[X86] Standardize floating point assembly comments

[RKSimon]

Summary

Consistently try to use APFloat::toString for floating point constant comments to get rid of differences between Constant / ConstantDataSequential values - it should help stop some of the linux-windows buildbot failures matching NaN/INF etc. as well.

Not sure if this is redundant if/when D37184 lands?

[craig.topper]

LGTM

[efriedma]

If the output of llvm::write_double isn't consistent between Linux and Windows, we should probably fix that.

That said, this is still fine.

lib/Target/X86/X86MCInstLower.cpp

2105 ↗

(On Diff #167641)

You could also check for isHalfTy(); I guess it's unlikely to be relevant on x86, though.

llvm/trunk/lib/Target/X86/X86MCInstLower.cpp

for (int i = 0, e = ShuffleMask.size(); i != e; ++i) {
     CS << ']';
     --i; // For loop increments element #.
   }
   CS.flush();
 
   return Comment;
 }
 
+static void printConstant(const APFloat& Flt, raw_ostream &CS) {
+  SmallString<32> Str;
+  Flt.toString(Str);
+  CS << Str;
+}
+
 static void printConstant(const Constant *COp, raw_ostream &CS) {
   if (isa<UndefValue>(COp)) {
     CS << "u";
   } else if (auto *CI = dyn_cast<ConstantInt>(COp)) {
     if (CI->getBitWidth() <= 64) {
       CS << CI->getZExtValue();
     } else {
       // print multi-word constant as (w0,w1)
       const auto &Val = CI->getValue();
       CS << "(";
       for (int i = 0, N = Val.getNumWords(); i < N; ++i) {
         if (i > 0)
           CS << ",";
         CS << Val.getRawData()[i];
       }
       CS << ")";
     }
   } else if (auto *CF = dyn_cast<ConstantFP>(COp)) {
-    SmallString<32> Str;
-    CF->getValueAPF().toString(Str);
-    CS << Str;
+    printConstant(CF->getValueAPF(), CS);
   } else {
     CS << "?";
   }
 }
 
 void X86AsmPrinter::EmitSEHInstruction(const MachineInstr *MI) {
   assert(MF->hasWinCFI() && "SEH_ instruction in function without WinCFI?");
   assert(getSubtarget().isOSWindows() && "SEH_ instruction Windows only");
if (auto *C = getConstantFromPool(*MI, MI->getOperand(4))) {
         CS << "[";
         for (int l = 0; l != NumLanes; ++l) {
           for (int i = 0, NumElements = CDS->getNumElements(); i < NumElements;
                ++i) {
             if (i != 0 || l != 0)
               CS << ",";
             if (CDS->getElementType()->isIntegerTy())
               CS << CDS->getElementAsInteger(i);
-            else if (CDS->getElementType()->isFloatTy())
-              CS << CDS->getElementAsFloat(i);
-            else if (CDS->getElementType()->isDoubleTy())
-              CS << CDS->getElementAsDouble(i);
+            else if (CDS->getElementType()->isHalfTy() ||
+                     CDS->getElementType()->isFloatTy() ||
+                     CDS->getElementType()->isDoubleTy())
+              printConstant(CDS->getElementAsAPFloat(i), CS);
             else
               CS << "?";
           }
         }
         CS << "]";
         OutStreamer->AddComment(CS.str(), !EnablePrintSchedInfo);
       } else if (auto *CV = dyn_cast<ConstantVector>(C)) {
         CS << "<";

llvm/trunk/test/CodeGen/X86/2011-20-21-zext-ui2fp.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mcpu=corei7 | FileCheck %s
 
 ; Check that the booleans are converted using zext and not via sext.
 ; 0x1 means that we only look at the first bit.
 
 define void @ui_to_fp_conv(<8 x float> * nocapture %aFOO, <8 x float>* nocapture %RET) nounwind {
 ; CHECK-LABEL: ui_to_fp_conv:
 ; CHECK:       # %bb.0: # %allocas
-; CHECK-NEXT:    movaps {{.*#+}} xmm0 = [1.000000e+00,1.000000e+00,0.000000e+00,0.000000e+00]
+; CHECK-NEXT:    movaps {{.*#+}} xmm0 = [1,1,0,0]
 ; CHECK-NEXT:    xorps %xmm1, %xmm1
 ; CHECK-NEXT:    movups %xmm1, 16(%rsi)
 ; CHECK-NEXT:    movups %xmm0, (%rsi)
 ; CHECK-NEXT:    retq
 allocas:
   %bincmp = fcmp olt <8 x float> <float 1.000000e+00, float 1.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00> , <float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00, float 3.000000e+00>
   %bool2float = uitofp <8 x i1> %bincmp to <8 x float>
   store <8 x float> %bool2float, <8 x float>* %RET, align 4
   ret void
 }
 
