>
  ret <4 x float> %bc
}

define void @test_mm_prefetch(i8* %a0) {
; X86-LABEL: test_mm_prefetch:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    prefetchnta (%eax) # encoding: [0x0f,0x18,0x00]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_mm_prefetch:
; X64:       # %bb.0:
; X64-NEXT:    prefetchnta (%rdi) # encoding: [0x0f,0x18,0x07]
; X64-NEXT:    retq # encoding: [0xc3]
  call void @llvm.prefetch(i8* %a0, i32 0, i32 0, i32 1)
  ret void
}
declare void @llvm.prefetch(i8* nocapture, i32, i32, i32) nounwind readnone

define <4 x float> @test_mm_rcp_ps(<4 x float> %a0) {
; SSE-LABEL: test_mm_rcp_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    rcpps %xmm0, %xmm0 # encoding: [0x0f,0x53,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_rcp_ps:
; AVX:       # %bb.0:
; AVX-NEXT:    vrcpps %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x53,0xc0]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.sse.rcp.ps(<4 x float> %a0)
  ret <4 x float> %res
}
declare <4 x float> @llvm.x86.sse.rcp.ps(<4 x float>) nounwind readnone

define <4 x float> @test_mm_rcp_ss(<4 x float> %a0) {
; SSE-LABEL: test_mm_rcp_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    rcpss %xmm0, %xmm0 # encoding: [0xf3,0x0f,0x53,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_rcp_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    vrcpss %xmm0, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0x53,0xc0]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %rcp = call <4 x float> @llvm.x86.sse.rcp.ss(<4 x float> %a0)
  ret <4 x float> %rcp
}
declare <4 x float> @llvm.x86.sse.rcp.ss(<4 x float>) nounwind readnone

define <4 x float> @test_mm_rsqrt_ps(<4 x float> %a0) {
; SSE-LABEL: test_mm_rsqrt_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    rsqrtps %xmm0, %xmm0 # encoding: [0x0f,0x52,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_rsqrt_ps:
; AVX:       # %bb.0:
; AVX-NEXT:    vrsqrtps %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x52,0xc0]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.sse.rsqrt.ps(<4 x float> %a0)
  ret <4 x float> %res
}
declare <4 x float> @llvm.x86.sse.rsqrt.ps(<4 x float>) nounwind readnone

define <4 x float> @test_mm_rsqrt_ss(<4 x float> %a0) {
; SSE-LABEL: test_mm_rsqrt_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    rsqrtss %xmm0, %xmm0 # encoding: [0xf3,0x0f,0x52,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX-LABEL: test_mm_rsqrt_ss:
; AVX:       # %bb.0:
; AVX-NEXT:    vrsqrtss %xmm0, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0x52,0xc0]
; AVX-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %rsqrt = call <4 x float> @llvm.x86.sse.rsqrt.ss(<4 x float> %a0)
  ret <4 x float> %rsqrt
}
declare <4 x float> @llvm.x86.sse.rsqrt.ss(<4 x float>) nounwind readnone

define void @test_MM_SET_EXCEPTION_MASK(i32 %a0) nounwind {
; X86-SSE-LABEL: test_MM_SET_EXCEPTION_MASK:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    pushl %eax # encoding: [0x50]
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x08]
; X86-SSE-NEXT:    movl %esp, %ecx # encoding: [0x89,0xe1]
; X86-SSE-NEXT:    stmxcsr (%ecx) # encoding: [0x0f,0xae,0x19]
; X86-SSE-NEXT:    movl (%esp), %edx # encoding: [0x8b,0x14,0x24]
; X86-SSE-NEXT:    andl $-8065, %edx # encoding: [0x81,0xe2,0x7f,0xe0,0xff,0xff]
; X86-SSE-NEXT:    # imm = 0xE07F
; X86-SSE-NEXT:    orl %eax, %edx # encoding: [0x09,0xc2]
; X86-SSE-NEXT:    movl %edx, (%esp) # encoding: [0x89,0x14,0x24]
; X86-SSE-NEXT:    ldmxcsr (%ecx) # encoding: [0x0f,0xae,0x11]
; X86-SSE-NEXT:    popl %eax # encoding: [0x58]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX-LABEL: test_MM_SET_EXCEPTION_MASK:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    pushl %eax # encoding: [0x50]
; X86-AVX-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x08]
; X86-AVX-NEXT:    movl %esp, %ecx # encoding: [0x89,0xe1]
; X86-AVX-NEXT:    vstmxcsr (%ecx) # encoding: [0xc5,0xf8,0xae,0x19]
; X86-AVX-NEXT:    movl (%esp), %edx # encoding: [0x8b,0x14,0x24]
; X86-AVX-NEXT:    andl $-8065, %edx # encoding: [0x81,0xe2,0x7f,0xe0,0xff,0xff]
; X86-AVX-NEXT:    # imm = 0xE07F
; X86-AVX-NEXT:    orl %eax, %edx # encoding: [0x09,0xc2]
; X86-AVX-NEXT:    movl %edx, (%esp) # encoding: [0x89,0x14,0x24]
; X86-AVX-NEXT:    vldmxcsr (%ecx) # encoding: [0xc5,0xf8,0xae,0x11]
; X86-AVX-NEXT:    popl %eax # encoding: [0x58]
; X86-AVX-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_MM_SET_EXCEPTION_MASK:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-SSE-NEXT:    stmxcsr (%rax) # encoding: [0x0f,0xae,0x18]
; X64-SSE-NEXT:    movl -{{[0-9]+}}(%rsp), %ecx # encoding: [0x8b,0x4c,0x24,0xfc]
; X64-SSE-NEXT:    andl $-8065, %ecx # encoding: [0x81,0xe1,0x7f,0xe0,0xff,0xff]
; X64-SSE-NEXT:    # imm = 0xE07F
; X64-SSE-NEXT:    orl %edi, %ecx # encoding: [0x09,0xf9]
; X64-SSE-NEXT:    movl %ecx, -{{[0-9]+}}(%rsp) # encoding: [0x89,0x4c,0x24,0xfc]
; X64-SSE-NEXT:    ldmxcsr (%rax) # encoding: [0x0f,0xae,0x10]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX-LABEL: test_MM_SET_EXCEPTION_MASK:
; X64-AVX:       # %bb.0:
; X64-AVX-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-AVX-NEXT:    vstmxcsr (%rax) # encoding: [0xc5,0xf8,0xae,0x18]
; X64-AVX-NEXT:    movl -{{[0-9]+}}(%rsp), %ecx # encoding: [0x8b,0x4c,0x24,0xfc]
; X64-AVX-NEXT:    andl $-8065, %ecx # encoding: [0x81,0xe1,0x7f,0xe0,0xff,0xff]
; X64-AVX-NEXT:    # imm = 0xE07F
; X64-AVX-NEXT:    orl %edi, %ecx # encoding: [0x09,0xf9]
; X64-AVX-NEXT:    movl %ecx, -{{[0-9]+}}(%rsp) # encoding: [0x89,0x4c,0x24,0xfc]
; X64-AVX-NEXT:    vldmxcsr (%rax) # encoding: [0xc5,0xf8,0xae,0x10]
; X64-AVX-NEXT:    retq # encoding: [0xc3]
  %1 = alloca i32, align 4
  %2 = bitcast i32* %1 to i8*
  call void @llvm.x86.sse.stmxcsr(i8* %2)
  %3 = load i32, i32* %1
  %4 = and i32 %3, -8065
  %5 = or i32 %4, %a0
  store i32 %5, i32* %1
  call void @llvm.x86.sse.ldmxcsr(i8* %2)
  ret void
}
declare void @llvm.x86.sse.ldmxcsr(i8*) nounwind readnone

define void @test_MM_SET_EXCEPTION_STATE(i32 %a0) nounwind {
; X86-SSE-LABEL: test_MM_SET_EXCEPTION_STATE:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    pushl %eax # encoding: [0x50]
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x08]
; X86-SSE-NEXT:    movl %esp, %ecx # encoding: [0x89,0xe1]
; X86-SSE-NEXT:    stmxcsr (%ecx) # encoding: [0x0f,0xae,0x19]
; X86-SSE-NEXT:    movl (%esp), %edx # encoding: [0x8b,0x14,0x24]
; X86-SSE-NEXT:    andl $-64, %edx # encoding: [0x83,0xe2,0xc0]
; X86-SSE-NEXT:    orl %eax, %edx # encoding: [0x09,0xc2]
; X86-SSE-NEXT:    movl %edx, (%esp) # encoding: [0x89,0x14,0x24]
; X86-SSE-NEXT:    ldmxcsr (%ecx) # encoding: [0x0f,0xae,0x11]
; X86-SSE-NEXT:    popl %eax # encoding: [0x58]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX-LABEL: test_MM_SET_EXCEPTION_STATE:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    pushl %eax # encoding: [0x50]
; X86-AVX-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x08]
; X86-AVX-NEXT:    movl %esp, %ecx # encoding: [0x89,0xe1]
; X86-AVX-NEXT:    vstmxcsr (%ecx) # encoding: [0xc5,0xf8,0xae,0x19]
; X86-AVX-NEXT:    movl (%esp), %edx # encoding: [0x8b,0x14,0x24]
; X86-AVX-NEXT:    andl $-64, %edx # encoding: [0x83,0xe2,0xc0]
; X86-AVX-NEXT:    orl %eax, %edx # encoding: [0x09,0xc2]
; X86-AVX-NEXT:    movl %edx, (%esp) # encoding: [0x89,0x14,0x24]
; X86-AVX-NEXT:    vldmxcsr (%ecx) # encoding: [0xc5,0xf8,0xae,0x11]
; X86-AVX-NEXT:    popl %eax # encoding: [0x58]
; X86-AVX-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_MM_SET_EXCEPTION_STATE:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-SSE-NEXT:    stmxcsr (%rax) # encoding: [0x0f,0xae,0x18]
; X64-SSE-NEXT:    movl -{{[0-9]+}}(%rsp), %ecx # encoding: [0x8b,0x4c,0x24,0xfc]
; X64-SSE-NEXT:    andl $-64, %ecx # encoding: [0x83,0xe1,0xc0]
; X64-SSE-NEXT:    orl %edi, %ecx # encoding: [0x09,0xf9]
; X64-SSE-NEXT:    movl %ecx, -{{[0-9]+}}(%rsp) # encoding: [0x89,0x4c,0x24,0xfc]
; X64-SSE-NEXT:    ldmxcsr (%rax) # encoding: [0x0f,0xae,0x10]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX-LABEL: test_MM_SET_EXCEPTION_STATE:
; X64-AVX:       # %bb.0:
; X64-AVX-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-AVX-NEXT:    vstmxcsr (%rax) # encoding: [0xc5,0xf8,0xae,0x18]
; X64-AVX-NEXT:    movl -{{[0-9]+}}(%rsp), %ecx # encoding: [0x8b,0x4c,0x24,0xfc]
; X64-AVX-NEXT:    andl $-64, %ecx # encoding: [0x83,0xe1,0xc0]
; X64-AVX-NEXT:    orl %edi, %ecx # encoding: [0x09,0xf9]
; X64-AVX-NEXT:    movl %ecx, -{{[0-9]+}}(%rsp) # encoding: [0x89,0x4c,0x24,0xfc]
; X64-AVX-NEXT:    vldmxcsr (%rax) # encoding: [0xc5,0xf8,0xae,0x10]
; X64-AVX-NEXT:    retq # encoding: [0xc3]
  %1 = alloca i32, align 4
  %2 = bitcast i32* %1 to i8*
  call void @llvm.x86.sse.stmxcsr(i8* %2)
  %3 = load i32, i32* %1
  %4 = and i32 %3, -64
  %5 = or i32 %4, %a0
  store i32 %5, i32* %1
  call void @llvm.x86.sse.ldmxcsr(i8* %2)
  ret void
}

