[llvm] 9d4c3e9 - [X86] Enable v8f16 FNEG custom lowering

Thu Nov 30 02:07:22 PST 2023

Author: Simon Pilgrim
Date: 2023-11-30T10:07:00Z
New Revision: 9d4c3e90356f968781ea3fe3c38b011d13d95c59

URL: https://github.com/llvm/llvm-project/commit/9d4c3e90356f968781ea3fe3c38b011d13d95c59
DIFF: https://github.com/llvm/llvm-project/commit/9d4c3e90356f968781ea3fe3c38b011d13d95c59.diff

LOG: [X86] Enable v8f16 FNEG custom lowering

Added: 
    

Modified: 
    llvm/lib/Target/X86/X86ISelLowering.cpp
    llvm/test/CodeGen/X86/vec_fneg.ll

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index ddba534c7df5f77..981f32b77fffad5 100644

--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -1135,6 +1135,7 @@ X86TargetLowering::X86TargetLowering(const X86TargetMachine &TM,
     setOperationAction(ISD::FSUB, MVT::v8f16, Expand);
     setOperationAction(ISD::FMUL, MVT::v8f16, Expand);
     setOperationAction(ISD::FDIV, MVT::v8f16, Expand);
+    setOperationAction(ISD::FNEG, MVT::v8f16, Custom);
 
     // Custom lower v2i64 and v2f64 selects.
     setOperationAction(ISD::SELECT,             MVT::v2f64, Custom);

diff  --git a/llvm/test/CodeGen/X86/vec_fneg.ll b/llvm/test/CodeGen/X86/vec_fneg.ll
index 121ae3c0f12fdc5..fad14c9bad8aa6e 100644
--- a/llvm/test/CodeGen/X86/vec_fneg.ll
+++ b/llvm/test/CodeGen/X86/vec_fneg.ll
@@ -139,905 +139,55 @@ define <4 x float> @fneg_v4f32(<4 x float> %p) nounwind {
 define <8 x half> @fneg_v8f16(ptr %p) nounwind {
 ; X86-SSE-LABEL: fneg_v8f16:
 ; X86-SSE:       # %bb.0:
-; X86-SSE-NEXT:    subl $148, %esp
 ; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; X86-SSE-NEXT:    movdqa (%eax), %xmm0
-; X86-SSE-NEXT:    movdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,1,1,1]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrlq $48, %xmm0
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1,1]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,3,3,3]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrld $16, %xmm0
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm1, (%esp)
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm2 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; X86-SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; X86-SSE-NEXT:    movdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; X86-SSE-NEXT:    movdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
-; X86-SSE-NEXT:    movhps {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; X86-SSE-NEXT:    # xmm1 = xmm1[0,1],mem[0,1]
-; X86-SSE-NEXT:    movaps %xmm1, %xmm0
-; X86-SSE-NEXT:    addl $148, %esp
-; X86-SSE-NEXT:    retl
-;
-; X86-AVX1-LABEL: fneg_v8f16:
-; X86-AVX1:       # %bb.0:
-; X86-AVX1-NEXT:    pushl %esi
-; X86-AVX1-NEXT:    subl $148, %esp
-; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %esi
-; X86-AVX1-NEXT:    vmovdqa (%esi), %xmm0
-; X86-AVX1-NEXT:    vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX1-NEXT:    calll __extendhfsf2
-; X86-AVX1-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX1-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; X86-AVX1-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX1-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX1-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX1-NEXT:    vxorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0, %xmm0
-; X86-AVX1-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    calll __extendhfsf2
-; X86-AVX1-NEXT:    vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX1-NEXT:    vmovss %xmm0, (%esp)
-; X86-AVX1-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX1-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX1-NEXT:    vxorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0, %xmm0
-; X86-AVX1-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    calll __truncsfhf2
-; X86-AVX1-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX1-NEXT:    vmovss %xmm0, (%esp)
-; X86-AVX1-NEXT:    calll __truncsfhf2
-; X86-AVX1-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    vbroadcastss 4(%esi), %xmm0
-; X86-AVX1-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX1-NEXT:    calll __extendhfsf2
-; X86-AVX1-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX1-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-AVX1-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX1-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX1-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX1-NEXT:    vxorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0, %xmm0
-; X86-AVX1-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    calll __extendhfsf2
-; X86-AVX1-NEXT:    vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX1-NEXT:    vmovss %xmm0, (%esp)
-; X86-AVX1-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX1-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX1-NEXT:    vxorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0, %xmm0
-; X86-AVX1-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    calll __truncsfhf2
-; X86-AVX1-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX1-NEXT:    vmovss %xmm0, (%esp)
-; X86-AVX1-NEXT:    calll __truncsfhf2
-; X86-AVX1-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    vbroadcastss 8(%esi), %xmm0
-; X86-AVX1-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX1-NEXT:    calll __extendhfsf2
-; X86-AVX1-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX1-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-AVX1-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX1-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX1-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX1-NEXT:    vxorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0, %xmm0
-; X86-AVX1-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    calll __extendhfsf2
-; X86-AVX1-NEXT:    vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX1-NEXT:    vmovss %xmm0, (%esp)
-; X86-AVX1-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX1-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX1-NEXT:    vxorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0, %xmm0
-; X86-AVX1-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    calll __truncsfhf2
-; X86-AVX1-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX1-NEXT:    vmovss %xmm0, (%esp)
-; X86-AVX1-NEXT:    calll __truncsfhf2
-; X86-AVX1-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    vbroadcastss 12(%esi), %xmm0
-; X86-AVX1-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX1-NEXT:    calll __extendhfsf2
-; X86-AVX1-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX1-NEXT:    vpsrld $16, %xmm0, %xmm0
-; X86-AVX1-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX1-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX1-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX1-NEXT:    vxorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0, %xmm0
-; X86-AVX1-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    calll __extendhfsf2
-; X86-AVX1-NEXT:    vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX1-NEXT:    vmovss %xmm0, (%esp)
-; X86-AVX1-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX1-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX1-NEXT:    vpxor {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0, %xmm0
-; X86-AVX1-NEXT:    vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    calll __truncsfhf2
-; X86-AVX1-NEXT:    vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-AVX1-NEXT:    vmovss %xmm1, (%esp)