 
 

llvm/trunk/test/CodeGen/X86/avx-basic.ll

; CHECK-NEXT:    retq
   ret <8 x i32> %updatedret.i30.i
 }
 
 ;;;; Don't crash on fneg
 ; rdar://10566486
 define <16 x float> @fneg(<16 x float> %a) nounwind {
 ; CHECK-LABEL: fneg:
 ; CHECK:       ## %bb.0:
-; CHECK-NEXT:    vmovaps {{.*#+}} ymm2 = [-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00]
+; CHECK-NEXT:    vmovaps {{.*#+}} ymm2 = [-0,-0,-0,-0,-0,-0,-0,-0]
 ; CHECK-NEXT:    vxorps %ymm2, %ymm0, %ymm0
 ; CHECK-NEXT:    vxorps %ymm2, %ymm1, %ymm1
 ; CHECK-NEXT:    retq
   %1 = fsub <16 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %a
   ret <16 x float> %1
 }
 
 ;;; Don't crash on build vector

llvm/trunk/test/CodeGen/X86/avx-vbroadcast.ll

entry:
   %vecinit6.i = insertelement <4 x float> %vecinit4.i, float %q, i32 3
   ret <4 x float> %vecinit6.i
 }
 
 ; Don't broadcast constants on pre-AVX2 hardware.
 define <4 x float> @_e2(float* %ptr) nounwind uwtable readnone ssp {
 ; X32-LABEL: _e2:
 ; X32:       ## %bb.0: ## %entry
-; X32-NEXT:    vmovaps {{.*#+}} xmm0 = [-7.812500e-03,-7.812500e-03,-7.812500e-03,-7.812500e-03]
+; X32-NEXT:    vmovaps {{.*#+}} xmm0 = [-0.0078125,-0.0078125,-0.0078125,-0.0078125]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: _e2:
 ; X64:       ## %bb.0: ## %entry
-; X64-NEXT:    vmovaps {{.*#+}} xmm0 = [-7.812500e-03,-7.812500e-03,-7.812500e-03,-7.812500e-03]
+; X64-NEXT:    vmovaps {{.*#+}} xmm0 = [-0.0078125,-0.0078125,-0.0078125,-0.0078125]
 ; X64-NEXT:    retq
 entry:
    %vecinit.i = insertelement <4 x float> undef, float       0xbf80000000000000, i32 0
   %vecinit2.i = insertelement <4 x float> %vecinit.i, float  0xbf80000000000000, i32 1
   %vecinit4.i = insertelement <4 x float> %vecinit2.i, float 0xbf80000000000000, i32 2
   %vecinit6.i = insertelement <4 x float> %vecinit4.i, float 0xbf80000000000000, i32 3
   ret <4 x float> %vecinit6.i
 }

llvm/trunk/test/CodeGen/X86/broadcast-elm-cross-splat-vec.ll

; ALL64-NEXT:    retq
   %res2 = fdiv <8 x float> <float 2.0, float 1.0, float 2.0, float 1.0, float 2.0, float 1.0, float 2.0, float 1.0>, %res1
   ret <8 x float> %res2
 }
 
 
 define <8 x float> @f8xf32_f128(<8 x float> %a) {
 ; AVX-LABEL: f8xf32_f128:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; AVX-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [4,1,2,3,4,1,2,3]
 ; AVX-NEXT:    # ymm1 = mem[0,1,0,1]
 ; AVX-NEXT:    vaddps %ymm1, %ymm0, %ymm0
 ; AVX-NEXT:    vdivps %ymm0, %ymm1, %ymm0
 ; AVX-NEXT:    retl
 ;
 ; ALL32-LABEL: f8xf32_f128:
 ; ALL32:       # %bb.0:
-; ALL32-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; ALL32-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [4,1,2,3,4,1,2,3]
 ; ALL32-NEXT:    # ymm1 = mem[0,1,0,1]
 ; ALL32-NEXT:    vaddps %ymm1, %ymm0, %ymm0
 ; ALL32-NEXT:    vdivps %ymm0, %ymm1, %ymm0
 ; ALL32-NEXT:    retl
 ;
 ; AVX-64-LABEL: f8xf32_f128:
 ; AVX-64:       # %bb.0:
-; AVX-64-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; AVX-64-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [4,1,2,3,4,1,2,3]
 ; AVX-64-NEXT:    # ymm1 = mem[0,1,0,1]
 ; AVX-64-NEXT:    vaddps %ymm1, %ymm0, %ymm0
 ; AVX-64-NEXT:    vdivps %ymm0, %ymm1, %ymm0
 ; AVX-64-NEXT:    retq
 ;
 ; ALL64-LABEL: f8xf32_f128:
 ; ALL64:       # %bb.0:
-; ALL64-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; ALL64-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [4,1,2,3,4,1,2,3]
 ; ALL64-NEXT:    # ymm1 = mem[0,1,0,1]
 ; ALL64-NEXT:    vaddps %ymm1, %ymm0, %ymm0
 ; ALL64-NEXT:    vdivps %ymm0, %ymm1, %ymm0
 ; ALL64-NEXT:    retq
   %res1 = fadd <8 x float> <float 4.0, float 1.0, float 2.0, float 3.0, float 4.0, float 1.0, float 2.0, float 3.0>, %a
   %res2 = fdiv <8 x float> <float 4.0, float 1.0, float 2.0, float 3.0, float 4.0, float 1.0, float 2.0, float 3.0>, %res1
   ret <8 x float> %res2
 }
; AVX512F-64-NEXT:    retq
   %res2 = fdiv <16 x float> <float 2.0, float 1.0, float 2.0, float 1.0, float 2.0, float 1.0, float 2.0, float 1.0, float 2.0, float 1.0, float 2.0, float 1.0, float 2.0, float 1.0, float 2.0, float 1.0>, %res1
   ret <16 x float> %res2
 }
 