define void @test_MM_SET_FLUSH_ZERO_MODE(i32 %a0) nounwind {
; X86-SSE-LABEL: test_MM_SET_FLUSH_ZERO_MODE:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    pushl %eax # encoding: [0x50]
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x08]
; X86-SSE-NEXT:    movl %esp, %ecx # encoding: [0x89,0xe1]
; X86-SSE-NEXT:    stmxcsr (%ecx) # encoding: [0x0f,0xae,0x19]
; X86-SSE-NEXT:    movl (%esp), %edx # encoding: [0x8b,0x14,0x24]
; X86-SSE-NEXT:    andl $-32769, %edx # encoding: [0x81,0xe2,0xff,0x7f,0xff,0xff]
; X86-SSE-NEXT:    # imm = 0xFFFF7FFF
; X86-SSE-NEXT:    orl %eax, %edx # encoding: [0x09,0xc2]
; X86-SSE-NEXT:    movl %edx, (%esp) # encoding: [0x89,0x14,0x24]
; X86-SSE-NEXT:    ldmxcsr (%ecx) # encoding: [0x0f,0xae,0x11]
; X86-SSE-NEXT:    popl %eax # encoding: [0x58]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX-LABEL: test_MM_SET_FLUSH_ZERO_MODE:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    pushl %eax # encoding: [0x50]
; X86-AVX-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x08]
; X86-AVX-NEXT:    movl %esp, %ecx # encoding: [0x89,0xe1]
; X86-AVX-NEXT:    vstmxcsr (%ecx) # encoding: [0xc5,0xf8,0xae,0x19]
; X86-AVX-NEXT:    movl (%esp), %edx # encoding: [0x8b,0x14,0x24]
; X86-AVX-NEXT:    andl $-32769, %edx # encoding: [0x81,0xe2,0xff,0x7f,0xff,0xff]
; X86-AVX-NEXT:    # imm = 0xFFFF7FFF
; X86-AVX-NEXT:    orl %eax, %edx # encoding: [0x09,0xc2]
; X86-AVX-NEXT:    movl %edx, (%esp) # encoding: [0x89,0x14,0x24]
; X86-AVX-NEXT:    vldmxcsr (%ecx) # encoding: [0xc5,0xf8,0xae,0x11]
; X86-AVX-NEXT:    popl %eax # encoding: [0x58]
; X86-AVX-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_MM_SET_FLUSH_ZERO_MODE:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-SSE-NEXT:    stmxcsr (%rax) # encoding: [0x0f,0xae,0x18]
; X64-SSE-NEXT:    movl -{{[0-9]+}}(%rsp), %ecx # encoding: [0x8b,0x4c,0x24,0xfc]
; X64-SSE-NEXT:    andl $-32769, %ecx # encoding: [0x81,0xe1,0xff,0x7f,0xff,0xff]
; X64-SSE-NEXT:    # imm = 0xFFFF7FFF
; X64-SSE-NEXT:    orl %edi, %ecx # encoding: [0x09,0xf9]
; X64-SSE-NEXT:    movl %ecx, -{{[0-9]+}}(%rsp) # encoding: [0x89,0x4c,0x24,0xfc]
; X64-SSE-NEXT:    ldmxcsr (%rax) # encoding: [0x0f,0xae,0x10]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX-LABEL: test_MM_SET_FLUSH_ZERO_MODE:
; X64-AVX:       # %bb.0:
; X64-AVX-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-AVX-NEXT:    vstmxcsr (%rax) # encoding: [0xc5,0xf8,0xae,0x18]
; X64-AVX-NEXT:    movl -{{[0-9]+}}(%rsp), %ecx # encoding: [0x8b,0x4c,0x24,0xfc]
; X64-AVX-NEXT:    andl $-32769, %ecx # encoding: [0x81,0xe1,0xff,0x7f,0xff,0xff]
; X64-AVX-NEXT:    # imm = 0xFFFF7FFF
; X64-AVX-NEXT:    orl %edi, %ecx # encoding: [0x09,0xf9]
; X64-AVX-NEXT:    movl %ecx, -{{[0-9]+}}(%rsp) # encoding: [0x89,0x4c,0x24,0xfc]
; X64-AVX-NEXT:    vldmxcsr (%rax) # encoding: [0xc5,0xf8,0xae,0x10]
; X64-AVX-NEXT:    retq # encoding: [0xc3]
  %1 = alloca i32, align 4
  %2 = bitcast i32* %1 to i8*
  call void @llvm.x86.sse.stmxcsr(i8* %2)
  %3 = load i32, i32* %1
  %4 = and i32 %3, -32769
  %5 = or i32 %4, %a0
  store i32 %5, i32* %1
  call void @llvm.x86.sse.ldmxcsr(i8* %2)
  ret void
}

define <4 x float> @test_mm_set_ps(float %a0, float %a1, float %a2, float %a3) nounwind {
; X86-SSE-LABEL: test_mm_set_ps:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movss {{[0-9]+}}(%esp), %xmm0 # encoding: [0xf3,0x0f,0x10,0x44,0x24,0x10]
; X86-SSE-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    movss {{[0-9]+}}(%esp), %xmm1 # encoding: [0xf3,0x0f,0x10,0x4c,0x24,0x0c]
; X86-SSE-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    unpcklps %xmm1, %xmm0 # encoding: [0x0f,0x14,0xc1]
; X86-SSE-NEXT:    # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; X86-SSE-NEXT:    movss {{[0-9]+}}(%esp), %xmm1 # encoding: [0xf3,0x0f,0x10,0x4c,0x24,0x08]
; X86-SSE-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    movss {{[0-9]+}}(%esp), %xmm2 # encoding: [0xf3,0x0f,0x10,0x54,0x24,0x04]
; X86-SSE-NEXT:    # xmm2 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    unpcklps %xmm2, %xmm1 # encoding: [0x0f,0x14,0xca]
; X86-SSE-NEXT:    # xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
; X86-SSE-NEXT:    movlhps %xmm1, %xmm0 # encoding: [0x0f,0x16,0xc1]
; X86-SSE-NEXT:    # xmm0 = xmm0[0],xmm1[0]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_set_ps:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm0 # encoding: [0xc5,0xfa,0x10,0x44,0x24,0x10]
; X86-AVX1-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-AVX1-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm1 # encoding: [0xc5,0xfa,0x10,0x4c,0x24,0x0c]
; X86-AVX1-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X86-AVX1-NEXT:    vinsertps $16, %xmm1, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x21,0xc1,0x10]
; X86-AVX1-NEXT:    # xmm0 = xmm0[0],xmm1[0],xmm0[2,3]
; X86-AVX1-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm1 # encoding: [0xc5,0xfa,0x10,0x4c,0x24,0x08]
; X86-AVX1-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X86-AVX1-NEXT:    vinsertps $32, %xmm1, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x21,0xc1,0x20]
; X86-AVX1-NEXT:    # xmm0 = xmm0[0,1],xmm1[0],xmm0[3]
; X86-AVX1-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm1 # encoding: [0xc5,0xfa,0x10,0x4c,0x24,0x04]
; X86-AVX1-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X86-AVX1-NEXT:    vinsertps $48, %xmm1, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x21,0xc1,0x30]
; X86-AVX1-NEXT:    # xmm0 = xmm0[0,1,2],xmm1[0]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_set_ps:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x44,0x24,0x10]
; X86-AVX512-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-AVX512-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x4c,0x24,0x0c]
; X86-AVX512-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X86-AVX512-NEXT:    vinsertps $16, %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x79,0x21,0xc1,0x10]
; X86-AVX512-NEXT:    # xmm0 = xmm0[0],xmm1[0],xmm0[2,3]
; X86-AVX512-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x4c,0x24,0x08]
; X86-AVX512-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X86-AVX512-NEXT:    vinsertps $32, %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x79,0x21,0xc1,0x20]
; X86-AVX512-NEXT:    # xmm0 = xmm0[0,1],xmm1[0],xmm0[3]
; X86-AVX512-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x4c,0x24,0x04]
; X86-AVX512-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X86-AVX512-NEXT:    vinsertps $48, %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x79,0x21,0xc1,0x30]
; X86-AVX512-NEXT:    # xmm0 = xmm0[0,1,2],xmm1[0]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_set_ps:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    unpcklps %xmm0, %xmm1 # encoding: [0x0f,0x14,0xc8]
; X64-SSE-NEXT:    # xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; X64-SSE-NEXT:    unpcklps %xmm2, %xmm3 # encoding: [0x0f,0x14,0xda]
; X64-SSE-NEXT:    # xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
; X64-SSE-NEXT:    movlhps %xmm1, %xmm3 # encoding: [0x0f,0x16,0xd9]
; X64-SSE-NEXT:    # xmm3 = xmm3[0],xmm1[0]
; X64-SSE-NEXT:    movaps %xmm3, %xmm0 # encoding: [0x0f,0x28,0xc3]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_set_ps:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vinsertps $16, %xmm2, %xmm3, %xmm2 # encoding: [0xc4,0xe3,0x61,0x21,0xd2,0x10]
; X64-AVX1-NEXT:    # xmm2 = xmm3[0],xmm2[0],xmm3[2,3]
; X64-AVX1-NEXT:    vinsertps $32, %xmm1, %xmm2, %xmm1 # encoding: [0xc4,0xe3,0x69,0x21,0xc9,0x20]
; X64-AVX1-NEXT:    # xmm1 = xmm2[0,1],xmm1[0],xmm2[3]
; X64-AVX1-NEXT:    vinsertps $48, %xmm0, %xmm1, %xmm0 # encoding: [0xc4,0xe3,0x71,0x21,0xc0,0x30]
; X64-AVX1-NEXT:    # xmm0 = xmm1[0,1,2],xmm0[0]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_set_ps:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vinsertps $16, %xmm2, %xmm3, %xmm2 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x61,0x21,0xd2,0x10]
; X64-AVX512-NEXT:    # xmm2 = xmm3[0],xmm2[0],xmm3[2,3]
; X64-AVX512-NEXT:    vinsertps $32, %xmm1, %xmm2, %xmm1 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x69,0x21,0xc9,0x20]
; X64-AVX512-NEXT:    # xmm1 = xmm2[0,1],xmm1[0],xmm2[3]
; X64-AVX512-NEXT:    vinsertps $48, %xmm0, %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x71,0x21,0xc0,0x30]
; X64-AVX512-NEXT:    # xmm0 = xmm1[0,1,2],xmm0[0]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %res0  = insertelement <4 x float> undef, float %a3, i32 0
  %res1  = insertelement <4 x float> %res0, float %a2, i32 1
  %res2  = insertelement <4 x float> %res1, float %a1, i32 2
  %res3  = insertelement <4 x float> %res2, float %a0, i32 3
  ret <4 x float> %res3
}