-; X86-AVX1-NEXT:    vpunpcklwd {{[-0-9]+}}(%e{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X86-AVX1-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X86-AVX1-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-AVX1-NEXT:    vpunpcklwd {{[-0-9]+}}(%e{{[sb]}}p), %xmm1, %xmm1 # 16-byte Folded Reload
-; X86-AVX1-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1],xmm1[2],mem[2],xmm1[3],mem[3]
-; X86-AVX1-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
-; X86-AVX1-NEXT:    vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX1-NEXT:    vpunpcklwd {{[-0-9]+}}(%e{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X86-AVX1-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X86-AVX1-NEXT:    vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX1-NEXT:    calll __truncsfhf2
-; X86-AVX1-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X86-AVX1-NEXT:    vpunpckldq {{[-0-9]+}}(%e{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X86-AVX1-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; X86-AVX1-NEXT:    vpunpcklqdq {{[-0-9]+}}(%e{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X86-AVX1-NEXT:    # xmm0 = xmm0[0],mem[0]
-; X86-AVX1-NEXT:    addl $148, %esp
-; X86-AVX1-NEXT:    popl %esi
-; X86-AVX1-NEXT:    retl
-;
-; X86-AVX2-LABEL: fneg_v8f16:
-; X86-AVX2:       # %bb.0:
-; X86-AVX2-NEXT:    pushl %esi
-; X86-AVX2-NEXT:    subl $180, %esp
-; X86-AVX2-NEXT:    movl {{[0-9]+}}(%esp), %esi
-; X86-AVX2-NEXT:    vmovdqa (%esi), %xmm0
-; X86-AVX2-NEXT:    vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX2-NEXT:    calll __extendhfsf2
-; X86-AVX2-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX2-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; X86-AVX2-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX2-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX2-NEXT:    vbroadcastss {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
-; X86-AVX2-NEXT:    vmovups %xmm1, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX2-NEXT:    vxorps %xmm1, %xmm0, %xmm0
-; X86-AVX2-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    calll __extendhfsf2
-; X86-AVX2-NEXT:    vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX2-NEXT:    vmovss %xmm0, (%esp)
-; X86-AVX2-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX2-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX2-NEXT:    vxorps {{[-0-9]+}}(%e{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X86-AVX2-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    calll __truncsfhf2
-; X86-AVX2-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX2-NEXT:    vmovd %xmm0, (%esp)
-; X86-AVX2-NEXT:    vpinsrw $0, 4(%esi), %xmm0, %xmm0
-; X86-AVX2-NEXT:    vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    calll __truncsfhf2
-; X86-AVX2-NEXT:    vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX2-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX2-NEXT:    calll __extendhfsf2
-; X86-AVX2-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX2-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-AVX2-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX2-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX2-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX2-NEXT:    vxorps {{[-0-9]+}}(%e{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X86-AVX2-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    calll __extendhfsf2
-; X86-AVX2-NEXT:    vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX2-NEXT:    vmovss %xmm0, (%esp)
-; X86-AVX2-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX2-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX2-NEXT:    vxorps {{[-0-9]+}}(%e{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X86-AVX2-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    calll __truncsfhf2
-; X86-AVX2-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX2-NEXT:    vmovd %xmm0, (%esp)
-; X86-AVX2-NEXT:    vpinsrw $0, 8(%esi), %xmm0, %xmm0
-; X86-AVX2-NEXT:    vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    calll __truncsfhf2
-; X86-AVX2-NEXT:    vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX2-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX2-NEXT:    calll __extendhfsf2
-; X86-AVX2-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX2-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-AVX2-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX2-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX2-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX2-NEXT:    vxorps {{[-0-9]+}}(%e{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X86-AVX2-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    calll __extendhfsf2
-; X86-AVX2-NEXT:    vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX2-NEXT:    vmovss %xmm0, (%esp)
-; X86-AVX2-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX2-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX2-NEXT:    vxorps {{[-0-9]+}}(%e{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X86-AVX2-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    calll __truncsfhf2
-; X86-AVX2-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX2-NEXT:    vmovd %xmm0, (%esp)
-; X86-AVX2-NEXT:    vpinsrw $0, 12(%esi), %xmm0, %xmm0
-; X86-AVX2-NEXT:    vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    calll __truncsfhf2
-; X86-AVX2-NEXT:    vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX2-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX2-NEXT:    calll __extendhfsf2
-; X86-AVX2-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX2-NEXT:    vpsrld $16, %xmm0, %xmm0
-; X86-AVX2-NEXT:    vpextrw $0, %xmm0, (%esp)
-; X86-AVX2-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX2-NEXT:    vmovss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX2-NEXT:    vxorps {{[-0-9]+}}(%e{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X86-AVX2-NEXT:    vmovups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    calll __extendhfsf2
-; X86-AVX2-NEXT:    vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX2-NEXT:    vmovss %xmm0, (%esp)
-; X86-AVX2-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-AVX2-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-AVX2-NEXT:    vpxor {{[-0-9]+}}(%e{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X86-AVX2-NEXT:    vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    calll __truncsfhf2
-; X86-AVX2-NEXT:    vmovups {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-AVX2-NEXT:    vmovss %xmm1, (%esp)
-; X86-AVX2-NEXT:    vpunpcklwd {{[-0-9]+}}(%e{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X86-AVX2-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X86-AVX2-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-AVX2-NEXT:    vpunpcklwd {{[-0-9]+}}(%e{{[sb]}}p), %xmm1, %xmm1 # 16-byte Folded Reload
-; X86-AVX2-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1],xmm1[2],mem[2],xmm1[3],mem[3]
-; X86-AVX2-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
-; X86-AVX2-NEXT:    vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-AVX2-NEXT:    vpunpcklwd {{[-0-9]+}}(%e{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X86-AVX2-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X86-AVX2-NEXT:    vmovdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-AVX2-NEXT:    calll __truncsfhf2