 
 define <16 x float> @f16xf32_f128(<16 x float> %a) {
 ; AVX-LABEL: f16xf32_f128:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; AVX-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [4,1,2,3,4,1,2,3]
 ; AVX-NEXT:    # ymm2 = mem[0,1,0,1]
 ; AVX-NEXT:    vaddps %ymm2, %ymm1, %ymm1
 ; AVX-NEXT:    vaddps %ymm2, %ymm0, %ymm0
 ; AVX-NEXT:    vdivps %ymm0, %ymm2, %ymm0
 ; AVX-NEXT:    vdivps %ymm1, %ymm2, %ymm1
 ; AVX-NEXT:    retl
 ;
 ; AVX2-LABEL: f16xf32_f128:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; AVX2-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [4,1,2,3,4,1,2,3]
 ; AVX2-NEXT:    # ymm2 = mem[0,1,0,1]
 ; AVX2-NEXT:    vaddps %ymm2, %ymm1, %ymm1
 ; AVX2-NEXT:    vaddps %ymm2, %ymm0, %ymm0
 ; AVX2-NEXT:    vdivps %ymm0, %ymm2, %ymm0
 ; AVX2-NEXT:    vdivps %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    retl
 ;
 ; AVX512-LABEL: f16xf32_f128:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vbroadcastf32x4 {{.*#+}} zmm1 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; AVX512-NEXT:    vbroadcastf32x4 {{.*#+}} zmm1 = [4,1,2,3,4,1,2,3,4,1,2,3,4,1,2,3]
 ; AVX512-NEXT:    # zmm1 = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
 ; AVX512-NEXT:    vaddps %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vdivps %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    retl
 ;
 ; AVX-64-LABEL: f16xf32_f128:
 ; AVX-64:       # %bb.0:
-; AVX-64-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; AVX-64-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [4,1,2,3,4,1,2,3]
 ; AVX-64-NEXT:    # ymm2 = mem[0,1,0,1]
 ; AVX-64-NEXT:    vaddps %ymm2, %ymm1, %ymm1
 ; AVX-64-NEXT:    vaddps %ymm2, %ymm0, %ymm0
 ; AVX-64-NEXT:    vdivps %ymm0, %ymm2, %ymm0
 ; AVX-64-NEXT:    vdivps %ymm1, %ymm2, %ymm1
 ; AVX-64-NEXT:    retq
 ;
 ; AVX2-64-LABEL: f16xf32_f128:
 ; AVX2-64:       # %bb.0:
-; AVX2-64-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; AVX2-64-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [4,1,2,3,4,1,2,3]
 ; AVX2-64-NEXT:    # ymm2 = mem[0,1,0,1]
 ; AVX2-64-NEXT:    vaddps %ymm2, %ymm1, %ymm1
 ; AVX2-64-NEXT:    vaddps %ymm2, %ymm0, %ymm0
 ; AVX2-64-NEXT:    vdivps %ymm0, %ymm2, %ymm0
 ; AVX2-64-NEXT:    vdivps %ymm1, %ymm2, %ymm1
 ; AVX2-64-NEXT:    retq
 ;
 ; AVX512F-64-LABEL: f16xf32_f128:
 ; AVX512F-64:       # %bb.0:
-; AVX512F-64-NEXT:    vbroadcastf32x4 {{.*#+}} zmm1 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; AVX512F-64-NEXT:    vbroadcastf32x4 {{.*#+}} zmm1 = [4,1,2,3,4,1,2,3,4,1,2,3,4,1,2,3]
 ; AVX512F-64-NEXT:    # zmm1 = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
 ; AVX512F-64-NEXT:    vaddps %zmm1, %zmm0, %zmm0
 ; AVX512F-64-NEXT:    vdivps %zmm0, %zmm1, %zmm0
 ; AVX512F-64-NEXT:    retq
   %res1 = fadd <16 x float> <float 4.0, float 1.0, float 2.0, float 3.0, float 4.0, float 1.0, float 2.0, float 3.0, float 4.0, float 1.0, float 2.0, float 3.0, float 4.0, float 1.0, float 2.0, float 3.0>, %a
   %res2 = fdiv <16 x float> <float 4.0, float 1.0, float 2.0, float 3.0, float 4.0, float 1.0, float 2.0, float 3.0, float 4.0, float 1.0, float 2.0, float 3.0, float 4.0, float 1.0, float 2.0, float 3.0>, %res1
   ret <16 x float> %res2
 }
 