define <4 x float> @test_mm_set_ps1(float %a0) nounwind {
; X86-SSE-LABEL: test_mm_set_ps1:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movss {{[0-9]+}}(%esp), %xmm0 # encoding: [0xf3,0x0f,0x10,0x44,0x24,0x04]
; X86-SSE-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    shufps $0, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x00]
; X86-SSE-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_set_ps1:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm0 # encoding: [0xc5,0xfa,0x10,0x44,0x24,0x04]
; X86-AVX1-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-AVX1-NEXT:    vpermilps $0, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x04,0xc0,0x00]
; X86-AVX1-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_set_ps1:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x44,0x24,0x04]
; X86-AVX512-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-AVX512-NEXT:    vbroadcastss %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x18,0xc0]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_set_ps1:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    shufps $0, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x00]
; X64-SSE-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_set_ps1:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vpermilps $0, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x04,0xc0,0x00]
; X64-AVX1-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_set_ps1:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vbroadcastss %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x18,0xc0]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %res0  = insertelement <4 x float> undef, float %a0, i32 0
  %res1  = insertelement <4 x float> %res0, float %a0, i32 1
  %res2  = insertelement <4 x float> %res1, float %a0, i32 2
  %res3  = insertelement <4 x float> %res2, float %a0, i32 3
  ret <4 x float> %res3
}

define void @test_MM_SET_ROUNDING_MODE(i32 %a0) nounwind {
; X86-SSE-LABEL: test_MM_SET_ROUNDING_MODE:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    pushl %eax # encoding: [0x50]
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x08]
; X86-SSE-NEXT:    movl %esp, %ecx # encoding: [0x89,0xe1]
; X86-SSE-NEXT:    stmxcsr (%ecx) # encoding: [0x0f,0xae,0x19]
; X86-SSE-NEXT:    movl (%esp), %edx # encoding: [0x8b,0x14,0x24]
; X86-SSE-NEXT:    andl $-24577, %edx # encoding: [0x81,0xe2,0xff,0x9f,0xff,0xff]
; X86-SSE-NEXT:    # imm = 0x9FFF
; X86-SSE-NEXT:    orl %eax, %edx # encoding: [0x09,0xc2]
; X86-SSE-NEXT:    movl %edx, (%esp) # encoding: [0x89,0x14,0x24]
; X86-SSE-NEXT:    ldmxcsr (%ecx) # encoding: [0x0f,0xae,0x11]
; X86-SSE-NEXT:    popl %eax # encoding: [0x58]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX-LABEL: test_MM_SET_ROUNDING_MODE:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    pushl %eax # encoding: [0x50]
; X86-AVX-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x08]
; X86-AVX-NEXT:    movl %esp, %ecx # encoding: [0x89,0xe1]
; X86-AVX-NEXT:    vstmxcsr (%ecx) # encoding: [0xc5,0xf8,0xae,0x19]
; X86-AVX-NEXT:    movl (%esp), %edx # encoding: [0x8b,0x14,0x24]
; X86-AVX-NEXT:    andl $-24577, %edx # encoding: [0x81,0xe2,0xff,0x9f,0xff,0xff]
; X86-AVX-NEXT:    # imm = 0x9FFF
; X86-AVX-NEXT:    orl %eax, %edx # encoding: [0x09,0xc2]
; X86-AVX-NEXT:    movl %edx, (%esp) # encoding: [0x89,0x14,0x24]
; X86-AVX-NEXT:    vldmxcsr (%ecx) # encoding: [0xc5,0xf8,0xae,0x11]
; X86-AVX-NEXT:    popl %eax # encoding: [0x58]
; X86-AVX-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_MM_SET_ROUNDING_MODE:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-SSE-NEXT:    stmxcsr (%rax) # encoding: [0x0f,0xae,0x18]
; X64-SSE-NEXT:    movl -{{[0-9]+}}(%rsp), %ecx # encoding: [0x8b,0x4c,0x24,0xfc]
; X64-SSE-NEXT:    andl $-24577, %ecx # encoding: [0x81,0xe1,0xff,0x9f,0xff,0xff]
; X64-SSE-NEXT:    # imm = 0x9FFF
; X64-SSE-NEXT:    orl %edi, %ecx # encoding: [0x09,0xf9]
; X64-SSE-NEXT:    movl %ecx, -{{[0-9]+}}(%rsp) # encoding: [0x89,0x4c,0x24,0xfc]
; X64-SSE-NEXT:    ldmxcsr (%rax) # encoding: [0x0f,0xae,0x10]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX-LABEL: test_MM_SET_ROUNDING_MODE:
; X64-AVX:       # %bb.0:
; X64-AVX-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-AVX-NEXT:    vstmxcsr (%rax) # encoding: [0xc5,0xf8,0xae,0x18]
; X64-AVX-NEXT:    movl -{{[0-9]+}}(%rsp), %ecx # encoding: [0x8b,0x4c,0x24,0xfc]
; X64-AVX-NEXT:    andl $-24577, %ecx # encoding: [0x81,0xe1,0xff,0x9f,0xff,0xff]
; X64-AVX-NEXT:    # imm = 0x9FFF
; X64-AVX-NEXT:    orl %edi, %ecx # encoding: [0x09,0xf9]
; X64-AVX-NEXT:    movl %ecx, -{{[0-9]+}}(%rsp) # encoding: [0x89,0x4c,0x24,0xfc]
; X64-AVX-NEXT:    vldmxcsr (%rax) # encoding: [0xc5,0xf8,0xae,0x10]
; X64-AVX-NEXT:    retq # encoding: [0xc3]
  %1 = alloca i32, align 4
  %2 = bitcast i32* %1 to i8*
  call void @llvm.x86.sse.stmxcsr(i8* %2)
  %3 = load i32, i32* %1
  %4 = and i32 %3, -24577
  %5 = or i32 %4, %a0
  store i32 %5, i32* %1
  call void @llvm.x86.sse.ldmxcsr(i8* %2)
  ret void
}

define <4 x float> @test_mm_set_ss(float %a0) nounwind {
; X86-SSE-LABEL: test_mm_set_ss:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movss {{[0-9]+}}(%esp), %xmm1 # encoding: [0xf3,0x0f,0x10,0x4c,0x24,0x04]
; X86-SSE-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    xorps %xmm0, %xmm0 # encoding: [0x0f,0x57,0xc0]
; X86-SSE-NEXT:    movss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0x10,0xc1]
; X86-SSE-NEXT:    # xmm0 = xmm1[0],xmm0[1,2,3]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_set_ss:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm0 # encoding: [0xc5,0xfa,0x10,0x44,0x24,0x04]
; X86-AVX1-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-AVX1-NEXT:    vxorps %xmm1, %xmm1, %xmm1 # encoding: [0xc5,0xf0,0x57,0xc9]
; X86-AVX1-NEXT:    vblendps $1, %xmm0, %xmm1, %xmm0 # encoding: [0xc4,0xe3,0x71,0x0c,0xc0,0x01]
; X86-AVX1-NEXT:    # xmm0 = xmm0[0],xmm1[1,2,3]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_set_ss:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x44,0x24,0x04]
; X86-AVX512-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-AVX512-NEXT:    vxorps %xmm1, %xmm1, %xmm1 # encoding: [0xc5,0xf0,0x57,0xc9]
; X86-AVX512-NEXT:    vblendps $1, %xmm0, %xmm1, %xmm0 # encoding: [0xc4,0xe3,0x71,0x0c,0xc0,0x01]
; X86-AVX512-NEXT:    # xmm0 = xmm0[0],xmm1[1,2,3]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_set_ss:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    xorps %xmm1, %xmm1 # encoding: [0x0f,0x57,0xc9]
; X64-SSE-NEXT:    movss %xmm0, %xmm1 # encoding: [0xf3,0x0f,0x10,0xc8]
; X64-SSE-NEXT:    # xmm1 = xmm0[0],xmm1[1,2,3]
; X64-SSE-NEXT:    movaps %xmm1, %xmm0 # encoding: [0x0f,0x28,0xc1]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX-LABEL: test_mm_set_ss:
; X64-AVX:       # %bb.0:
; X64-AVX-NEXT:    vxorps %xmm1, %xmm1, %xmm1 # encoding: [0xc5,0xf0,0x57,0xc9]
; X64-AVX-NEXT:    vblendps $1, %xmm0, %xmm1, %xmm0 # encoding: [0xc4,0xe3,0x71,0x0c,0xc0,0x01]
; X64-AVX-NEXT:    # xmm0 = xmm0[0],xmm1[1,2,3]
; X64-AVX-NEXT:    retq # encoding: [0xc3]
  %res0  = insertelement <4 x float> undef, float %a0, i32 0
  %res1  = insertelement <4 x float> %res0, float 0.0, i32 1
  %res2  = insertelement <4 x float> %res1, float 0.0, i32 2
  %res3  = insertelement <4 x float> %res2, float 0.0, i32 3
  ret <4 x float> %res3
}

define <4 x float> @test_mm_set1_ps(float %a0) nounwind {
; X86-SSE-LABEL: test_mm_set1_ps:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movss {{[0-9]+}}(%esp), %xmm0 # encoding: [0xf3,0x0f,0x10,0x44,0x24,0x04]
; X86-SSE-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    shufps $0, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x00]
; X86-SSE-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_set1_ps:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm0 # encoding: [0xc5,0xfa,0x10,0x44,0x24,0x04]
; X86-AVX1-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-AVX1-NEXT:    vpermilps $0, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x04,0xc0,0x00]
; X86-AVX1-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_set1_ps:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x44,0x24,0x04]
; X86-AVX512-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-AVX512-NEXT:    vbroadcastss %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x18,0xc0]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_set1_ps:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    shufps $0, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x00]
; X64-SSE-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_set1_ps:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vpermilps $0, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x04,0xc0,0x00]
; X64-AVX1-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_set1_ps:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vbroadcastss %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x18,0xc0]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %res0  = insertelement <4 x float> undef, float %a0, i32 0
  %res1  = insertelement <4 x float> %res0, float %a0, i32 1
  %res2  = insertelement <4 x float> %res1, float %a0, i32 2
  %res3  = insertelement <4 x float> %res2, float %a0, i32 3
  ret <4 x float> %res3
}

define void @test_mm_setcsr(i32 %a0) nounwind {
; X86-SSE-LABEL: test_mm_setcsr:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    leal {{[0-9]+}}(%esp), %eax # encoding: [0x8d,0x44,0x24,0x04]
; X86-SSE-NEXT:    ldmxcsr (%eax) # encoding: [0x0f,0xae,0x10]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX-LABEL: test_mm_setcsr:
; X86-AVX:       # %bb.0:
; X86-AVX-NEXT:    leal {{[0-9]+}}(%esp), %eax # encoding: [0x8d,0x44,0x24,0x04]
; X86-AVX-NEXT:    vldmxcsr (%eax) # encoding: [0xc5,0xf8,0xae,0x10]
; X86-AVX-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_setcsr:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movl %edi, -{{[0-9]+}}(%rsp) # encoding: [0x89,0x7c,0x24,0xfc]
; X64-SSE-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-SSE-NEXT:    ldmxcsr (%rax) # encoding: [0x0f,0xae,0x10]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX-LABEL: test_mm_setcsr:
; X64-AVX:       # %bb.0:
; X64-AVX-NEXT:    movl %edi, -{{[0-9]+}}(%rsp) # encoding: [0x89,0x7c,0x24,0xfc]
; X64-AVX-NEXT:    leaq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8d,0x44,0x24,0xfc]
; X64-AVX-NEXT:    vldmxcsr (%rax) # encoding: [0xc5,0xf8,0xae,0x10]
; X64-AVX-NEXT:    retq # encoding: [0xc3]
  %st = alloca i32, align 4
  store i32 %a0, i32* %st, align 4
  %bc = bitcast i32* %st to i8*
  call void @llvm.x86.sse.ldmxcsr(i8* %bc)
  ret void
}