-; X86-AVX2-NEXT:    vmovdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X86-AVX2-NEXT:    vpunpckldq {{[-0-9]+}}(%e{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X86-AVX2-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; X86-AVX2-NEXT:    vpunpcklqdq {{[-0-9]+}}(%e{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X86-AVX2-NEXT:    # xmm0 = xmm0[0],mem[0]
-; X86-AVX2-NEXT:    addl $180, %esp
-; X86-AVX2-NEXT:    popl %esi
-; X86-AVX2-NEXT:    retl
-;
-; X86-AVX512VL-LABEL: fneg_v8f16:
-; X86-AVX512VL:       # %bb.0:
-; X86-AVX512VL-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; X86-AVX512VL-NEXT:    vmovdqa (%eax), %xmm1
-; X86-AVX512VL-NEXT:    movzwl 12(%eax), %ecx
-; X86-AVX512VL-NEXT:    vmovd %ecx, %xmm0
-; X86-AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm2
-; X86-AVX512VL-NEXT:    vpbroadcastd {{.*#+}} xmm0 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
-; X86-AVX512VL-NEXT:    vpxor %xmm0, %xmm2, %xmm2
-; X86-AVX512VL-NEXT:    vcvtps2ph $4, %xmm2, %xmm2
-; X86-AVX512VL-NEXT:    vmovd %xmm2, %ecx
-; X86-AVX512VL-NEXT:    vpinsrw $0, %ecx, %xmm0, %xmm2
-; X86-AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm3 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-AVX512VL-NEXT:    vpextrw $0, %xmm3, %ecx
-; X86-AVX512VL-NEXT:    movzwl %cx, %ecx
-; X86-AVX512VL-NEXT:    vmovd %ecx, %xmm3
-; X86-AVX512VL-NEXT:    vcvtph2ps %xmm3, %xmm3
-; X86-AVX512VL-NEXT:    vpxor %xmm0, %xmm3, %xmm3
-; X86-AVX512VL-NEXT:    vcvtps2ph $4, %xmm3, %xmm3
-; X86-AVX512VL-NEXT:    vmovd %xmm3, %ecx
-; X86-AVX512VL-NEXT:    vpinsrw $0, %ecx, %xmm0, %xmm3
-; X86-AVX512VL-NEXT:    vpunpcklwd {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3]
-; X86-AVX512VL-NEXT:    movzwl 8(%eax), %ecx
-; X86-AVX512VL-NEXT:    vmovd %ecx, %xmm3
-; X86-AVX512VL-NEXT:    vcvtph2ps %xmm3, %xmm3
-; X86-AVX512VL-NEXT:    vpxor %xmm0, %xmm3, %xmm3
-; X86-AVX512VL-NEXT:    vcvtps2ph $4, %xmm3, %xmm3
-; X86-AVX512VL-NEXT:    vmovd %xmm3, %ecx
-; X86-AVX512VL-NEXT:    vpinsrw $0, %ecx, %xmm0, %xmm3
-; X86-AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm4 = xmm1[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-AVX512VL-NEXT:    vpextrw $0, %xmm4, %ecx
-; X86-AVX512VL-NEXT:    movzwl %cx, %ecx
-; X86-AVX512VL-NEXT:    vmovd %ecx, %xmm4
-; X86-AVX512VL-NEXT:    vcvtph2ps %xmm4, %xmm4
-; X86-AVX512VL-NEXT:    vpxor %xmm0, %xmm4, %xmm4
-; X86-AVX512VL-NEXT:    vcvtps2ph $4, %xmm4, %xmm4
-; X86-AVX512VL-NEXT:    vmovd %xmm4, %ecx
-; X86-AVX512VL-NEXT:    vpinsrw $0, %ecx, %xmm0, %xmm4
-; X86-AVX512VL-NEXT:    vpunpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
-; X86-AVX512VL-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
-; X86-AVX512VL-NEXT:    movzwl 4(%eax), %eax
-; X86-AVX512VL-NEXT:    vmovd %eax, %xmm3
-; X86-AVX512VL-NEXT:    vcvtph2ps %xmm3, %xmm3
-; X86-AVX512VL-NEXT:    vpxor %xmm0, %xmm3, %xmm3
-; X86-AVX512VL-NEXT:    vcvtps2ph $4, %xmm3, %xmm3
-; X86-AVX512VL-NEXT:    vmovd %xmm3, %eax
-; X86-AVX512VL-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm3
-; X86-AVX512VL-NEXT:    vpsrlq $48, %xmm1, %xmm4
-; X86-AVX512VL-NEXT:    vpextrw $0, %xmm4, %eax
-; X86-AVX512VL-NEXT:    movzwl %ax, %eax
-; X86-AVX512VL-NEXT:    vmovd %eax, %xmm4
-; X86-AVX512VL-NEXT:    vcvtph2ps %xmm4, %xmm4
-; X86-AVX512VL-NEXT:    vpxor %xmm0, %xmm4, %xmm4
-; X86-AVX512VL-NEXT:    vcvtps2ph $4, %xmm4, %xmm4
-; X86-AVX512VL-NEXT:    vmovd %xmm4, %eax
-; X86-AVX512VL-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm4
-; X86-AVX512VL-NEXT:    vpunpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
-; X86-AVX512VL-NEXT:    vpextrw $0, %xmm1, %eax
-; X86-AVX512VL-NEXT:    movzwl %ax, %eax
-; X86-AVX512VL-NEXT:    vmovd %eax, %xmm4
-; X86-AVX512VL-NEXT:    vcvtph2ps %xmm4, %xmm4
-; X86-AVX512VL-NEXT:    vpxor %xmm0, %xmm4, %xmm4
-; X86-AVX512VL-NEXT:    vcvtps2ph $4, %xmm4, %xmm4
-; X86-AVX512VL-NEXT:    vmovd %xmm4, %eax
-; X86-AVX512VL-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm4
-; X86-AVX512VL-NEXT:    vpsrld $16, %xmm1, %xmm1
-; X86-AVX512VL-NEXT:    vpextrw $0, %xmm1, %eax
-; X86-AVX512VL-NEXT:    movzwl %ax, %eax
-; X86-AVX512VL-NEXT:    vmovd %eax, %xmm1
-; X86-AVX512VL-NEXT:    vcvtph2ps %xmm1, %xmm1
-; X86-AVX512VL-NEXT:    vpxor %xmm0, %xmm1, %xmm0
-; X86-AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; X86-AVX512VL-NEXT:    vmovd %xmm0, %eax
-; X86-AVX512VL-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm0
-; X86-AVX512VL-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm4[0],xmm0[0],xmm4[1],xmm0[1],xmm4[2],xmm0[2],xmm4[3],xmm0[3]
-; X86-AVX512VL-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
-; X86-AVX512VL-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
-; X86-AVX512VL-NEXT:    retl
-;
-; X86-AVX512FP16-LABEL: fneg_v8f16:
-; X86-AVX512FP16:       # %bb.0:
-; X86-AVX512FP16-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; X86-AVX512FP16-NEXT:    vpbroadcastw {{.*#+}} xmm0 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
-; X86-AVX512FP16-NEXT:    vpxor (%eax), %xmm0, %xmm0
-; X86-AVX512FP16-NEXT:    retl
-;
-; X86-AVX512VLDQ-LABEL: fneg_v8f16:
-; X86-AVX512VLDQ:       # %bb.0:
-; X86-AVX512VLDQ-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; X86-AVX512VLDQ-NEXT:    vmovdqa (%eax), %xmm1
-; X86-AVX512VLDQ-NEXT:    movzwl 12(%eax), %ecx
-; X86-AVX512VLDQ-NEXT:    vmovd %ecx, %xmm0
-; X86-AVX512VLDQ-NEXT:    vcvtph2ps %xmm0, %xmm2
-; X86-AVX512VLDQ-NEXT:    vpbroadcastd {{.*#+}} xmm0 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
-; X86-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm2, %xmm2
-; X86-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm2, %xmm2
-; X86-AVX512VLDQ-NEXT:    vmovd %xmm2, %ecx
-; X86-AVX512VLDQ-NEXT:    vpinsrw $0, %ecx, %xmm0, %xmm2
-; X86-AVX512VLDQ-NEXT:    vpsrldq {{.*#+}} xmm3 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-AVX512VLDQ-NEXT:    vpextrw $0, %xmm3, %ecx
-; X86-AVX512VLDQ-NEXT:    movzwl %cx, %ecx
-; X86-AVX512VLDQ-NEXT:    vmovd %ecx, %xmm3
-; X86-AVX512VLDQ-NEXT:    vcvtph2ps %xmm3, %xmm3
-; X86-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm3, %xmm3
-; X86-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm3, %xmm3
-; X86-AVX512VLDQ-NEXT:    vmovd %xmm3, %ecx
-; X86-AVX512VLDQ-NEXT:    vpinsrw $0, %ecx, %xmm0, %xmm3
-; X86-AVX512VLDQ-NEXT:    vpunpcklwd {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3]
-; X86-AVX512VLDQ-NEXT:    movzwl 8(%eax), %ecx
-; X86-AVX512VLDQ-NEXT:    vmovd %ecx, %xmm3
-; X86-AVX512VLDQ-NEXT:    vcvtph2ps %xmm3, %xmm3
-; X86-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm3, %xmm3
-; X86-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm3, %xmm3
-; X86-AVX512VLDQ-NEXT:    vmovd %xmm3, %ecx
-; X86-AVX512VLDQ-NEXT:    vpinsrw $0, %ecx, %xmm0, %xmm3
-; X86-AVX512VLDQ-NEXT:    vpsrldq {{.*#+}} xmm4 = xmm1[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-AVX512VLDQ-NEXT:    vpextrw $0, %xmm4, %ecx
-; X86-AVX512VLDQ-NEXT:    movzwl %cx, %ecx
-; X86-AVX512VLDQ-NEXT:    vmovd %ecx, %xmm4
-; X86-AVX512VLDQ-NEXT:    vcvtph2ps %xmm4, %xmm4
-; X86-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm4, %xmm4
-; X86-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm4, %xmm4
-; X86-AVX512VLDQ-NEXT:    vmovd %xmm4, %ecx
-; X86-AVX512VLDQ-NEXT:    vpinsrw $0, %ecx, %xmm0, %xmm4
-; X86-AVX512VLDQ-NEXT:    vpunpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
-; X86-AVX512VLDQ-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
-; X86-AVX512VLDQ-NEXT:    movzwl 4(%eax), %eax
-; X86-AVX512VLDQ-NEXT:    vmovd %eax, %xmm3