 
 define <16 x float> @f16xf32_f256(<16 x float> %a) {
 ; AVX-LABEL: f16xf32_f256:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vmovaps {{.*#+}} ymm2 = [8.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,5.000000e+00,6.000000e+00,7.000000e+00]
+; AVX-NEXT:    vmovaps {{.*#+}} ymm2 = [8,1,2,3,4,5,6,7]
 ; AVX-NEXT:    vaddps %ymm2, %ymm1, %ymm1
 ; AVX-NEXT:    vaddps %ymm2, %ymm0, %ymm0
 ; AVX-NEXT:    vdivps %ymm0, %ymm2, %ymm0
 ; AVX-NEXT:    vdivps %ymm1, %ymm2, %ymm1
 ; AVX-NEXT:    retl
 ;
 ; AVX2-LABEL: f16xf32_f256:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = [8.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,5.000000e+00,6.000000e+00,7.000000e+00]
+; AVX2-NEXT:    vmovaps {{.*#+}} ymm2 = [8,1,2,3,4,5,6,7]
 ; AVX2-NEXT:    vaddps %ymm2, %ymm1, %ymm1
 ; AVX2-NEXT:    vaddps %ymm2, %ymm0, %ymm0
 ; AVX2-NEXT:    vdivps %ymm0, %ymm2, %ymm0
 ; AVX2-NEXT:    vdivps %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    retl
 ;
 ; AVX512-LABEL: f16xf32_f256:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vbroadcastf64x4 {{.*#+}} zmm1 = [8.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,5.000000e+00,6.000000e+00,7.000000e+00,8.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,5.000000e+00,6.000000e+00,7.000000e+00]
+; AVX512-NEXT:    vbroadcastf64x4 {{.*#+}} zmm1 = [8,1,2,3,4,5,6,7,8,1,2,3,4,5,6,7]
 ; AVX512-NEXT:    # zmm1 = mem[0,1,2,3,0,1,2,3]
 ; AVX512-NEXT:    vaddps %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vdivps %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    retl
 ;
 ; AVX-64-LABEL: f16xf32_f256:
 ; AVX-64:       # %bb.0:
-; AVX-64-NEXT:    vmovaps {{.*#+}} ymm2 = [8.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,5.000000e+00,6.000000e+00,7.000000e+00]
+; AVX-64-NEXT:    vmovaps {{.*#+}} ymm2 = [8,1,2,3,4,5,6,7]
 ; AVX-64-NEXT:    vaddps %ymm2, %ymm1, %ymm1
 ; AVX-64-NEXT:    vaddps %ymm2, %ymm0, %ymm0
 ; AVX-64-NEXT:    vdivps %ymm0, %ymm2, %ymm0
 ; AVX-64-NEXT:    vdivps %ymm1, %ymm2, %ymm1
 ; AVX-64-NEXT:    retq
 ;
 ; AVX2-64-LABEL: f16xf32_f256:
 ; AVX2-64:       # %bb.0:
-; AVX2-64-NEXT:    vmovaps {{.*#+}} ymm2 = [8.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,5.000000e+00,6.000000e+00,7.000000e+00]
+; AVX2-64-NEXT:    vmovaps {{.*#+}} ymm2 = [8,1,2,3,4,5,6,7]
 ; AVX2-64-NEXT:    vaddps %ymm2, %ymm1, %ymm1
 ; AVX2-64-NEXT:    vaddps %ymm2, %ymm0, %ymm0
 ; AVX2-64-NEXT:    vdivps %ymm0, %ymm2, %ymm0
 ; AVX2-64-NEXT:    vdivps %ymm1, %ymm2, %ymm1
 ; AVX2-64-NEXT:    retq
 ;
 ; AVX512F-64-LABEL: f16xf32_f256:
 ; AVX512F-64:       # %bb.0:
-; AVX512F-64-NEXT:    vbroadcastf64x4 {{.*#+}} zmm1 = [8.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,5.000000e+00,6.000000e+00,7.000000e+00,8.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,5.000000e+00,6.000000e+00,7.000000e+00]
+; AVX512F-64-NEXT:    vbroadcastf64x4 {{.*#+}} zmm1 = [8,1,2,3,4,5,6,7,8,1,2,3,4,5,6,7]
 ; AVX512F-64-NEXT:    # zmm1 = mem[0,1,2,3,0,1,2,3]
 ; AVX512F-64-NEXT:    vaddps %zmm1, %zmm0, %zmm0
 ; AVX512F-64-NEXT:    vdivps %zmm0, %zmm1, %zmm0
 ; AVX512F-64-NEXT:    retq
   %res1 = fadd <16 x float> <float 8.0, float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0, float 8.0, float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0>, %a
   %res2 = fdiv <16 x float> <float 8.0, float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0, float 8.0, float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0>, %res1
   ret <16 x float> %res2
 }
 