define <4 x float> @test_mm_setr_ps(float %a0, float %a1, float %a2, float %a3) nounwind {
; X86-SSE-LABEL: test_mm_setr_ps:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movss {{[0-9]+}}(%esp), %xmm0 # encoding: [0xf3,0x0f,0x10,0x44,0x24,0x10]
; X86-SSE-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    movss {{[0-9]+}}(%esp), %xmm1 # encoding: [0xf3,0x0f,0x10,0x4c,0x24,0x0c]
; X86-SSE-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    unpcklps %xmm0, %xmm1 # encoding: [0x0f,0x14,0xc8]
; X86-SSE-NEXT:    # xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
; X86-SSE-NEXT:    movss {{[0-9]+}}(%esp), %xmm2 # encoding: [0xf3,0x0f,0x10,0x54,0x24,0x08]
; X86-SSE-NEXT:    # xmm2 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    movss {{[0-9]+}}(%esp), %xmm0 # encoding: [0xf3,0x0f,0x10,0x44,0x24,0x04]
; X86-SSE-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    unpcklps %xmm2, %xmm0 # encoding: [0x0f,0x14,0xc2]
; X86-SSE-NEXT:    # xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
; X86-SSE-NEXT:    movlhps %xmm1, %xmm0 # encoding: [0x0f,0x16,0xc1]
; X86-SSE-NEXT:    # xmm0 = xmm0[0],xmm1[0]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_setr_ps:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm0 # encoding: [0xc5,0xfa,0x10,0x44,0x24,0x10]
; X86-AVX1-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-AVX1-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm1 # encoding: [0xc5,0xfa,0x10,0x4c,0x24,0x0c]
; X86-AVX1-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X86-AVX1-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm2 # encoding: [0xc5,0xfa,0x10,0x54,0x24,0x08]
; X86-AVX1-NEXT:    # xmm2 = mem[0],zero,zero,zero
; X86-AVX1-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm3 # encoding: [0xc5,0xfa,0x10,0x5c,0x24,0x04]
; X86-AVX1-NEXT:    # xmm3 = mem[0],zero,zero,zero
; X86-AVX1-NEXT:    vinsertps $16, %xmm2, %xmm3, %xmm2 # encoding: [0xc4,0xe3,0x61,0x21,0xd2,0x10]
; X86-AVX1-NEXT:    # xmm2 = xmm3[0],xmm2[0],xmm3[2,3]
; X86-AVX1-NEXT:    vinsertps $32, %xmm1, %xmm2, %xmm1 # encoding: [0xc4,0xe3,0x69,0x21,0xc9,0x20]
; X86-AVX1-NEXT:    # xmm1 = xmm2[0,1],xmm1[0],xmm2[3]
; X86-AVX1-NEXT:    vinsertps $48, %xmm0, %xmm1, %xmm0 # encoding: [0xc4,0xe3,0x71,0x21,0xc0,0x30]
; X86-AVX1-NEXT:    # xmm0 = xmm1[0,1,2],xmm0[0]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_setr_ps:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x44,0x24,0x10]
; X86-AVX512-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-AVX512-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x4c,0x24,0x0c]
; X86-AVX512-NEXT:    # xmm1 = mem[0],zero,zero,zero
; X86-AVX512-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm2 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x54,0x24,0x08]
; X86-AVX512-NEXT:    # xmm2 = mem[0],zero,zero,zero
; X86-AVX512-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x5c,0x24,0x04]
; X86-AVX512-NEXT:    # xmm3 = mem[0],zero,zero,zero
; X86-AVX512-NEXT:    vinsertps $16, %xmm2, %xmm3, %xmm2 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x61,0x21,0xd2,0x10]
; X86-AVX512-NEXT:    # xmm2 = xmm3[0],xmm2[0],xmm3[2,3]
; X86-AVX512-NEXT:    vinsertps $32, %xmm1, %xmm2, %xmm1 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x69,0x21,0xc9,0x20]
; X86-AVX512-NEXT:    # xmm1 = xmm2[0,1],xmm1[0],xmm2[3]
; X86-AVX512-NEXT:    vinsertps $48, %xmm0, %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x71,0x21,0xc0,0x30]
; X86-AVX512-NEXT:    # xmm0 = xmm1[0,1,2],xmm0[0]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_setr_ps:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    unpcklps %xmm3, %xmm2 # encoding: [0x0f,0x14,0xd3]
; X64-SSE-NEXT:    # xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; X64-SSE-NEXT:    unpcklps %xmm1, %xmm0 # encoding: [0x0f,0x14,0xc1]
; X64-SSE-NEXT:    # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; X64-SSE-NEXT:    movlhps %xmm2, %xmm0 # encoding: [0x0f,0x16,0xc2]
; X64-SSE-NEXT:    # xmm0 = xmm0[0],xmm2[0]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_setr_ps:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vinsertps $16, %xmm1, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x21,0xc1,0x10]
; X64-AVX1-NEXT:    # xmm0 = xmm0[0],xmm1[0],xmm0[2,3]
; X64-AVX1-NEXT:    vinsertps $32, %xmm2, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x21,0xc2,0x20]
; X64-AVX1-NEXT:    # xmm0 = xmm0[0,1],xmm2[0],xmm0[3]
; X64-AVX1-NEXT:    vinsertps $48, %xmm3, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x21,0xc3,0x30]
; X64-AVX1-NEXT:    # xmm0 = xmm0[0,1,2],xmm3[0]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_setr_ps:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vinsertps $16, %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x79,0x21,0xc1,0x10]
; X64-AVX512-NEXT:    # xmm0 = xmm0[0],xmm1[0],xmm0[2,3]
; X64-AVX512-NEXT:    vinsertps $32, %xmm2, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x79,0x21,0xc2,0x20]
; X64-AVX512-NEXT:    # xmm0 = xmm0[0,1],xmm2[0],xmm0[3]
; X64-AVX512-NEXT:    vinsertps $48, %xmm3, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x79,0x21,0xc3,0x30]
; X64-AVX512-NEXT:    # xmm0 = xmm0[0,1,2],xmm3[0]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %res0  = insertelement <4 x float> undef, float %a0, i32 0
  %res1  = insertelement <4 x float> %res0, float %a1, i32 1
  %res2  = insertelement <4 x float> %res1, float %a2, i32 2
  %res3  = insertelement <4 x float> %res2, float %a3, i32 3
  ret <4 x float> %res3
}

define <4 x float> @test_mm_setzero_ps() {
; SSE-LABEL: test_mm_setzero_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    xorps %xmm0, %xmm0 # encoding: [0x0f,0x57,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_setzero_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vxorps %xmm0, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x57,0xc0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_setzero_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vxorps %xmm0, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x57,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  ret <4 x float> zeroinitializer
}

define void @test_mm_sfence() nounwind {
; CHECK-LABEL: test_mm_sfence:
; CHECK:       # %bb.0:
; CHECK-NEXT:    sfence # encoding: [0x0f,0xae,0xf8]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  call void @llvm.x86.sse.sfence()
  ret void
}
declare void @llvm.x86.sse.sfence() nounwind readnone

define <4 x float> @test_mm_shuffle_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_shuffle_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    shufps $0, %xmm1, %xmm0 # encoding: [0x0f,0xc6,0xc1,0x00]
; SSE-NEXT:    # xmm0 = xmm0[0,0],xmm1[0,0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_shuffle_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vshufps $0, %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0xc6,0xc1,0x00]
; AVX1-NEXT:    # xmm0 = xmm0[0,0],xmm1[0,0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_shuffle_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vshufps $0, %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0xc6,0xc1,0x00]
; AVX512-NEXT:    # xmm0 = xmm0[0,0],xmm1[0,0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = shufflevector <4 x float> %a0, <4 x float> %a1, <4 x i32> <i32 0, i32 0, i32 4, i32 4>
  ret <4 x float> %res
}

define <4 x float> @test_mm_sqrt_ps(<4 x float> %a0) {
; SSE-LABEL: test_mm_sqrt_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    sqrtps %xmm0, %xmm0 # encoding: [0x0f,0x51,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_sqrt_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vsqrtps %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x51,0xc0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_sqrt_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vsqrtps %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x51,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <4 x float> @llvm.sqrt.v4f32(<4 x float> %a0)
  ret <4 x float> %res
}
declare <4 x float> @llvm.sqrt.v4f32(<4 x float>) nounwind readnone

define <4 x float> @test_mm_sqrt_ss(<4 x float> %a0) {
; SSE-LABEL: test_mm_sqrt_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    sqrtss %xmm0, %xmm0 # encoding: [0xf3,0x0f,0x51,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_sqrt_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vsqrtss %xmm0, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0x51,0xc0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_sqrt_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vsqrtss %xmm0, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x51,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %ext = extractelement <4 x float> %a0, i32 0
  %sqrt = call float @llvm.sqrt.f32(float %ext)
  %ins = insertelement <4 x float> %a0, float %sqrt, i32 0
  ret <4 x float> %ins
}
declare float @llvm.sqrt.f32(float) nounwind readnone

define float @test_mm_sqrt_ss_scalar(float %a0) {
; X86-SSE-LABEL: test_mm_sqrt_ss_scalar:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    pushl %eax # encoding: [0x50]
; X86-SSE-NEXT:    .cfi_def_cfa_offset 8
; X86-SSE-NEXT:    movss {{[0-9]+}}(%esp), %xmm0 # encoding: [0xf3,0x0f,0x10,0x44,0x24,0x08]
; X86-SSE-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-SSE-NEXT:    sqrtss %xmm0, %xmm0 # encoding: [0xf3,0x0f,0x51,0xc0]
; X86-SSE-NEXT:    movss %xmm0, (%esp) # encoding: [0xf3,0x0f,0x11,0x04,0x24]
; X86-SSE-NEXT:    flds (%esp) # encoding: [0xd9,0x04,0x24]
; X86-SSE-NEXT:    popl %eax # encoding: [0x58]
; X86-SSE-NEXT:    .cfi_def_cfa_offset 4
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_sqrt_ss_scalar:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    pushl %eax # encoding: [0x50]
; X86-AVX1-NEXT:    .cfi_def_cfa_offset 8
; X86-AVX1-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm0 # encoding: [0xc5,0xfa,0x10,0x44,0x24,0x08]
; X86-AVX1-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-AVX1-NEXT:    vsqrtss %xmm0, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0x51,0xc0]
; X86-AVX1-NEXT:    vmovss %xmm0, (%esp) # encoding: [0xc5,0xfa,0x11,0x04,0x24]
; X86-AVX1-NEXT:    flds (%esp) # encoding: [0xd9,0x04,0x24]
; X86-AVX1-NEXT:    popl %eax # encoding: [0x58]
; X86-AVX1-NEXT:    .cfi_def_cfa_offset 4
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_sqrt_ss_scalar:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    pushl %eax # encoding: [0x50]
; X86-AVX512-NEXT:    .cfi_def_cfa_offset 8
; X86-AVX512-NEXT:    vmovss {{[0-9]+}}(%esp), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x10,0x44,0x24,0x08]
; X86-AVX512-NEXT:    # xmm0 = mem[0],zero,zero,zero
; X86-AVX512-NEXT:    vsqrtss %xmm0, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x51,0xc0]
; X86-AVX512-NEXT:    vmovss %xmm0, (%esp) # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x11,0x04,0x24]
; X86-AVX512-NEXT:    flds (%esp) # encoding: [0xd9,0x04,0x24]
; X86-AVX512-NEXT:    popl %eax # encoding: [0x58]
; X86-AVX512-NEXT:    .cfi_def_cfa_offset 4
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_sqrt_ss_scalar:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    sqrtss %xmm0, %xmm0 # encoding: [0xf3,0x0f,0x51,0xc0]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_sqrt_ss_scalar:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vsqrtss %xmm0, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0x51,0xc0]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_sqrt_ss_scalar:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vsqrtss %xmm0, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x51,0xc0]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %sqrt = call float @llvm.sqrt.f32(float %a0)
  ret float %sqrt
}