-; X86-AVX512VLDQ-NEXT:    vcvtph2ps %xmm3, %xmm3
-; X86-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm3, %xmm3
-; X86-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm3, %xmm3
-; X86-AVX512VLDQ-NEXT:    vmovd %xmm3, %eax
-; X86-AVX512VLDQ-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm3
-; X86-AVX512VLDQ-NEXT:    vpsrlq $48, %xmm1, %xmm4
-; X86-AVX512VLDQ-NEXT:    vpextrw $0, %xmm4, %eax
-; X86-AVX512VLDQ-NEXT:    movzwl %ax, %eax
-; X86-AVX512VLDQ-NEXT:    vmovd %eax, %xmm4
-; X86-AVX512VLDQ-NEXT:    vcvtph2ps %xmm4, %xmm4
-; X86-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm4, %xmm4
-; X86-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm4, %xmm4
-; X86-AVX512VLDQ-NEXT:    vmovd %xmm4, %eax
-; X86-AVX512VLDQ-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm4
-; X86-AVX512VLDQ-NEXT:    vpunpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
-; X86-AVX512VLDQ-NEXT:    vpextrw $0, %xmm1, %eax
-; X86-AVX512VLDQ-NEXT:    movzwl %ax, %eax
-; X86-AVX512VLDQ-NEXT:    vmovd %eax, %xmm4
-; X86-AVX512VLDQ-NEXT:    vcvtph2ps %xmm4, %xmm4
-; X86-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm4, %xmm4
-; X86-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm4, %xmm4
-; X86-AVX512VLDQ-NEXT:    vmovd %xmm4, %eax
-; X86-AVX512VLDQ-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm4
-; X86-AVX512VLDQ-NEXT:    vpsrld $16, %xmm1, %xmm1
-; X86-AVX512VLDQ-NEXT:    vpextrw $0, %xmm1, %eax
-; X86-AVX512VLDQ-NEXT:    movzwl %ax, %eax
-; X86-AVX512VLDQ-NEXT:    vmovd %eax, %xmm1
-; X86-AVX512VLDQ-NEXT:    vcvtph2ps %xmm1, %xmm1
-; X86-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm1, %xmm0
-; X86-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; X86-AVX512VLDQ-NEXT:    vmovd %xmm0, %eax
-; X86-AVX512VLDQ-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm0
-; X86-AVX512VLDQ-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm4[0],xmm0[0],xmm4[1],xmm0[1],xmm4[2],xmm0[2],xmm4[3],xmm0[3]
-; X86-AVX512VLDQ-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
-; X86-AVX512VLDQ-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
-; X86-AVX512VLDQ-NEXT:    retl
-;
-; X64-SSE-LABEL: fneg_v8f16:
-; X64-SSE:       # %bb.0:
-; X64-SSE-NEXT:    subq $72, %rsp
-; X64-SSE-NEXT:    movdqa (%rdi), %xmm0
-; X64-SSE-NEXT:    movdqa %xmm0, (%rsp) # 16-byte Spill
-; X64-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,3,3,3]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1,1]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrlq $48, %xmm0
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,1,1,1]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrld $16, %xmm0
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; X64-SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X64-SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
-; X64-SSE-NEXT:    punpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm1 = xmm1[0],mem[0]
-; X64-SSE-NEXT:    movdqa %xmm1, %xmm0
-; X64-SSE-NEXT:    addq $72, %rsp
-; X64-SSE-NEXT:    retq
-;
-; X64-AVX1-LABEL: fneg_v8f16:
-; X64-AVX1:       # %bb.0:
-; X64-AVX1-NEXT:    pushq %rbx
-; X64-AVX1-NEXT:    subq $64, %rsp
-; X64-AVX1-NEXT:    movq %rdi, %rbx
-; X64-AVX1-NEXT:    vmovaps (%rdi), %xmm0
-; X64-AVX1-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-AVX1-NEXT:    vbroadcastss 12(%rdi), %xmm0
-; X64-AVX1-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX1-NEXT:    vxorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
-; X64-AVX1-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX1-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-AVX1-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-AVX1-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-AVX1-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX1-NEXT:    vpxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
-; X64-AVX1-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX1-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; X64-AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X64-AVX1-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-AVX1-NEXT:    vbroadcastss 8(%rbx), %xmm0
-; X64-AVX1-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX1-NEXT:    vxorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
-; X64-AVX1-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX1-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; X64-AVX1-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-AVX1-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-AVX1-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX1-NEXT:    vpxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
-; X64-AVX1-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX1-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; X64-AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X64-AVX1-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X64-AVX1-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; X64-AVX1-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-AVX1-NEXT:    vbroadcastss 4(%rbx), %xmm0
-; X64-AVX1-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX1-NEXT:    vxorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
-; X64-AVX1-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX1-NEXT:    vmovaps %xmm0, (%rsp) # 16-byte Spill
-; X64-AVX1-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-AVX1-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; X64-AVX1-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX1-NEXT:    vpxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
-; X64-AVX1-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX1-NEXT:    vmovdqa (%rsp), %xmm1 # 16-byte Reload
-; X64-AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X64-AVX1-NEXT:    vmovdqa %xmm0, (%rsp) # 16-byte Spill
-; X64-AVX1-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-AVX1-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX1-NEXT:    vxorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
-; X64-AVX1-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX1-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-AVX1-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-AVX1-NEXT:    vpsrld $16, %xmm0, %xmm0
-; X64-AVX1-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX1-NEXT:    vpxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
-; X64-AVX1-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX1-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; X64-AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X64-AVX1-NEXT:    vpunpckldq (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; X64-AVX1-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; X64-AVX1-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X64-AVX1-NEXT:    # xmm0 = xmm0[0],mem[0]
-; X64-AVX1-NEXT:    addq $64, %rsp
-; X64-AVX1-NEXT:    popq %rbx
-; X64-AVX1-NEXT:    retq
-;
-; X64-AVX2-LABEL: fneg_v8f16:
-; X64-AVX2:       # %bb.0:
-; X64-AVX2-NEXT:    pushq %rbx
-; X64-AVX2-NEXT:    subq $80, %rsp
-; X64-AVX2-NEXT:    movq %rdi, %rbx
-; X64-AVX2-NEXT:    vmovdqa (%rdi), %xmm0
-; X64-AVX2-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-AVX2-NEXT:    vpinsrw $0, 12(%rdi), %xmm0, %xmm0
-; X64-AVX2-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
-; X64-AVX2-NEXT:    vmovdqa %xmm1, (%rsp) # 16-byte Spill
-; X64-AVX2-NEXT:    vpxor %xmm1, %xmm0, %xmm0