 
 define <4 x double> @f4xf64_f128(<4 x double> %a) {
 ; AVX-LABEL: f4xf64_f128:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [2.000000e+00,1.000000e+00,2.000000e+00,1.000000e+00]
+; AVX-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [2,1,2,1]
 ; AVX-NEXT:    # ymm1 = mem[0,1,0,1]
 ; AVX-NEXT:    vaddpd %ymm1, %ymm0, %ymm0
 ; AVX-NEXT:    vdivpd %ymm0, %ymm1, %ymm0
 ; AVX-NEXT:    retl
 ;
 ; ALL32-LABEL: f4xf64_f128:
 ; ALL32:       # %bb.0:
-; ALL32-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [2.000000e+00,1.000000e+00,2.000000e+00,1.000000e+00]
+; ALL32-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [2,1,2,1]
 ; ALL32-NEXT:    # ymm1 = mem[0,1,0,1]
 ; ALL32-NEXT:    vaddpd %ymm1, %ymm0, %ymm0
 ; ALL32-NEXT:    vdivpd %ymm0, %ymm1, %ymm0
 ; ALL32-NEXT:    retl
 ;
 ; AVX-64-LABEL: f4xf64_f128:
 ; AVX-64:       # %bb.0:
-; AVX-64-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [2.000000e+00,1.000000e+00,2.000000e+00,1.000000e+00]
+; AVX-64-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [2,1,2,1]
 ; AVX-64-NEXT:    # ymm1 = mem[0,1,0,1]
 ; AVX-64-NEXT:    vaddpd %ymm1, %ymm0, %ymm0
 ; AVX-64-NEXT:    vdivpd %ymm0, %ymm1, %ymm0
 ; AVX-64-NEXT:    retq
 ;
 ; ALL64-LABEL: f4xf64_f128:
 ; ALL64:       # %bb.0:
-; ALL64-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [2.000000e+00,1.000000e+00,2.000000e+00,1.000000e+00]
+; ALL64-NEXT:    vbroadcastf128 {{.*#+}} ymm1 = [2,1,2,1]
 ; ALL64-NEXT:    # ymm1 = mem[0,1,0,1]
 ; ALL64-NEXT:    vaddpd %ymm1, %ymm0, %ymm0
 ; ALL64-NEXT:    vdivpd %ymm0, %ymm1, %ymm0
 ; ALL64-NEXT:    retq
   %res1 = fadd <4 x double> <double 2.0, double 1.0, double 2.0, double 1.0>, %a
   %res2 = fdiv <4 x double> <double 2.0, double 1.0, double 2.0, double 1.0>, %res1
   ret <4 x double> %res2
 }
 
 
 define <8 x double> @f8xf64_f128(<8 x double> %a) {
 ; AVX-LABEL: f8xf64_f128:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [2.000000e+00,1.000000e+00,2.000000e+00,1.000000e+00]
+; AVX-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [2,1,2,1]
 ; AVX-NEXT:    # ymm2 = mem[0,1,0,1]
 ; AVX-NEXT:    vaddpd %ymm2, %ymm1, %ymm1
 ; AVX-NEXT:    vaddpd %ymm2, %ymm0, %ymm0
 ; AVX-NEXT:    vdivpd %ymm0, %ymm2, %ymm0
 ; AVX-NEXT:    vdivpd %ymm1, %ymm2, %ymm1
 ; AVX-NEXT:    retl
 ;
 ; AVX2-LABEL: f8xf64_f128:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [2.000000e+00,1.000000e+00,2.000000e+00,1.000000e+00]
+; AVX2-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [2,1,2,1]
 ; AVX2-NEXT:    # ymm2 = mem[0,1,0,1]
 ; AVX2-NEXT:    vaddpd %ymm2, %ymm1, %ymm1
 ; AVX2-NEXT:    vaddpd %ymm2, %ymm0, %ymm0
 ; AVX2-NEXT:    vdivpd %ymm0, %ymm2, %ymm0
 ; AVX2-NEXT:    vdivpd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    retl
 ;
 ; AVX512-LABEL: f8xf64_f128:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vbroadcastf32x4 {{.*#+}} zmm1 = [2.000000e+00,1.000000e+00,2.000000e+00,1.000000e+00,2.000000e+00,1.000000e+00,2.000000e+00,1.000000e+00]
+; AVX512-NEXT:    vbroadcastf32x4 {{.*#+}} zmm1 = [2,1,2,1,2,1,2,1]
 ; AVX512-NEXT:    # zmm1 = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
 ; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vdivpd %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    retl
 ;
 ; AVX-64-LABEL: f8xf64_f128:
 ; AVX-64:       # %bb.0:
-; AVX-64-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [2.000000e+00,1.000000e+00,2.000000e+00,1.000000e+00]
+; AVX-64-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [2,1,2,1]
 ; AVX-64-NEXT:    # ymm2 = mem[0,1,0,1]
 ; AVX-64-NEXT:    vaddpd %ymm2, %ymm1, %ymm1
 ; AVX-64-NEXT:    vaddpd %ymm2, %ymm0, %ymm0
 ; AVX-64-NEXT:    vdivpd %ymm0, %ymm2, %ymm0
 ; AVX-64-NEXT:    vdivpd %ymm1, %ymm2, %ymm1
 ; AVX-64-NEXT:    retq
 ;
 ; AVX2-64-LABEL: f8xf64_f128:
 ; AVX2-64:       # %bb.0:
-; AVX2-64-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [2.000000e+00,1.000000e+00,2.000000e+00,1.000000e+00]
+; AVX2-64-NEXT:    vbroadcastf128 {{.*#+}} ymm2 = [2,1,2,1]
 ; AVX2-64-NEXT:    # ymm2 = mem[0,1,0,1]
 ; AVX2-64-NEXT:    vaddpd %ymm2, %ymm1, %ymm1
 ; AVX2-64-NEXT:    vaddpd %ymm2, %ymm0, %ymm0
 ; AVX2-64-NEXT:    vdivpd %ymm0, %ymm2, %ymm0
 ; AVX2-64-NEXT:    vdivpd %ymm1, %ymm2, %ymm1
 ; AVX2-64-NEXT:    retq
 ;
 ; AVX512F-64-LABEL: f8xf64_f128:
 ; AVX512F-64:       # %bb.0:
-; AVX512F-64-NEXT:    vbroadcastf32x4 {{.*#+}} zmm1 = [2.000000e+00,1.000000e+00,2.000000e+00,1.000000e+00,2.000000e+00,1.000000e+00,2.000000e+00,1.000000e+00]
+; AVX512F-64-NEXT:    vbroadcastf32x4 {{.*#+}} zmm1 = [2,1,2,1,2,1,2,1]
 ; AVX512F-64-NEXT:    # zmm1 = mem[0,1,2,3,0,1,2,3,0,1,2,3,0,1,2,3]
 ; AVX512F-64-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
 ; AVX512F-64-NEXT:    vdivpd %zmm0, %zmm1, %zmm0
 ; AVX512F-64-NEXT:    retq
   %res1 = fadd <8 x double> <double 2.0, double 1.0, double 2.0, double 1.0, double 2.0, double 1.0, double 2.0, double 1.0>, %a
   %res2 = fdiv <8 x double> <double 2.0, double 1.0, double 2.0, double 1.0, double 2.0, double 1.0, double 2.0, double 1.0>, %res1
   ret <8 x double> %res2
 }
 