define void @test_mm_store_ps(float *%a0, <4 x float> %a1) {
; X86-SSE-LABEL: test_mm_store_ps:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-SSE-NEXT:    movaps %xmm0, (%eax) # encoding: [0x0f,0x29,0x00]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_store_ps:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vmovaps %xmm0, (%eax) # encoding: [0xc5,0xf8,0x29,0x00]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_store_ps:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vmovaps %xmm0, (%eax) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x00]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_store_ps:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movaps %xmm0, (%rdi) # encoding: [0x0f,0x29,0x07]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_store_ps:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vmovaps %xmm0, (%rdi) # encoding: [0xc5,0xf8,0x29,0x07]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_store_ps:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vmovaps %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x07]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %arg0 = bitcast float* %a0 to <4 x float>*
  store <4 x float> %a1, <4 x float>* %arg0, align 16
  ret void
}

define void @test_mm_store_ps1(float *%a0, <4 x float> %a1) {
; X86-SSE-LABEL: test_mm_store_ps1:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-SSE-NEXT:    shufps $0, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x00]
; X86-SSE-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X86-SSE-NEXT:    movaps %xmm0, (%eax) # encoding: [0x0f,0x29,0x00]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_store_ps1:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vpermilps $0, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x04,0xc0,0x00]
; X86-AVX1-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X86-AVX1-NEXT:    vmovaps %xmm0, (%eax) # encoding: [0xc5,0xf8,0x29,0x00]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_store_ps1:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vbroadcastss %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x18,0xc0]
; X86-AVX512-NEXT:    vmovaps %xmm0, (%eax) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x00]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_store_ps1:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    shufps $0, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x00]
; X64-SSE-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X64-SSE-NEXT:    movaps %xmm0, (%rdi) # encoding: [0x0f,0x29,0x07]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_store_ps1:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vpermilps $0, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x04,0xc0,0x00]
; X64-AVX1-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X64-AVX1-NEXT:    vmovaps %xmm0, (%rdi) # encoding: [0xc5,0xf8,0x29,0x07]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_store_ps1:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vbroadcastss %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x18,0xc0]
; X64-AVX512-NEXT:    vmovaps %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x07]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %arg0 = bitcast float* %a0 to <4 x float>*
  %shuf = shufflevector <4 x float> %a1, <4 x float> undef, <4 x i32> zeroinitializer
  store <4 x float> %shuf, <4 x float>* %arg0, align 16
  ret void
}

define void @test_mm_store_ss(float *%a0, <4 x float> %a1) {
; X86-SSE-LABEL: test_mm_store_ss:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-SSE-NEXT:    movss %xmm0, (%eax) # encoding: [0xf3,0x0f,0x11,0x00]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_store_ss:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vmovss %xmm0, (%eax) # encoding: [0xc5,0xfa,0x11,0x00]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_store_ss:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vmovss %xmm0, (%eax) # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x11,0x00]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_store_ss:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movss %xmm0, (%rdi) # encoding: [0xf3,0x0f,0x11,0x07]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_store_ss:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vmovss %xmm0, (%rdi) # encoding: [0xc5,0xfa,0x11,0x07]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_store_ss:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vmovss %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x11,0x07]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %ext = extractelement <4 x float> %a1, i32 0
  store float %ext, float* %a0, align 1
  ret void
}

define void @test_mm_store1_ps(float *%a0, <4 x float> %a1) {
; X86-SSE-LABEL: test_mm_store1_ps:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-SSE-NEXT:    shufps $0, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x00]
; X86-SSE-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X86-SSE-NEXT:    movaps %xmm0, (%eax) # encoding: [0x0f,0x29,0x00]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_store1_ps:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vpermilps $0, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x04,0xc0,0x00]
; X86-AVX1-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X86-AVX1-NEXT:    vmovaps %xmm0, (%eax) # encoding: [0xc5,0xf8,0x29,0x00]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_store1_ps:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vbroadcastss %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x18,0xc0]
; X86-AVX512-NEXT:    vmovaps %xmm0, (%eax) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x00]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_store1_ps:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    shufps $0, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x00]
; X64-SSE-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X64-SSE-NEXT:    movaps %xmm0, (%rdi) # encoding: [0x0f,0x29,0x07]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_store1_ps:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vpermilps $0, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x04,0xc0,0x00]
; X64-AVX1-NEXT:    # xmm0 = xmm0[0,0,0,0]
; X64-AVX1-NEXT:    vmovaps %xmm0, (%rdi) # encoding: [0xc5,0xf8,0x29,0x07]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_store1_ps:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vbroadcastss %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x18,0xc0]
; X64-AVX512-NEXT:    vmovaps %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x07]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %arg0 = bitcast float* %a0 to <4 x float>*
  %shuf = shufflevector <4 x float> %a1, <4 x float> undef, <4 x i32> zeroinitializer
  store <4 x float> %shuf, <4 x float>* %arg0, align 16
  ret void
}

define void @test_mm_storeh_ps(x86_mmx *%a0, <4 x float> %a1) nounwind {
; X86-SSE-LABEL: test_mm_storeh_ps:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    pushl %ebp # encoding: [0x55]
; X86-SSE-NEXT:    movl %esp, %ebp # encoding: [0x89,0xe5]
; X86-SSE-NEXT:    andl $-16, %esp # encoding: [0x83,0xe4,0xf0]
; X86-SSE-NEXT:    subl $32, %esp # encoding: [0x83,0xec,0x20]
; X86-SSE-NEXT:    movl 8(%ebp), %eax # encoding: [0x8b,0x45,0x08]
; X86-SSE-NEXT:    movaps %xmm0, (%esp) # encoding: [0x0f,0x29,0x04,0x24]
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %ecx # encoding: [0x8b,0x4c,0x24,0x08]
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %edx # encoding: [0x8b,0x54,0x24,0x0c]
; X86-SSE-NEXT:    movl %edx, 4(%eax) # encoding: [0x89,0x50,0x04]
; X86-SSE-NEXT:    movl %ecx, (%eax) # encoding: [0x89,0x08]
; X86-SSE-NEXT:    movl %ebp, %esp # encoding: [0x89,0xec]
; X86-SSE-NEXT:    popl %ebp # encoding: [0x5d]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_storeh_ps:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vmovhpd %xmm0, (%eax) # encoding: [0xc5,0xf9,0x17,0x00]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_storeh_ps:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vmovhpd %xmm0, (%eax) # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x17,0x00]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_storeh_ps:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp) # encoding: [0x0f,0x29,0x44,0x24,0xe8]
; X64-SSE-NEXT:    movq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8b,0x44,0x24,0xf0]
; X64-SSE-NEXT:    movq %rax, (%rdi) # encoding: [0x48,0x89,0x07]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_storeh_ps:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vpextrq $1, %xmm0, %rax # encoding: [0xc4,0xe3,0xf9,0x16,0xc0,0x01]
; X64-AVX1-NEXT:    movq %rax, (%rdi) # encoding: [0x48,0x89,0x07]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_storeh_ps:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vpextrq $1, %xmm0, %rax # EVEX TO VEX Compression encoding: [0xc4,0xe3,0xf9,0x16,0xc0,0x01]
; X64-AVX512-NEXT:    movq %rax, (%rdi) # encoding: [0x48,0x89,0x07]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %ptr = bitcast x86_mmx* %a0 to i64*
  %bc  = bitcast <4 x float> %a1 to <2 x i64>
  %ext = extractelement <2 x i64> %bc, i32 1
  store i64 %ext, i64* %ptr
  ret void
}

define void @test_mm_storel_ps(x86_mmx *%a0, <4 x float> %a1) nounwind {
; X86-SSE-LABEL: test_mm_storel_ps:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    pushl %ebp # encoding: [0x55]
; X86-SSE-NEXT:    movl %esp, %ebp # encoding: [0x89,0xe5]
; X86-SSE-NEXT:    andl $-16, %esp # encoding: [0x83,0xe4,0xf0]
; X86-SSE-NEXT:    subl $32, %esp # encoding: [0x83,0xec,0x20]
; X86-SSE-NEXT:    movl 8(%ebp), %eax # encoding: [0x8b,0x45,0x08]
; X86-SSE-NEXT:    movaps %xmm0, (%esp) # encoding: [0x0f,0x29,0x04,0x24]
; X86-SSE-NEXT:    movl (%esp), %ecx # encoding: [0x8b,0x0c,0x24]
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %edx # encoding: [0x8b,0x54,0x24,0x04]
; X86-SSE-NEXT:    movl %edx, 4(%eax) # encoding: [0x89,0x50,0x04]
; X86-SSE-NEXT:    movl %ecx, (%eax) # encoding: [0x89,0x08]
; X86-SSE-NEXT:    movl %ebp, %esp # encoding: [0x89,0xec]
; X86-SSE-NEXT:    popl %ebp # encoding: [0x5d]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_storel_ps:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vmovlps %xmm0, (%eax) # encoding: [0xc5,0xf8,0x13,0x00]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_storel_ps:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vmovlps %xmm0, (%eax) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x13,0x00]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_storel_ps:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movaps %xmm0, -{{[0-9]+}}(%rsp) # encoding: [0x0f,0x29,0x44,0x24,0xe8]
; X64-SSE-NEXT:    movq -{{[0-9]+}}(%rsp), %rax # encoding: [0x48,0x8b,0x44,0x24,0xe8]
; X64-SSE-NEXT:    movq %rax, (%rdi) # encoding: [0x48,0x89,0x07]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_storel_ps:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vmovq %xmm0, %rax # encoding: [0xc4,0xe1,0xf9,0x7e,0xc0]
; X64-AVX1-NEXT:    movq %rax, (%rdi) # encoding: [0x48,0x89,0x07]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_storel_ps:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vmovq %xmm0, %rax # EVEX TO VEX Compression encoding: [0xc4,0xe1,0xf9,0x7e,0xc0]
; X64-AVX512-NEXT:    movq %rax, (%rdi) # encoding: [0x48,0x89,0x07]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %ptr = bitcast x86_mmx* %a0 to i64*
  %bc  = bitcast <4 x float> %a1 to <2 x i64>
  %ext = extractelement <2 x i64> %bc, i32 0
  store i64 %ext, i64* %ptr
  ret void
}