-; X64-AVX2-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX2-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-AVX2-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-AVX2-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-AVX2-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX2-NEXT:    vpxor (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; X64-AVX2-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX2-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; X64-AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X64-AVX2-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-AVX2-NEXT:    vpinsrw $0, 8(%rbx), %xmm0, %xmm0
-; X64-AVX2-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX2-NEXT:    vpxor (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; X64-AVX2-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX2-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-AVX2-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-AVX2-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-AVX2-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX2-NEXT:    vpxor (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; X64-AVX2-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX2-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; X64-AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X64-AVX2-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X64-AVX2-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; X64-AVX2-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-AVX2-NEXT:    vpinsrw $0, 4(%rbx), %xmm0, %xmm0
-; X64-AVX2-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX2-NEXT:    vpxor (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; X64-AVX2-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX2-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-AVX2-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-AVX2-NEXT:    vpsrlq $48, %xmm0, %xmm0
-; X64-AVX2-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX2-NEXT:    vpxor (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; X64-AVX2-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX2-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; X64-AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X64-AVX2-NEXT:    vmovdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-AVX2-NEXT:    vmovaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-AVX2-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX2-NEXT:    vxorps (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; X64-AVX2-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX2-NEXT:    vmovaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-AVX2-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-AVX2-NEXT:    vpsrld $16, %xmm0, %xmm0
-; X64-AVX2-NEXT:    callq __extendhfsf2 at PLT
-; X64-AVX2-NEXT:    vpxor (%rsp), %xmm0, %xmm0 # 16-byte Folded Reload
-; X64-AVX2-NEXT:    callq __truncsfhf2 at PLT
-; X64-AVX2-NEXT:    vmovdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; X64-AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X64-AVX2-NEXT:    vpunpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X64-AVX2-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; X64-AVX2-NEXT:    vpunpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0, %xmm0 # 16-byte Folded Reload
-; X64-AVX2-NEXT:    # xmm0 = xmm0[0],mem[0]
-; X64-AVX2-NEXT:    addq $80, %rsp
-; X64-AVX2-NEXT:    popq %rbx
-; X64-AVX2-NEXT:    retq
+; X86-SSE-NEXT:    movaps (%eax), %xmm0
+; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
+; X86-SSE-NEXT:    retl
 ;
-; X64-AVX512VL-LABEL: fneg_v8f16:
-; X64-AVX512VL:       # %bb.0:
-; X64-AVX512VL-NEXT:    vmovdqa (%rdi), %xmm1
-; X64-AVX512VL-NEXT:    movzwl 12(%rdi), %eax
-; X64-AVX512VL-NEXT:    vmovd %eax, %xmm0
-; X64-AVX512VL-NEXT:    vcvtph2ps %xmm0, %xmm2
-; X64-AVX512VL-NEXT:    vpbroadcastd {{.*#+}} xmm0 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
-; X64-AVX512VL-NEXT:    vpxor %xmm0, %xmm2, %xmm2
-; X64-AVX512VL-NEXT:    vcvtps2ph $4, %xmm2, %xmm2
-; X64-AVX512VL-NEXT:    vmovd %xmm2, %eax
-; X64-AVX512VL-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm2
-; X64-AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm3 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-AVX512VL-NEXT:    vpextrw $0, %xmm3, %eax
-; X64-AVX512VL-NEXT:    movzwl %ax, %eax
-; X64-AVX512VL-NEXT:    vmovd %eax, %xmm3
-; X64-AVX512VL-NEXT:    vcvtph2ps %xmm3, %xmm3
-; X64-AVX512VL-NEXT:    vpxor %xmm0, %xmm3, %xmm3
-; X64-AVX512VL-NEXT:    vcvtps2ph $4, %xmm3, %xmm3
-; X64-AVX512VL-NEXT:    vmovd %xmm3, %eax
-; X64-AVX512VL-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm3
-; X64-AVX512VL-NEXT:    vpunpcklwd {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3]
-; X64-AVX512VL-NEXT:    movzwl 8(%rdi), %eax
-; X64-AVX512VL-NEXT:    vmovd %eax, %xmm3
-; X64-AVX512VL-NEXT:    vcvtph2ps %xmm3, %xmm3
-; X64-AVX512VL-NEXT:    vpxor %xmm0, %xmm3, %xmm3
-; X64-AVX512VL-NEXT:    vcvtps2ph $4, %xmm3, %xmm3
-; X64-AVX512VL-NEXT:    vmovd %xmm3, %eax
-; X64-AVX512VL-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm3
-; X64-AVX512VL-NEXT:    vpsrldq {{.*#+}} xmm4 = xmm1[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-AVX512VL-NEXT:    vpextrw $0, %xmm4, %eax
-; X64-AVX512VL-NEXT:    movzwl %ax, %eax
-; X64-AVX512VL-NEXT:    vmovd %eax, %xmm4
-; X64-AVX512VL-NEXT:    vcvtph2ps %xmm4, %xmm4
-; X64-AVX512VL-NEXT:    vpxor %xmm0, %xmm4, %xmm4
-; X64-AVX512VL-NEXT:    vcvtps2ph $4, %xmm4, %xmm4
-; X64-AVX512VL-NEXT:    vmovd %xmm4, %eax
-; X64-AVX512VL-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm4
-; X64-AVX512VL-NEXT:    vpunpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
-; X64-AVX512VL-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
-; X64-AVX512VL-NEXT:    movzwl 4(%rdi), %eax
-; X64-AVX512VL-NEXT:    vmovd %eax, %xmm3
-; X64-AVX512VL-NEXT:    vcvtph2ps %xmm3, %xmm3
-; X64-AVX512VL-NEXT:    vpxor %xmm0, %xmm3, %xmm3
-; X64-AVX512VL-NEXT:    vcvtps2ph $4, %xmm3, %xmm3
-; X64-AVX512VL-NEXT:    vmovd %xmm3, %eax
-; X64-AVX512VL-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm3
-; X64-AVX512VL-NEXT:    vpsrlq $48, %xmm1, %xmm4
-; X64-AVX512VL-NEXT:    vpextrw $0, %xmm4, %eax
-; X64-AVX512VL-NEXT:    movzwl %ax, %eax
-; X64-AVX512VL-NEXT:    vmovd %eax, %xmm4
-; X64-AVX512VL-NEXT:    vcvtph2ps %xmm4, %xmm4
-; X64-AVX512VL-NEXT:    vpxor %xmm0, %xmm4, %xmm4
-; X64-AVX512VL-NEXT:    vcvtps2ph $4, %xmm4, %xmm4
-; X64-AVX512VL-NEXT:    vmovd %xmm4, %eax
-; X64-AVX512VL-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm4
-; X64-AVX512VL-NEXT:    vpunpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
-; X64-AVX512VL-NEXT:    vpextrw $0, %xmm1, %eax
-; X64-AVX512VL-NEXT:    movzwl %ax, %eax
-; X64-AVX512VL-NEXT:    vmovd %eax, %xmm4
-; X64-AVX512VL-NEXT:    vcvtph2ps %xmm4, %xmm4
-; X64-AVX512VL-NEXT:    vpxor %xmm0, %xmm4, %xmm4
-; X64-AVX512VL-NEXT:    vcvtps2ph $4, %xmm4, %xmm4
-; X64-AVX512VL-NEXT:    vmovd %xmm4, %eax
-; X64-AVX512VL-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm4
-; X64-AVX512VL-NEXT:    vpsrld $16, %xmm1, %xmm1
-; X64-AVX512VL-NEXT:    vpextrw $0, %xmm1, %eax
-; X64-AVX512VL-NEXT:    movzwl %ax, %eax
-; X64-AVX512VL-NEXT:    vmovd %eax, %xmm1
-; X64-AVX512VL-NEXT:    vcvtph2ps %xmm1, %xmm1
-; X64-AVX512VL-NEXT:    vpxor %xmm0, %xmm1, %xmm0
-; X64-AVX512VL-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; X64-AVX512VL-NEXT:    vmovd %xmm0, %eax
-; X64-AVX512VL-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm0