 
 ; AVX512:       .LCPI37
 ; AVX512-NEXT:  .quad 4616189618054758400     # double 4
 ; AVX512-NEXT:  .quad 4607182418800017408     # double 1
 ; AVX512-NEXT:  .quad 4611686018427387904     # double 2
 ; AVX512-NEXT:  .quad 4613937818241073152     # double 3
 ; AVX512-NOT:   .quad
 
 define <8 x double> @f8xf64_f256(<8 x double> %a) {
 ; AVX-LABEL: f8xf64_f256:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vmovapd {{.*#+}} ymm2 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; AVX-NEXT:    vmovapd {{.*#+}} ymm2 = [4,1,2,3]
 ; AVX-NEXT:    vaddpd %ymm2, %ymm1, %ymm1
 ; AVX-NEXT:    vaddpd %ymm2, %ymm0, %ymm0
 ; AVX-NEXT:    vdivpd %ymm0, %ymm2, %ymm0
 ; AVX-NEXT:    vdivpd %ymm1, %ymm2, %ymm1
 ; AVX-NEXT:    retl
 ;
 ; AVX2-LABEL: f8xf64_f256:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vmovapd {{.*#+}} ymm2 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; AVX2-NEXT:    vmovapd {{.*#+}} ymm2 = [4,1,2,3]
 ; AVX2-NEXT:    vaddpd %ymm2, %ymm1, %ymm1
 ; AVX2-NEXT:    vaddpd %ymm2, %ymm0, %ymm0
 ; AVX2-NEXT:    vdivpd %ymm0, %ymm2, %ymm0
 ; AVX2-NEXT:    vdivpd %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    retl
 ;
 ; AVX512-LABEL: f8xf64_f256:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vbroadcastf64x4 {{.*#+}} zmm1 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; AVX512-NEXT:    vbroadcastf64x4 {{.*#+}} zmm1 = [4,1,2,3,4,1,2,3]
 ; AVX512-NEXT:    # zmm1 = mem[0,1,2,3,0,1,2,3]
 ; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vdivpd %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    retl
 ;
 ; AVX-64-LABEL: f8xf64_f256:
 ; AVX-64:       # %bb.0:
-; AVX-64-NEXT:    vmovapd {{.*#+}} ymm2 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; AVX-64-NEXT:    vmovapd {{.*#+}} ymm2 = [4,1,2,3]
 ; AVX-64-NEXT:    vaddpd %ymm2, %ymm1, %ymm1
 ; AVX-64-NEXT:    vaddpd %ymm2, %ymm0, %ymm0
 ; AVX-64-NEXT:    vdivpd %ymm0, %ymm2, %ymm0
 ; AVX-64-NEXT:    vdivpd %ymm1, %ymm2, %ymm1
 ; AVX-64-NEXT:    retq
 ;
 ; AVX2-64-LABEL: f8xf64_f256:
 ; AVX2-64:       # %bb.0:
-; AVX2-64-NEXT:    vmovapd {{.*#+}} ymm2 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; AVX2-64-NEXT:    vmovapd {{.*#+}} ymm2 = [4,1,2,3]
 ; AVX2-64-NEXT:    vaddpd %ymm2, %ymm1, %ymm1
 ; AVX2-64-NEXT:    vaddpd %ymm2, %ymm0, %ymm0
 ; AVX2-64-NEXT:    vdivpd %ymm0, %ymm2, %ymm0
 ; AVX2-64-NEXT:    vdivpd %ymm1, %ymm2, %ymm1
 ; AVX2-64-NEXT:    retq
 ;
 ; AVX512F-64-LABEL: f8xf64_f256:
 ; AVX512F-64:       # %bb.0:
-; AVX512F-64-NEXT:    vbroadcastf64x4 {{.*#+}} zmm1 = [4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00,1.000000e+00,2.000000e+00,3.000000e+00]
+; AVX512F-64-NEXT:    vbroadcastf64x4 {{.*#+}} zmm1 = [4,1,2,3,4,1,2,3]
 ; AVX512F-64-NEXT:    # zmm1 = mem[0,1,2,3,0,1,2,3]
 ; AVX512F-64-NEXT:    vaddpd %zmm1, %zmm0, %zmm0
 ; AVX512F-64-NEXT:    vdivpd %zmm0, %zmm1, %zmm0
 ; AVX512F-64-NEXT:    retq
   %res1 = fadd <8 x double> <double 4.0, double 1.0, double 2.0, double 3.0, double 4.0, double 1.0, double 2.0, double 3.0>, %a
   %res2 = fdiv <8 x double> <double 4.0, double 1.0, double 2.0, double 3.0, double 4.0, double 1.0, double 2.0, double 3.0>, %res1
   ret <8 x double> %res2
 }