define void @test_mm_storer_ps(float *%a0, <4 x float> %a1) {
; X86-SSE-LABEL: test_mm_storer_ps:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-SSE-NEXT:    shufps $27, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x1b]
; X86-SSE-NEXT:    # xmm0 = xmm0[3,2,1,0]
; X86-SSE-NEXT:    movaps %xmm0, (%eax) # encoding: [0x0f,0x29,0x00]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_storer_ps:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vpermilps $27, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x04,0xc0,0x1b]
; X86-AVX1-NEXT:    # xmm0 = xmm0[3,2,1,0]
; X86-AVX1-NEXT:    vmovaps %xmm0, (%eax) # encoding: [0xc5,0xf8,0x29,0x00]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_storer_ps:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vpermilps $27, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x79,0x04,0xc0,0x1b]
; X86-AVX512-NEXT:    # xmm0 = xmm0[3,2,1,0]
; X86-AVX512-NEXT:    vmovaps %xmm0, (%eax) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x00]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_storer_ps:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    shufps $27, %xmm0, %xmm0 # encoding: [0x0f,0xc6,0xc0,0x1b]
; X64-SSE-NEXT:    # xmm0 = xmm0[3,2,1,0]
; X64-SSE-NEXT:    movaps %xmm0, (%rdi) # encoding: [0x0f,0x29,0x07]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_storer_ps:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vpermilps $27, %xmm0, %xmm0 # encoding: [0xc4,0xe3,0x79,0x04,0xc0,0x1b]
; X64-AVX1-NEXT:    # xmm0 = xmm0[3,2,1,0]
; X64-AVX1-NEXT:    vmovaps %xmm0, (%rdi) # encoding: [0xc5,0xf8,0x29,0x07]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_storer_ps:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vpermilps $27, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc4,0xe3,0x79,0x04,0xc0,0x1b]
; X64-AVX512-NEXT:    # xmm0 = xmm0[3,2,1,0]
; X64-AVX512-NEXT:    vmovaps %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x07]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %arg0 = bitcast float* %a0 to <4 x float>*
  %shuf = shufflevector <4 x float> %a1, <4 x float> undef, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
  store <4 x float> %shuf, <4 x float>* %arg0, align 16
  ret void
}

define void @test_mm_storeu_ps(float *%a0, <4 x float> %a1) {
; X86-SSE-LABEL: test_mm_storeu_ps:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-SSE-NEXT:    movups %xmm0, (%eax) # encoding: [0x0f,0x11,0x00]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_storeu_ps:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vmovups %xmm0, (%eax) # encoding: [0xc5,0xf8,0x11,0x00]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_storeu_ps:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vmovups %xmm0, (%eax) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x11,0x00]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_storeu_ps:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movups %xmm0, (%rdi) # encoding: [0x0f,0x11,0x07]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_storeu_ps:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vmovups %xmm0, (%rdi) # encoding: [0xc5,0xf8,0x11,0x07]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_storeu_ps:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vmovups %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x11,0x07]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %arg0 = bitcast float* %a0 to <4 x float>*
  store <4 x float> %a1, <4 x float>* %arg0, align 1
  ret void
}

define void @test_mm_stream_ps(float *%a0, <4 x float> %a1) {
; X86-SSE-LABEL: test_mm_stream_ps:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-SSE-NEXT:    movntps %xmm0, (%eax) # encoding: [0x0f,0x2b,0x00]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_mm_stream_ps:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX1-NEXT:    vmovntps %xmm0, (%eax) # encoding: [0xc5,0xf8,0x2b,0x00]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_mm_stream_ps:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-AVX512-NEXT:    vmovntps %xmm0, (%eax) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x2b,0x00]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_mm_stream_ps:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movntps %xmm0, (%rdi) # encoding: [0x0f,0x2b,0x07]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_mm_stream_ps:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vmovntps %xmm0, (%rdi) # encoding: [0xc5,0xf8,0x2b,0x07]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_mm_stream_ps:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vmovntps %xmm0, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x2b,0x07]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %arg0 = bitcast float* %a0 to <4 x float>*
  store <4 x float> %a1, <4 x float>* %arg0, align 16, !nontemporal !0
  ret void
}

define <4 x float> @test_mm_sub_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_sub_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    subps %xmm1, %xmm0 # encoding: [0x0f,0x5c,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_sub_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vsubps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x5c,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_sub_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vsubps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x5c,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = fsub <4 x float> %a0, %a1
  ret <4 x float> %res
}

define <4 x float> @test_mm_sub_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_sub_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    subss %xmm1, %xmm0 # encoding: [0xf3,0x0f,0x5c,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_sub_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vsubss %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xfa,0x5c,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_sub_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vsubss %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xfa,0x5c,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %ext0 = extractelement <4 x float> %a0, i32 0
  %ext1 = extractelement <4 x float> %a1, i32 0
  %fsub = fsub float %ext0, %ext1
  %res = insertelement <4 x float> %a0, float %fsub, i32 0
  ret <4 x float> %res
}