-; X64-AVX512VL-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm4[0],xmm0[0],xmm4[1],xmm0[1],xmm4[2],xmm0[2],xmm4[3],xmm0[3]
-; X64-AVX512VL-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
-; X64-AVX512VL-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
-; X64-AVX512VL-NEXT:    retq
+; X86-AVX1-LABEL: fneg_v8f16:
+; X86-AVX1:       # %bb.0:
+; X86-AVX1-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X86-AVX1-NEXT:    vmovaps (%eax), %xmm0
+; X86-AVX1-NEXT:    vxorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0, %xmm0
+; X86-AVX1-NEXT:    retl
 ;
-; X64-AVX512FP16-LABEL: fneg_v8f16:
-; X64-AVX512FP16:       # %bb.0:
-; X64-AVX512FP16-NEXT:    vpbroadcastw {{.*#+}} xmm0 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
-; X64-AVX512FP16-NEXT:    vpxor (%rdi), %xmm0, %xmm0
-; X64-AVX512FP16-NEXT:    retq
+; X86-AVX2-LABEL: fneg_v8f16:
+; X86-AVX2:       # %bb.0:
+; X86-AVX2-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X86-AVX2-NEXT:    vpbroadcastw {{.*#+}} xmm0 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
+; X86-AVX2-NEXT:    vpxor (%eax), %xmm0, %xmm0
+; X86-AVX2-NEXT:    retl
 ;
-; X64-AVX512VLDQ-LABEL: fneg_v8f16:
-; X64-AVX512VLDQ:       # %bb.0:
-; X64-AVX512VLDQ-NEXT:    vmovdqa (%rdi), %xmm1
-; X64-AVX512VLDQ-NEXT:    movzwl 12(%rdi), %eax
-; X64-AVX512VLDQ-NEXT:    vmovd %eax, %xmm0
-; X64-AVX512VLDQ-NEXT:    vcvtph2ps %xmm0, %xmm2
-; X64-AVX512VLDQ-NEXT:    vpbroadcastd {{.*#+}} xmm0 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
-; X64-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm2, %xmm2
-; X64-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm2, %xmm2
-; X64-AVX512VLDQ-NEXT:    vmovd %xmm2, %eax
-; X64-AVX512VLDQ-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm2
-; X64-AVX512VLDQ-NEXT:    vpsrldq {{.*#+}} xmm3 = xmm1[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-AVX512VLDQ-NEXT:    vpextrw $0, %xmm3, %eax
-; X64-AVX512VLDQ-NEXT:    movzwl %ax, %eax
-; X64-AVX512VLDQ-NEXT:    vmovd %eax, %xmm3
-; X64-AVX512VLDQ-NEXT:    vcvtph2ps %xmm3, %xmm3
-; X64-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm3, %xmm3
-; X64-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm3, %xmm3
-; X64-AVX512VLDQ-NEXT:    vmovd %xmm3, %eax
-; X64-AVX512VLDQ-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm3
-; X64-AVX512VLDQ-NEXT:    vpunpcklwd {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1],xmm2[2],xmm3[2],xmm2[3],xmm3[3]
-; X64-AVX512VLDQ-NEXT:    movzwl 8(%rdi), %eax
-; X64-AVX512VLDQ-NEXT:    vmovd %eax, %xmm3
-; X64-AVX512VLDQ-NEXT:    vcvtph2ps %xmm3, %xmm3
-; X64-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm3, %xmm3
-; X64-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm3, %xmm3
-; X64-AVX512VLDQ-NEXT:    vmovd %xmm3, %eax
-; X64-AVX512VLDQ-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm3
-; X64-AVX512VLDQ-NEXT:    vpsrldq {{.*#+}} xmm4 = xmm1[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-AVX512VLDQ-NEXT:    vpextrw $0, %xmm4, %eax
-; X64-AVX512VLDQ-NEXT:    movzwl %ax, %eax
-; X64-AVX512VLDQ-NEXT:    vmovd %eax, %xmm4
-; X64-AVX512VLDQ-NEXT:    vcvtph2ps %xmm4, %xmm4
-; X64-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm4, %xmm4
-; X64-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm4, %xmm4
-; X64-AVX512VLDQ-NEXT:    vmovd %xmm4, %eax
-; X64-AVX512VLDQ-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm4
-; X64-AVX512VLDQ-NEXT:    vpunpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
-; X64-AVX512VLDQ-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
-; X64-AVX512VLDQ-NEXT:    movzwl 4(%rdi), %eax
-; X64-AVX512VLDQ-NEXT:    vmovd %eax, %xmm3
-; X64-AVX512VLDQ-NEXT:    vcvtph2ps %xmm3, %xmm3
-; X64-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm3, %xmm3
-; X64-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm3, %xmm3
-; X64-AVX512VLDQ-NEXT:    vmovd %xmm3, %eax
-; X64-AVX512VLDQ-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm3
-; X64-AVX512VLDQ-NEXT:    vpsrlq $48, %xmm1, %xmm4
-; X64-AVX512VLDQ-NEXT:    vpextrw $0, %xmm4, %eax
-; X64-AVX512VLDQ-NEXT:    movzwl %ax, %eax
-; X64-AVX512VLDQ-NEXT:    vmovd %eax, %xmm4
-; X64-AVX512VLDQ-NEXT:    vcvtph2ps %xmm4, %xmm4
-; X64-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm4, %xmm4
-; X64-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm4, %xmm4
-; X64-AVX512VLDQ-NEXT:    vmovd %xmm4, %eax
-; X64-AVX512VLDQ-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm4
-; X64-AVX512VLDQ-NEXT:    vpunpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
-; X64-AVX512VLDQ-NEXT:    vpextrw $0, %xmm1, %eax
-; X64-AVX512VLDQ-NEXT:    movzwl %ax, %eax
-; X64-AVX512VLDQ-NEXT:    vmovd %eax, %xmm4
-; X64-AVX512VLDQ-NEXT:    vcvtph2ps %xmm4, %xmm4
-; X64-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm4, %xmm4
-; X64-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm4, %xmm4
-; X64-AVX512VLDQ-NEXT:    vmovd %xmm4, %eax
-; X64-AVX512VLDQ-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm4
-; X64-AVX512VLDQ-NEXT:    vpsrld $16, %xmm1, %xmm1
-; X64-AVX512VLDQ-NEXT:    vpextrw $0, %xmm1, %eax
-; X64-AVX512VLDQ-NEXT:    movzwl %ax, %eax
-; X64-AVX512VLDQ-NEXT:    vmovd %eax, %xmm1
-; X64-AVX512VLDQ-NEXT:    vcvtph2ps %xmm1, %xmm1
-; X64-AVX512VLDQ-NEXT:    vpxor %xmm0, %xmm1, %xmm0
-; X64-AVX512VLDQ-NEXT:    vcvtps2ph $4, %xmm0, %xmm0
-; X64-AVX512VLDQ-NEXT:    vmovd %xmm0, %eax
-; X64-AVX512VLDQ-NEXT:    vpinsrw $0, %eax, %xmm0, %xmm0
-; X64-AVX512VLDQ-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm4[0],xmm0[0],xmm4[1],xmm0[1],xmm4[2],xmm0[2],xmm4[3],xmm0[3]
-; X64-AVX512VLDQ-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
-; X64-AVX512VLDQ-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
-; X64-AVX512VLDQ-NEXT:    retq
+; X86-AVX512-LABEL: fneg_v8f16:
+; X86-AVX512:       # %bb.0:
+; X86-AVX512-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X86-AVX512-NEXT:    vpbroadcastw {{.*#+}} xmm0 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
+; X86-AVX512-NEXT:    vpxor (%eax), %xmm0, %xmm0
+; X86-AVX512-NEXT:    retl
+;
+; X64-SSE-LABEL: fneg_v8f16:
+; X64-SSE:       # %bb.0:
+; X64-SSE-NEXT:    movaps (%rdi), %xmm0
+; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; X64-SSE-NEXT:    retq
+;
+; X64-AVX1-LABEL: fneg_v8f16:
+; X64-AVX1:       # %bb.0:
+; X64-AVX1-NEXT:    vmovaps (%rdi), %xmm0
+; X64-AVX1-NEXT:    vxorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
+; X64-AVX1-NEXT:    retq
+;
+; X64-AVX2-LABEL: fneg_v8f16:
+; X64-AVX2:       # %bb.0:
+; X64-AVX2-NEXT:    vpbroadcastw {{.*#+}} xmm0 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
+; X64-AVX2-NEXT:    vpxor (%rdi), %xmm0, %xmm0
+; X64-AVX2-NEXT:    retq
+;
+; X64-AVX512-LABEL: fneg_v8f16:
+; X64-AVX512:       # %bb.0:
+; X64-AVX512-NEXT:    vpbroadcastw {{.*#+}} xmm0 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
+; X64-AVX512-NEXT:    vpxor (%rdi), %xmm0, %xmm0
+; X64-AVX512-NEXT:    retq
   %v = load <8 x half>, ptr %p, align 16
   %nnv = fsub <8 x half> <half -0.0, half -0.0, half -0.0, half -0.0, half -0.0, half -0.0, half -0.0, half -0.0>, %v
   ret <8 x half> %nnv
@@ -1190,251 +340,11 @@ define <8 x float> @fneg_v8f32(<8 x float> %p) nounwind {
 define <16 x half> @fneg_v16f16(ptr %p) nounwind {
 ; X86-SSE-LABEL: fneg_v16f16:
 ; X86-SSE:       # %bb.0:
-; X86-SSE-NEXT:    subl $324, %esp # imm = 0x144
 ; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X86-SSE-NEXT:    movaps {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