llvm/trunk/test/CodeGen/X86/combine-fabs.ll

 ; AVX-NEXT:    retq
   %1 = call float @llvm.fabs.f32(float -2.0)
   ret float %1
 }
 
 define <4 x float> @combine_vec_fabs_constant() {
 ; SSE-LABEL: combine_vec_fabs_constant:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movaps {{.*#+}} xmm0 = [0.000000e+00,0.000000e+00,2.000000e+00,2.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [0,0,2,2]
 ; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_vec_fabs_constant:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [0.000000e+00,0.000000e+00,2.000000e+00,2.000000e+00]
+; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [0,0,2,2]
 ; AVX-NEXT:    retq
   %1 = call <4 x float> @llvm.fabs.v4f32(<4 x float> <float 0.0, float -0.0, float 2.0, float -2.0>)
   ret <4 x float> %1
 }
 
 ; fabs(fabs(x)) -> fabs(x)
 define float @combine_fabs_fabs(float %a) {
 ; SSE-LABEL: combine_fabs_fabs:

llvm/trunk/test/CodeGen/X86/combine-fcopysign.ll

 define <4 x float> @combine_vec_fcopysign_neg_constant0(<4 x float> %x) {
 ; SSE-LABEL: combine_vec_fcopysign_neg_constant0:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    orps {{.*}}(%rip), %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_vec_fcopysign_neg_constant0:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vbroadcastss {{.*}}(%rip), %xmm1
+; AVX-NEXT:    vbroadcastss {{.*#+}} xmm1 = [-0,-0,-0,-0]
 ; AVX-NEXT:    vorps %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    retq
   %1 = call <4 x float> @llvm.copysign.v4f32(<4 x float> %x, <4 x float> <float -2.0, float -2.0, float -2.0, float -2.0>)
   ret <4 x float> %1
 }
 
 define <4 x float> @combine_vec_fcopysign_neg_constant1(<4 x float> %x) {
 ; SSE-LABEL: combine_vec_fcopysign_neg_constant1:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    orps {{.*}}(%rip), %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_vec_fcopysign_neg_constant1:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vbroadcastss {{.*}}(%rip), %xmm1
+; AVX-NEXT:    vbroadcastss {{.*#+}} xmm1 = [-0,-0,-0,-0]
 ; AVX-NEXT:    vorps %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    retq
   %1 = call <4 x float> @llvm.copysign.v4f32(<4 x float> %x, <4 x float> <float -0.0, float -2.0, float -4.0, float -8.0>)
   ret <4 x float> %1
 }
 
 define <4 x float> @combine_vec_fcopysign_fneg_fabs_sgn(<4 x float> %x, <4 x float> %y) {
 ; SSE-LABEL: combine_vec_fcopysign_fneg_fabs_sgn:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    orps {{.*}}(%rip), %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_vec_fcopysign_fneg_fabs_sgn:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vbroadcastss {{.*}}(%rip), %xmm1
+; AVX-NEXT:    vbroadcastss {{.*#+}} xmm1 = [-0,-0,-0,-0]
 ; AVX-NEXT:    vorps %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    retq
   %1 = call <4 x float> @llvm.fabs.v4f32(<4 x float> %y)
   %2 = fsub <4 x float> <float -0.0, float -0.0, float -0.0, float -0.0>, %1
   %3 = call <4 x float> @llvm.copysign.v4f32(<4 x float> %x, <4 x float> %2)
   ret <4 x float> %3
 }
 
 ; copysign(fabs(x), y) -> copysign(x, y)
 define <4 x float> @combine_vec_fcopysign_fabs_mag(<4 x float> %x, <4 x float> %y) {
 ; SSE-LABEL: combine_vec_fcopysign_fabs_mag:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    andps {{.*}}(%rip), %xmm1
 ; SSE-NEXT:    andps {{.*}}(%rip), %xmm0
 ; SSE-NEXT:    orps %xmm1, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_vec_fcopysign_fabs_mag:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vbroadcastss {{.*}}(%rip), %xmm2
+; AVX-NEXT:    vbroadcastss {{.*#+}} xmm2 = [-0,-0,-0,-0]
 ; AVX-NEXT:    vandps %xmm2, %xmm1, %xmm1
 ; AVX-NEXT:    vbroadcastss {{.*}}(%rip), %xmm2
 ; AVX-NEXT:    vandps %xmm2, %xmm0, %xmm0
 ; AVX-NEXT:    vorps %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    retq
   %1 = call <4 x float> @llvm.fabs.v4f32(<4 x float> %x)
   %2 = call <4 x float> @llvm.copysign.v4f32(<4 x float> %1, <4 x float> %y)
   ret <4 x float> %2
 }
 