define void @test_MM_TRANSPOSE4_PS(<4 x float>* %a0, <4 x float>* %a1, <4 x float>* %a2, <4 x float>* %a3) nounwind {
; X86-SSE-LABEL: test_MM_TRANSPOSE4_PS:
; X86-SSE:       # %bb.0:
; X86-SSE-NEXT:    pushl %esi # encoding: [0x56]
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x14]
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %ecx # encoding: [0x8b,0x4c,0x24,0x10]
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %edx # encoding: [0x8b,0x54,0x24,0x0c]
; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %esi # encoding: [0x8b,0x74,0x24,0x08]
; X86-SSE-NEXT:    movaps (%esi), %xmm0 # encoding: [0x0f,0x28,0x06]
; X86-SSE-NEXT:    movaps (%edx), %xmm1 # encoding: [0x0f,0x28,0x0a]
; X86-SSE-NEXT:    movaps (%ecx), %xmm2 # encoding: [0x0f,0x28,0x11]
; X86-SSE-NEXT:    movaps (%eax), %xmm3 # encoding: [0x0f,0x28,0x18]
; X86-SSE-NEXT:    movaps %xmm0, %xmm4 # encoding: [0x0f,0x28,0xe0]
; X86-SSE-NEXT:    unpcklps %xmm1, %xmm4 # encoding: [0x0f,0x14,0xe1]
; X86-SSE-NEXT:    # xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1]
; X86-SSE-NEXT:    movaps %xmm2, %xmm5 # encoding: [0x0f,0x28,0xea]
; X86-SSE-NEXT:    unpcklps %xmm3, %xmm5 # encoding: [0x0f,0x14,0xeb]
; X86-SSE-NEXT:    # xmm5 = xmm5[0],xmm3[0],xmm5[1],xmm3[1]
; X86-SSE-NEXT:    unpckhps %xmm1, %xmm0 # encoding: [0x0f,0x15,0xc1]
; X86-SSE-NEXT:    # xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; X86-SSE-NEXT:    unpckhps %xmm3, %xmm2 # encoding: [0x0f,0x15,0xd3]
; X86-SSE-NEXT:    # xmm2 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
; X86-SSE-NEXT:    movaps %xmm4, %xmm1 # encoding: [0x0f,0x28,0xcc]
; X86-SSE-NEXT:    movlhps %xmm5, %xmm1 # encoding: [0x0f,0x16,0xcd]
; X86-SSE-NEXT:    # xmm1 = xmm1[0],xmm5[0]
; X86-SSE-NEXT:    movhlps %xmm4, %xmm5 # encoding: [0x0f,0x12,0xec]
; X86-SSE-NEXT:    # xmm5 = xmm4[1],xmm5[1]
; X86-SSE-NEXT:    movaps %xmm0, %xmm3 # encoding: [0x0f,0x28,0xd8]
; X86-SSE-NEXT:    movlhps %xmm2, %xmm3 # encoding: [0x0f,0x16,0xda]
; X86-SSE-NEXT:    # xmm3 = xmm3[0],xmm2[0]
; X86-SSE-NEXT:    movhlps %xmm0, %xmm2 # encoding: [0x0f,0x12,0xd0]
; X86-SSE-NEXT:    # xmm2 = xmm0[1],xmm2[1]
; X86-SSE-NEXT:    movaps %xmm1, (%esi) # encoding: [0x0f,0x29,0x0e]
; X86-SSE-NEXT:    movaps %xmm5, (%edx) # encoding: [0x0f,0x29,0x2a]
; X86-SSE-NEXT:    movaps %xmm3, (%ecx) # encoding: [0x0f,0x29,0x19]
; X86-SSE-NEXT:    movaps %xmm2, (%eax) # encoding: [0x0f,0x29,0x10]
; X86-SSE-NEXT:    popl %esi # encoding: [0x5e]
; X86-SSE-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX1-LABEL: test_MM_TRANSPOSE4_PS:
; X86-AVX1:       # %bb.0:
; X86-AVX1-NEXT:    pushl %esi # encoding: [0x56]
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x14]
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %ecx # encoding: [0x8b,0x4c,0x24,0x10]
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %edx # encoding: [0x8b,0x54,0x24,0x0c]
; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %esi # encoding: [0x8b,0x74,0x24,0x08]
; X86-AVX1-NEXT:    vmovaps (%esi), %xmm0 # encoding: [0xc5,0xf8,0x28,0x06]
; X86-AVX1-NEXT:    vmovaps (%edx), %xmm1 # encoding: [0xc5,0xf8,0x28,0x0a]
; X86-AVX1-NEXT:    vmovaps (%ecx), %xmm2 # encoding: [0xc5,0xf8,0x28,0x11]
; X86-AVX1-NEXT:    vmovaps (%eax), %xmm3 # encoding: [0xc5,0xf8,0x28,0x18]
; X86-AVX1-NEXT:    vunpcklps %xmm1, %xmm0, %xmm4 # encoding: [0xc5,0xf8,0x14,0xe1]
; X86-AVX1-NEXT:    # xmm4 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; X86-AVX1-NEXT:    vunpcklps %xmm3, %xmm2, %xmm5 # encoding: [0xc5,0xe8,0x14,0xeb]
; X86-AVX1-NEXT:    # xmm5 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; X86-AVX1-NEXT:    vunpckhps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x15,0xc1]
; X86-AVX1-NEXT:    # xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; X86-AVX1-NEXT:    vunpckhps %xmm3, %xmm2, %xmm1 # encoding: [0xc5,0xe8,0x15,0xcb]
; X86-AVX1-NEXT:    # xmm1 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
; X86-AVX1-NEXT:    vmovlhps %xmm5, %xmm4, %xmm2 # encoding: [0xc5,0xd8,0x16,0xd5]
; X86-AVX1-NEXT:    # xmm2 = xmm4[0],xmm5[0]
; X86-AVX1-NEXT:    vunpckhpd %xmm5, %xmm4, %xmm3 # encoding: [0xc5,0xd9,0x15,0xdd]
; X86-AVX1-NEXT:    # xmm3 = xmm4[1],xmm5[1]
; X86-AVX1-NEXT:    vmovlhps %xmm1, %xmm0, %xmm4 # encoding: [0xc5,0xf8,0x16,0xe1]
; X86-AVX1-NEXT:    # xmm4 = xmm0[0],xmm1[0]
; X86-AVX1-NEXT:    vunpckhpd %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf9,0x15,0xc1]
; X86-AVX1-NEXT:    # xmm0 = xmm0[1],xmm1[1]
; X86-AVX1-NEXT:    vmovaps %xmm2, (%esi) # encoding: [0xc5,0xf8,0x29,0x16]
; X86-AVX1-NEXT:    vmovaps %xmm3, (%edx) # encoding: [0xc5,0xf8,0x29,0x1a]
; X86-AVX1-NEXT:    vmovaps %xmm4, (%ecx) # encoding: [0xc5,0xf8,0x29,0x21]
; X86-AVX1-NEXT:    vmovaps %xmm0, (%eax) # encoding: [0xc5,0xf8,0x29,0x00]
; X86-AVX1-NEXT:    popl %esi # encoding: [0x5e]
; X86-AVX1-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512-LABEL: test_MM_TRANSPOSE4_PS:
; X86-AVX512:       # %bb.0:
; X86-AVX512-NEXT:    pushl %esi # encoding: [0x56]
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x14]
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %ecx # encoding: [0x8b,0x4c,0x24,0x10]
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %edx # encoding: [0x8b,0x54,0x24,0x0c]
; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %esi # encoding: [0x8b,0x74,0x24,0x08]
; X86-AVX512-NEXT:    vmovaps (%esi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0x06]
; X86-AVX512-NEXT:    vmovaps (%edx), %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0x0a]
; X86-AVX512-NEXT:    vmovaps (%ecx), %xmm2 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0x11]
; X86-AVX512-NEXT:    vmovaps (%eax), %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0x18]
; X86-AVX512-NEXT:    vunpcklps %xmm1, %xmm0, %xmm4 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x14,0xe1]
; X86-AVX512-NEXT:    # xmm4 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; X86-AVX512-NEXT:    vunpcklps %xmm3, %xmm2, %xmm5 # EVEX TO VEX Compression encoding: [0xc5,0xe8,0x14,0xeb]
; X86-AVX512-NEXT:    # xmm5 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; X86-AVX512-NEXT:    vunpckhps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x15,0xc1]
; X86-AVX512-NEXT:    # xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; X86-AVX512-NEXT:    vunpckhps %xmm3, %xmm2, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xe8,0x15,0xcb]
; X86-AVX512-NEXT:    # xmm1 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
; X86-AVX512-NEXT:    vmovlhps %xmm5, %xmm4, %xmm2 # EVEX TO VEX Compression encoding: [0xc5,0xd8,0x16,0xd5]
; X86-AVX512-NEXT:    # xmm2 = xmm4[0],xmm5[0]
; X86-AVX512-NEXT:    vunpckhpd %xmm5, %xmm4, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xd9,0x15,0xdd]
; X86-AVX512-NEXT:    # xmm3 = xmm4[1],xmm5[1]
; X86-AVX512-NEXT:    vmovlhps %xmm1, %xmm0, %xmm4 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x16,0xe1]
; X86-AVX512-NEXT:    # xmm4 = xmm0[0],xmm1[0]
; X86-AVX512-NEXT:    vunpckhpd %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x15,0xc1]
; X86-AVX512-NEXT:    # xmm0 = xmm0[1],xmm1[1]
; X86-AVX512-NEXT:    vmovaps %xmm2, (%esi) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x16]
; X86-AVX512-NEXT:    vmovaps %xmm3, (%edx) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x1a]
; X86-AVX512-NEXT:    vmovaps %xmm4, (%ecx) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x21]
; X86-AVX512-NEXT:    vmovaps %xmm0, (%eax) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x00]
; X86-AVX512-NEXT:    popl %esi # encoding: [0x5e]
; X86-AVX512-NEXT:    retl # encoding: [0xc3]
;
; X64-SSE-LABEL: test_MM_TRANSPOSE4_PS:
; X64-SSE:       # %bb.0:
; X64-SSE-NEXT:    movaps (%rdi), %xmm0 # encoding: [0x0f,0x28,0x07]
; X64-SSE-NEXT:    movaps (%rsi), %xmm1 # encoding: [0x0f,0x28,0x0e]
; X64-SSE-NEXT:    movaps (%rdx), %xmm2 # encoding: [0x0f,0x28,0x12]
; X64-SSE-NEXT:    movaps (%rcx), %xmm3 # encoding: [0x0f,0x28,0x19]
; X64-SSE-NEXT:    movaps %xmm0, %xmm4 # encoding: [0x0f,0x28,0xe0]
; X64-SSE-NEXT:    unpcklps %xmm1, %xmm4 # encoding: [0x0f,0x14,0xe1]
; X64-SSE-NEXT:    # xmm4 = xmm4[0],xmm1[0],xmm4[1],xmm1[1]
; X64-SSE-NEXT:    movaps %xmm2, %xmm5 # encoding: [0x0f,0x28,0xea]
; X64-SSE-NEXT:    unpcklps %xmm3, %xmm5 # encoding: [0x0f,0x14,0xeb]
; X64-SSE-NEXT:    # xmm5 = xmm5[0],xmm3[0],xmm5[1],xmm3[1]
; X64-SSE-NEXT:    unpckhps %xmm1, %xmm0 # encoding: [0x0f,0x15,0xc1]
; X64-SSE-NEXT:    # xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; X64-SSE-NEXT:    unpckhps %xmm3, %xmm2 # encoding: [0x0f,0x15,0xd3]
; X64-SSE-NEXT:    # xmm2 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
; X64-SSE-NEXT:    movaps %xmm4, %xmm1 # encoding: [0x0f,0x28,0xcc]
; X64-SSE-NEXT:    movlhps %xmm5, %xmm1 # encoding: [0x0f,0x16,0xcd]
; X64-SSE-NEXT:    # xmm1 = xmm1[0],xmm5[0]
; X64-SSE-NEXT:    movhlps %xmm4, %xmm5 # encoding: [0x0f,0x12,0xec]
; X64-SSE-NEXT:    # xmm5 = xmm4[1],xmm5[1]
; X64-SSE-NEXT:    movaps %xmm0, %xmm3 # encoding: [0x0f,0x28,0xd8]
; X64-SSE-NEXT:    movlhps %xmm2, %xmm3 # encoding: [0x0f,0x16,0xda]
; X64-SSE-NEXT:    # xmm3 = xmm3[0],xmm2[0]
; X64-SSE-NEXT:    movhlps %xmm0, %xmm2 # encoding: [0x0f,0x12,0xd0]
; X64-SSE-NEXT:    # xmm2 = xmm0[1],xmm2[1]
; X64-SSE-NEXT:    movaps %xmm1, (%rdi) # encoding: [0x0f,0x29,0x0f]
; X64-SSE-NEXT:    movaps %xmm5, (%rsi) # encoding: [0x0f,0x29,0x2e]
; X64-SSE-NEXT:    movaps %xmm3, (%rdx) # encoding: [0x0f,0x29,0x1a]
; X64-SSE-NEXT:    movaps %xmm2, (%rcx) # encoding: [0x0f,0x29,0x11]
; X64-SSE-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX1-LABEL: test_MM_TRANSPOSE4_PS:
; X64-AVX1:       # %bb.0:
; X64-AVX1-NEXT:    vmovaps (%rdi), %xmm0 # encoding: [0xc5,0xf8,0x28,0x07]
; X64-AVX1-NEXT:    vmovaps (%rsi), %xmm1 # encoding: [0xc5,0xf8,0x28,0x0e]
; X64-AVX1-NEXT:    vmovaps (%rdx), %xmm2 # encoding: [0xc5,0xf8,0x28,0x12]
; X64-AVX1-NEXT:    vmovaps (%rcx), %xmm3 # encoding: [0xc5,0xf8,0x28,0x19]
; X64-AVX1-NEXT:    vunpcklps %xmm1, %xmm0, %xmm4 # encoding: [0xc5,0xf8,0x14,0xe1]
; X64-AVX1-NEXT:    # xmm4 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; X64-AVX1-NEXT:    vunpcklps %xmm3, %xmm2, %xmm5 # encoding: [0xc5,0xe8,0x14,0xeb]
; X64-AVX1-NEXT:    # xmm5 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; X64-AVX1-NEXT:    vunpckhps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x15,0xc1]
; X64-AVX1-NEXT:    # xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; X64-AVX1-NEXT:    vunpckhps %xmm3, %xmm2, %xmm1 # encoding: [0xc5,0xe8,0x15,0xcb]
; X64-AVX1-NEXT:    # xmm1 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
; X64-AVX1-NEXT:    vmovlhps %xmm5, %xmm4, %xmm2 # encoding: [0xc5,0xd8,0x16,0xd5]
; X64-AVX1-NEXT:    # xmm2 = xmm4[0],xmm5[0]
; X64-AVX1-NEXT:    vunpckhpd %xmm5, %xmm4, %xmm3 # encoding: [0xc5,0xd9,0x15,0xdd]
; X64-AVX1-NEXT:    # xmm3 = xmm4[1],xmm5[1]
; X64-AVX1-NEXT:    vmovlhps %xmm1, %xmm0, %xmm4 # encoding: [0xc5,0xf8,0x16,0xe1]
; X64-AVX1-NEXT:    # xmm4 = xmm0[0],xmm1[0]
; X64-AVX1-NEXT:    vunpckhpd %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf9,0x15,0xc1]
; X64-AVX1-NEXT:    # xmm0 = xmm0[1],xmm1[1]
; X64-AVX1-NEXT:    vmovaps %xmm2, (%rdi) # encoding: [0xc5,0xf8,0x29,0x17]
; X64-AVX1-NEXT:    vmovaps %xmm3, (%rsi) # encoding: [0xc5,0xf8,0x29,0x1e]
; X64-AVX1-NEXT:    vmovaps %xmm4, (%rdx) # encoding: [0xc5,0xf8,0x29,0x22]
; X64-AVX1-NEXT:    vmovaps %xmm0, (%rcx) # encoding: [0xc5,0xf8,0x29,0x01]
; X64-AVX1-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512-LABEL: test_MM_TRANSPOSE4_PS:
; X64-AVX512:       # %bb.0:
; X64-AVX512-NEXT:    vmovaps (%rdi), %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0x07]
; X64-AVX512-NEXT:    vmovaps (%rsi), %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0x0e]
; X64-AVX512-NEXT:    vmovaps (%rdx), %xmm2 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0x12]
; X64-AVX512-NEXT:    vmovaps (%rcx), %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x28,0x19]
; X64-AVX512-NEXT:    vunpcklps %xmm1, %xmm0, %xmm4 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x14,0xe1]
; X64-AVX512-NEXT:    # xmm4 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; X64-AVX512-NEXT:    vunpcklps %xmm3, %xmm2, %xmm5 # EVEX TO VEX Compression encoding: [0xc5,0xe8,0x14,0xeb]
; X64-AVX512-NEXT:    # xmm5 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
; X64-AVX512-NEXT:    vunpckhps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x15,0xc1]
; X64-AVX512-NEXT:    # xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; X64-AVX512-NEXT:    vunpckhps %xmm3, %xmm2, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xe8,0x15,0xcb]
; X64-AVX512-NEXT:    # xmm1 = xmm2[2],xmm3[2],xmm2[3],xmm3[3]
; X64-AVX512-NEXT:    vmovlhps %xmm5, %xmm4, %xmm2 # EVEX TO VEX Compression encoding: [0xc5,0xd8,0x16,0xd5]
; X64-AVX512-NEXT:    # xmm2 = xmm4[0],xmm5[0]
; X64-AVX512-NEXT:    vunpckhpd %xmm5, %xmm4, %xmm3 # EVEX TO VEX Compression encoding: [0xc5,0xd9,0x15,0xdd]
; X64-AVX512-NEXT:    # xmm3 = xmm4[1],xmm5[1]
; X64-AVX512-NEXT:    vmovlhps %xmm1, %xmm0, %xmm4 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x16,0xe1]
; X64-AVX512-NEXT:    # xmm4 = xmm0[0],xmm1[0]
; X64-AVX512-NEXT:    vunpckhpd %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf9,0x15,0xc1]
; X64-AVX512-NEXT:    # xmm0 = xmm0[1],xmm1[1]
; X64-AVX512-NEXT:    vmovaps %xmm2, (%rdi) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x17]
; X64-AVX512-NEXT:    vmovaps %xmm3, (%rsi) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x1e]
; X64-AVX512-NEXT:    vmovaps %xmm4, (%rdx) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x22]
; X64-AVX512-NEXT:    vmovaps %xmm0, (%rcx) # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x29,0x01]
; X64-AVX512-NEXT:    retq # encoding: [0xc3]
  %row0 = load <4 x float>, <4 x float>* %a0, align 16
  %row1 = load <4 x float>, <4 x float>* %a1, align 16
  %row2 = load <4 x float>, <4 x float>* %a2, align 16
  %row3 = load <4 x float>, <4 x float>* %a3, align 16
  %tmp0 = shufflevector <4 x float> %row0, <4 x float> %row1, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
  %tmp2 = shufflevector <4 x float> %row2, <4 x float> %row3, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
  %tmp1 = shufflevector <4 x float> %row0, <4 x float> %row1, <4 x i32> <i32 2, i32 6, i32 3, i32 7>
  %tmp3 = shufflevector <4 x float> %row2, <4 x float> %row3, <4 x i32> <i32 2, i32 6, i32 3, i32 7>
  %res0 = shufflevector <4 x float> %tmp0, <4 x float> %tmp2, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
  %res1 = shufflevector <4 x float> %tmp2, <4 x float> %tmp0, <4 x i32> <i32 6, i32 7, i32 2, i32 3>
  %res2 = shufflevector <4 x float> %tmp1, <4 x float> %tmp3, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
  %res3 = shufflevector <4 x float> %tmp3, <4 x float> %tmp1, <4 x i32> <i32 6, i32 7, i32 2, i32 3>
  store <4 x float> %res0, <4 x float>* %a0, align 16
  store <4 x float> %res1, <4 x float>* %a1, align 16
  store <4 x float> %res2, <4 x float>* %a2, align 16
  store <4 x float> %res3, <4 x float>* %a3, align 16
  ret void
}