 ; X86-SSE-NEXT:    movaps (%eax), %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqa 16(%eax), %xmm0
-; X86-SSE-NEXT:    movdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,1,1,1]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrlq $48, %xmm0
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1,1]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,3,3,3]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrld $16, %xmm0
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,1,1,1]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrlq $48, %xmm0
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1,1]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,3,3,3]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrld $16, %xmm0
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm1, (%esp)
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm2 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; X86-SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; X86-SSE-NEXT:    movdqa %xmm0, %xmm3
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm2 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X86-SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; X86-SSE-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm3[0]
-; X86-SSE-NEXT:    movdqu %xmm1, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm2 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; X86-SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; X86-SSE-NEXT:    movdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; X86-SSE-NEXT:    movdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
-; X86-SSE-NEXT:    movhps {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; X86-SSE-NEXT:    # xmm1 = xmm1[0,1],mem[0,1]
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    addl $324, %esp # imm = 0x144
+; X86-SSE-NEXT:    xorps %xmm1, %xmm0
+; X86-SSE-NEXT:    xorps 16(%eax), %xmm1
 ; X86-SSE-NEXT:    retl
 ;
 ; X86-AVX1-LABEL: fneg_v16f16:
@@ -2223,134 +1133,10 @@ define <16 x half> @fneg_v16f16(ptr %p) nounwind {
 ;
 ; X64-SSE-LABEL: fneg_v16f16:
 ; X64-SSE:       # %bb.0:
-; X64-SSE-NEXT:    subq $88, %rsp
-; X64-SSE-NEXT:    movdqa (%rdi), %xmm1
-; X64-SSE-NEXT:    movdqa %xmm1, (%rsp) # 16-byte Spill
-; X64-SSE-NEXT:    movaps 16(%rdi), %xmm0
-; X64-SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa %xmm1, %xmm0
-; X64-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,3,3,3]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1,1]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrlq $48, %xmm0
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,1,1,1]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrld $16, %xmm0
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; X64-SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X64-SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
-; X64-SSE-NEXT:    punpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm1 = xmm1[0],mem[0]
-; X64-SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, (%rsp) # 16-byte Spill
-; X64-SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,3,3,3]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd (%rsp), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    movdqa %xmm0, (%rsp) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1,1]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    punpckldq (%rsp), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; X64-SSE-NEXT:    movdqa %xmm0, (%rsp) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrlq $48, %xmm0
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,1,1,1]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrld $16, %xmm0
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; X64-SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X64-SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
-; X64-SSE-NEXT:    punpcklqdq (%rsp), %xmm1 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm1 = xmm1[0],mem[0]
-; X64-SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    addq $88, %rsp
+; X64-SSE-NEXT:    movaps {{.*#+}} xmm1 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
+; X64-SSE-NEXT:    movaps (%rdi), %xmm0
+; X64-SSE-NEXT:    xorps %xmm1, %xmm0
+; X64-SSE-NEXT:    xorps 16(%rdi), %xmm1
 ; X64-SSE-NEXT:    retq
 ;
 ; X64-AVX1-LABEL: fneg_v16f16:
@@ -3076,497 +1862,15 @@ define <16 x float> @fneg_v16f32(<16 x float> %p) nounwind {
 define <32 x half> @fneg_v32f16(ptr %p) nounwind {
 ; X86-SSE-LABEL: fneg_v32f16:
 ; X86-SSE:       # %bb.0:
-; X86-SSE-NEXT:    subl $644, %esp # imm = 0x284
 ; X86-SSE-NEXT:    movl {{[0-9]+}}(%esp), %eax
+; X86-SSE-NEXT:    movaps {{.*#+}} xmm3 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
 ; X86-SSE-NEXT:    movaps (%eax), %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movaps 16(%eax), %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movaps 32(%eax), %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqa 48(%eax), %xmm0
-; X86-SSE-NEXT:    movdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,1,1,1]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrlq $48, %xmm0
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1,1]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,3,3,3]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrld $16, %xmm0
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,1,1,1]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrlq $48, %xmm0
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1,1]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,3,3,3]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrld $16, %xmm0
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,1,1,1]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrlq $48, %xmm0
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1,1]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,3,3,3]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrld $16, %xmm0
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,1,1,1]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrlq $48, %xmm0
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1,1]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,3,3,3]
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    psrld $16, %xmm0
-; X86-SSE-NEXT:    pextrw $0, %xmm0, %eax
-; X86-SSE-NEXT:    movw %ax, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movups %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __extendhfsf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm0, (%esp)
-; X86-SSE-NEXT:    fstps {{[0-9]+}}(%esp)
-; X86-SSE-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; X86-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE-NEXT:    movdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    movss %xmm1, (%esp)
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm2 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; X86-SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; X86-SSE-NEXT:    movdqa %xmm0, %xmm3
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm2 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; X86-SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; X86-SSE-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm3[0]
-; X86-SSE-NEXT:    movdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm2 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; X86-SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; X86-SSE-NEXT:    movdqa %xmm0, %xmm3
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm2 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; X86-SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; X86-SSE-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm3[0]
-; X86-SSE-NEXT:    movdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm2 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; X86-SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; X86-SSE-NEXT:    movdqa %xmm0, %xmm3
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm2 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X86-SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; X86-SSE-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm3[0]
-; X86-SSE-NEXT:    movdqu %xmm1, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm2 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; X86-SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; X86-SSE-NEXT:    movdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; X86-SSE-NEXT:    movdqu %xmm0, {{[-0-9]+}}(%e{{[sb]}}p) # 16-byte Spill
-; X86-SSE-NEXT:    calll __truncsfhf2
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm3 # 16-byte Reload
-; X86-SSE-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
-; X86-SSE-NEXT:    movdqu {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1]
-; X86-SSE-NEXT:    movhps {{[-0-9]+}}(%e{{[sb]}}p), %xmm3 # 16-byte Folded Reload
-; X86-SSE-NEXT:    # xmm3 = xmm3[0,1],mem[0,1]
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm0 # 16-byte Reload
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm1 # 16-byte Reload
-; X86-SSE-NEXT:    movups {{[-0-9]+}}(%e{{[sb]}}p), %xmm2 # 16-byte Reload
-; X86-SSE-NEXT:    addl $644, %esp # imm = 0x284
+; X86-SSE-NEXT:    xorps %xmm3, %xmm0
+; X86-SSE-NEXT:    movaps 16(%eax), %xmm1
+; X86-SSE-NEXT:    xorps %xmm3, %xmm1
+; X86-SSE-NEXT:    movaps 32(%eax), %xmm2
+; X86-SSE-NEXT:    xorps %xmm3, %xmm2
+; X86-SSE-NEXT:    xorps 48(%eax), %xmm3
 ; X86-SSE-NEXT:    retl
 ;
 ; X86-AVX1-LABEL: fneg_v32f16:
@@ -5136,262 +3440,14 @@ define <32 x half> @fneg_v32f16(ptr %p) nounwind {
 ;
 ; X64-SSE-LABEL: fneg_v32f16:
 ; X64-SSE:       # %bb.0:
-; X64-SSE-NEXT:    subq $120, %rsp
-; X64-SSE-NEXT:    movdqa (%rdi), %xmm1
-; X64-SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps 16(%rdi), %xmm0
-; X64-SSE-NEXT:    movaps %xmm0, (%rsp) # 16-byte Spill
-; X64-SSE-NEXT:    movaps 32(%rdi), %xmm0
-; X64-SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps 48(%rdi), %xmm0
-; X64-SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa %xmm1, %xmm0
-; X64-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,3,3,3]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1,1]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrlq $48, %xmm0
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,1,1,1]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrld $16, %xmm0
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; X64-SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X64-SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
-; X64-SSE-NEXT:    punpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm1 = xmm1[0],mem[0]
-; X64-SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,3,3,3]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1,1]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrlq $48, %xmm0
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,1,1,1]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa (%rsp), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrld $16, %xmm0
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; X64-SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X64-SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
-; X64-SSE-NEXT:    punpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm1 = xmm1[0],mem[0]
-; X64-SSE-NEXT:    movdqa %xmm1, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, (%rsp) # 16-byte Spill
-; X64-SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,3,3,3]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd (%rsp), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    movdqa %xmm0, (%rsp) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1,1]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    punpckldq (%rsp), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrlq $48, %xmm0
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, (%rsp) # 16-byte Spill
-; X64-SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,1,1,1]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd (%rsp), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movaps %xmm0, (%rsp) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrld $16, %xmm0
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa (%rsp), %xmm1 # 16-byte Reload
-; X64-SSE-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; X64-SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm1 = xmm1[0],mem[0],xmm1[1],mem[1]
-; X64-SSE-NEXT:    punpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm1 = xmm1[0],mem[0]
-; X64-SSE-NEXT:    movdqa %xmm1, (%rsp) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,3,3,3]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrldq {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],zero,zero,zero,zero,zero,zero,zero,zero,zero,zero
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1,1]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrlq $48, %xmm0
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,1,1,1]
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    punpcklwd {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm0 = xmm0[0],mem[0],xmm0[1],mem[1],xmm0[2],mem[2],xmm0[3],mem[3]
-; X64-SSE-NEXT:    movdqa %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    xorps {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movaps %xmm0, {{[-0-9]+}}(%r{{[sb]}}p) # 16-byte Spill
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    psrld $16, %xmm0
-; X64-SSE-NEXT:    callq __extendhfsf2 at PLT
-; X64-SSE-NEXT:    pxor {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; X64-SSE-NEXT:    callq __truncsfhf2 at PLT
-; X64-SSE-NEXT:    movdqa {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Reload
-; X64-SSE-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
-; X64-SSE-NEXT:    punpckldq {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm3 = xmm3[0],mem[0],xmm3[1],mem[1]
-; X64-SSE-NEXT:    punpcklqdq {{[-0-9]+}}(%r{{[sb]}}p), %xmm3 # 16-byte Folded Reload
-; X64-SSE-NEXT:    # xmm3 = xmm3[0],mem[0]
-; X64-SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
-; X64-SSE-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm1 # 16-byte Reload
-; X64-SSE-NEXT:    movaps (%rsp), %xmm2 # 16-byte Reload
-; X64-SSE-NEXT:    addq $120, %rsp
+; X64-SSE-NEXT:    movaps {{.*#+}} xmm3 = [-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0,-0.0E+0]
+; X64-SSE-NEXT:    movaps (%rdi), %xmm0
+; X64-SSE-NEXT:    xorps %xmm3, %xmm0
+; X64-SSE-NEXT:    movaps 16(%rdi), %xmm1
+; X64-SSE-NEXT:    xorps %xmm3, %xmm1
+; X64-SSE-NEXT:    movaps 32(%rdi), %xmm2
+; X64-SSE-NEXT:    xorps %xmm3, %xmm2
+; X64-SSE-NEXT:    xorps 48(%rdi), %xmm3
 ; X64-SSE-NEXT:    retq
 ;
 ; X64-AVX1-LABEL: fneg_v32f16:
@@ -6509,6 +4565,4 @@ define <32 x half> @fneg_v32f16(ptr %p) nounwind {
 }
 ;; NOTE: These prefixes are unused and the list is autogenerated. Do not add tests below this line:
 ; X64: {{.*}}
-; X64-AVX512: {{.*}}
 ; X86: {{.*}}
-; X86-AVX512: {{.*}}


        