define i32 @test_mm_ucomieq_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_ucomieq_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    ucomiss %xmm1, %xmm0 # encoding: [0x0f,0x2e,0xc1]
; SSE-NEXT:    setnp %al # encoding: [0x0f,0x9b,0xc0]
; SSE-NEXT:    sete %cl # encoding: [0x0f,0x94,0xc1]
; SSE-NEXT:    andb %al, %cl # encoding: [0x20,0xc1]
; SSE-NEXT:    movzbl %cl, %eax # encoding: [0x0f,0xb6,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_ucomieq_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vucomiss %xmm1, %xmm0 # encoding: [0xc5,0xf8,0x2e,0xc1]
; AVX1-NEXT:    setnp %al # encoding: [0x0f,0x9b,0xc0]
; AVX1-NEXT:    sete %cl # encoding: [0x0f,0x94,0xc1]
; AVX1-NEXT:    andb %al, %cl # encoding: [0x20,0xc1]
; AVX1-NEXT:    movzbl %cl, %eax # encoding: [0x0f,0xb6,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_ucomieq_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vucomiss %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x2e,0xc1]
; AVX512-NEXT:    setnp %al # encoding: [0x0f,0x9b,0xc0]
; AVX512-NEXT:    sete %cl # encoding: [0x0f,0x94,0xc1]
; AVX512-NEXT:    andb %al, %cl # encoding: [0x20,0xc1]
; AVX512-NEXT:    movzbl %cl, %eax # encoding: [0x0f,0xb6,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.ucomieq.ss(<4 x float> %a0, <4 x float> %a1)
  ret i32 %res
}
declare i32 @llvm.x86.sse.ucomieq.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_mm_ucomige_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_ucomige_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; SSE-NEXT:    ucomiss %xmm1, %xmm0 # encoding: [0x0f,0x2e,0xc1]
; SSE-NEXT:    setae %al # encoding: [0x0f,0x93,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_ucomige_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX1-NEXT:    vucomiss %xmm1, %xmm0 # encoding: [0xc5,0xf8,0x2e,0xc1]
; AVX1-NEXT:    setae %al # encoding: [0x0f,0x93,0xc0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_ucomige_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX512-NEXT:    vucomiss %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x2e,0xc1]
; AVX512-NEXT:    setae %al # encoding: [0x0f,0x93,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.ucomige.ss(<4 x float> %a0, <4 x float> %a1)
  ret i32 %res
}
declare i32 @llvm.x86.sse.ucomige.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_mm_ucomigt_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_ucomigt_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; SSE-NEXT:    ucomiss %xmm1, %xmm0 # encoding: [0x0f,0x2e,0xc1]
; SSE-NEXT:    seta %al # encoding: [0x0f,0x97,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_ucomigt_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX1-NEXT:    vucomiss %xmm1, %xmm0 # encoding: [0xc5,0xf8,0x2e,0xc1]
; AVX1-NEXT:    seta %al # encoding: [0x0f,0x97,0xc0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_ucomigt_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX512-NEXT:    vucomiss %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x2e,0xc1]
; AVX512-NEXT:    seta %al # encoding: [0x0f,0x97,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.ucomigt.ss(<4 x float> %a0, <4 x float> %a1)
  ret i32 %res
}
declare i32 @llvm.x86.sse.ucomigt.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_mm_ucomile_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_ucomile_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; SSE-NEXT:    ucomiss %xmm0, %xmm1 # encoding: [0x0f,0x2e,0xc8]
; SSE-NEXT:    setae %al # encoding: [0x0f,0x93,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_ucomile_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX1-NEXT:    vucomiss %xmm0, %xmm1 # encoding: [0xc5,0xf8,0x2e,0xc8]
; AVX1-NEXT:    setae %al # encoding: [0x0f,0x93,0xc0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_ucomile_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX512-NEXT:    vucomiss %xmm0, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x2e,0xc8]
; AVX512-NEXT:    setae %al # encoding: [0x0f,0x93,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.ucomile.ss(<4 x float> %a0, <4 x float> %a1)
  ret i32 %res
}
declare i32 @llvm.x86.sse.ucomile.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_mm_ucomilt_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_ucomilt_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; SSE-NEXT:    ucomiss %xmm0, %xmm1 # encoding: [0x0f,0x2e,0xc8]
; SSE-NEXT:    seta %al # encoding: [0x0f,0x97,0xc0]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_ucomilt_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX1-NEXT:    vucomiss %xmm0, %xmm1 # encoding: [0xc5,0xf8,0x2e,0xc8]
; AVX1-NEXT:    seta %al # encoding: [0x0f,0x97,0xc0]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_ucomilt_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; AVX512-NEXT:    vucomiss %xmm0, %xmm1 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x2e,0xc8]
; AVX512-NEXT:    seta %al # encoding: [0x0f,0x97,0xc0]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.ucomilt.ss(<4 x float> %a0, <4 x float> %a1)
  ret i32 %res
}
declare i32 @llvm.x86.sse.ucomilt.ss(<4 x float>, <4 x float>) nounwind readnone

define i32 @test_mm_ucomineq_ss(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_ucomineq_ss:
; SSE:       # %bb.0:
; SSE-NEXT:    ucomiss %xmm1, %xmm0 # encoding: [0x0f,0x2e,0xc1]
; SSE-NEXT:    setp %al # encoding: [0x0f,0x9a,0xc0]
; SSE-NEXT:    setne %cl # encoding: [0x0f,0x95,0xc1]
; SSE-NEXT:    orb %al, %cl # encoding: [0x08,0xc1]
; SSE-NEXT:    movzbl %cl, %eax # encoding: [0x0f,0xb6,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_ucomineq_ss:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vucomiss %xmm1, %xmm0 # encoding: [0xc5,0xf8,0x2e,0xc1]
; AVX1-NEXT:    setp %al # encoding: [0x0f,0x9a,0xc0]
; AVX1-NEXT:    setne %cl # encoding: [0x0f,0x95,0xc1]
; AVX1-NEXT:    orb %al, %cl # encoding: [0x08,0xc1]
; AVX1-NEXT:    movzbl %cl, %eax # encoding: [0x0f,0xb6,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_ucomineq_ss:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vucomiss %xmm1, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x2e,0xc1]
; AVX512-NEXT:    setp %al # encoding: [0x0f,0x9a,0xc0]
; AVX512-NEXT:    setne %cl # encoding: [0x0f,0x95,0xc1]
; AVX512-NEXT:    orb %al, %cl # encoding: [0x08,0xc1]
; AVX512-NEXT:    movzbl %cl, %eax # encoding: [0x0f,0xb6,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i32 @llvm.x86.sse.ucomineq.ss(<4 x float> %a0, <4 x float> %a1)
  ret i32 %res
}
declare i32 @llvm.x86.sse.ucomineq.ss(<4 x float>, <4 x float>) nounwind readnone

define <4 x float> @test_mm_undefined_ps() {
; CHECK-LABEL: test_mm_undefined_ps:
; CHECK:       # %bb.0:
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  ret <4 x float> undef
}

define <4 x float> @test_mm_unpackhi_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_unpackhi_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    unpckhps %xmm1, %xmm0 # encoding: [0x0f,0x15,0xc1]
; SSE-NEXT:    # xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_unpackhi_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vunpckhps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x15,0xc1]
; AVX1-NEXT:    # xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_unpackhi_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vunpckhps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x15,0xc1]
; AVX512-NEXT:    # xmm0 = xmm0[2],xmm1[2],xmm0[3],xmm1[3]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = shufflevector <4 x float> %a0, <4 x float> %a1, <4 x i32> <i32 2, i32 6, i32 3, i32 7>
  ret <4 x float> %res
}

define <4 x float> @test_mm_unpacklo_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_unpacklo_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    unpcklps %xmm1, %xmm0 # encoding: [0x0f,0x14,0xc1]
; SSE-NEXT:    # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_unpacklo_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vunpcklps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x14,0xc1]
; AVX1-NEXT:    # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_unpacklo_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vunpcklps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x14,0xc1]
; AVX512-NEXT:    # xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = shufflevector <4 x float> %a0, <4 x float> %a1, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
  ret <4 x float> %res
}

define <4 x float> @test_mm_xor_ps(<4 x float> %a0, <4 x float> %a1) nounwind {
; SSE-LABEL: test_mm_xor_ps:
; SSE:       # %bb.0:
; SSE-NEXT:    xorps %xmm1, %xmm0 # encoding: [0x0f,0x57,0xc1]
; SSE-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX1-LABEL: test_mm_xor_ps:
; AVX1:       # %bb.0:
; AVX1-NEXT:    vxorps %xmm1, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x57,0xc1]
; AVX1-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
;
; AVX512-LABEL: test_mm_xor_ps:
; AVX512:       # %bb.0:
; AVX512-NEXT:    vxorps %xmm1, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x57,0xc1]
; AVX512-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %arg0 = bitcast <4 x float> %a0 to <4 x i32>
  %arg1 = bitcast <4 x float> %a1 to <4 x i32>
  %res = xor <4 x i32> %arg0, %arg1
  %bc = bitcast <4 x i32> %res to <4 x float>
  ret <4 x float> %bc
}

!0 = !{i32 1}