 ; copysign(fneg(x), y) -> copysign(x, y)
 define <4 x float> @combine_vec_fcopysign_fneg_mag(<4 x float> %x, <4 x float> %y) {
 ; SSE-LABEL: combine_vec_fcopysign_fneg_mag:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    andps {{.*}}(%rip), %xmm1
 ; SSE-NEXT:    andps {{.*}}(%rip), %xmm0
 ; SSE-NEXT:    orps %xmm1, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_vec_fcopysign_fneg_mag:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vbroadcastss {{.*}}(%rip), %xmm2
+; AVX-NEXT:    vbroadcastss {{.*#+}} xmm2 = [-0,-0,-0,-0]
 ; AVX-NEXT:    vandps %xmm2, %xmm1, %xmm1
 ; AVX-NEXT:    vbroadcastss {{.*}}(%rip), %xmm2
 ; AVX-NEXT:    vandps %xmm2, %xmm0, %xmm0
 ; AVX-NEXT:    vorps %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    retq
   %1 = fsub <4 x float> <float -0.0, float -0.0, float -0.0, float -0.0>, %x
   %2 = call <4 x float> @llvm.copysign.v4f32(<4 x float> %1, <4 x float> %y)
   ret <4 x float> %2
 }
 
 ; copysign(copysign(x,z), y) -> copysign(x, y)
 define <4 x float> @combine_vec_fcopysign_fcopysign_mag(<4 x float> %x, <4 x float> %y, <4 x float> %z) {
 ; SSE-LABEL: combine_vec_fcopysign_fcopysign_mag:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    andps {{.*}}(%rip), %xmm1
 ; SSE-NEXT:    andps {{.*}}(%rip), %xmm0
 ; SSE-NEXT:    orps %xmm1, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_vec_fcopysign_fcopysign_mag:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vbroadcastss {{.*}}(%rip), %xmm2
+; AVX-NEXT:    vbroadcastss {{.*#+}} xmm2 = [-0,-0,-0,-0]
 ; AVX-NEXT:    vandps %xmm2, %xmm1, %xmm1
 ; AVX-NEXT:    vbroadcastss {{.*}}(%rip), %xmm2
 ; AVX-NEXT:    vandps %xmm2, %xmm0, %xmm0
 ; AVX-NEXT:    vorps %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    retq
   %1 = call <4 x float> @llvm.copysign.v4f32(<4 x float> %x, <4 x float> %z)
   %2 = call <4 x float> @llvm.copysign.v4f32(<4 x float> %1, <4 x float> %y)
   ret <4 x float> %2
 }
 
 ; copysign(x, copysign(y,z)) -> copysign(x, z)
 define <4 x float> @combine_vec_fcopysign_fcopysign_sgn(<4 x float> %x, <4 x float> %y, <4 x float> %z) {
 ; SSE-LABEL: combine_vec_fcopysign_fcopysign_sgn:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    andps {{.*}}(%rip), %xmm2
 ; SSE-NEXT:    andps {{.*}}(%rip), %xmm0
 ; SSE-NEXT:    orps %xmm2, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_vec_fcopysign_fcopysign_sgn:
 ; AVX:       # %bb.0:
-; AVX-NEXT:    vbroadcastss {{.*}}(%rip), %xmm1
+; AVX-NEXT:    vbroadcastss {{.*#+}} xmm1 = [-0,-0,-0,-0]
 ; AVX-NEXT:    vandps %xmm1, %xmm2, %xmm1
 ; AVX-NEXT:    vbroadcastss {{.*}}(%rip), %xmm2
 ; AVX-NEXT:    vandps %xmm2, %xmm0, %xmm0
 ; AVX-NEXT:    vorps %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    retq
   %1 = call <4 x float> @llvm.copysign.v4f32(<4 x float> %y, <4 x float> %z)
   %2 = call <4 x float> @llvm.copysign.v4f32(<4 x float> %x, <4 x float> %1)
   ret <4 x float> %2
 }
 
 ; copysign(x, fp_extend(y)) -> copysign(x, y)
 define <4 x double> @combine_vec_fcopysign_fpext_sgn(<4 x double> %x, <4 x float> %y) {
 ; SSE-LABEL: combine_vec_fcopysign_fpext_sgn:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movaps %xmm2, %xmm3
 ; SSE-NEXT:    cvtss2sd %xmm2, %xmm4
 ; SSE-NEXT:    movshdup {{.*#+}} xmm5 = xmm2[1,1,3,3]
 ; SSE-NEXT:    movaps %xmm2, %xmm6
 ; SSE-NEXT:    unpckhpd {{.*#+}} xmm6 = xmm6[1],xmm2[1]
 ; SSE-NEXT:    shufps {{.*#+}} xmm3 = xmm3[3,1],xmm2[2,3]
 ; SSE-NEXT:    movaps {{.*#+}} xmm7
 ; SSE-NEXT:    movaps %xmm0, %xmm2
 ; SSE-NEXT:    andps %xmm7, %xmm2
-; SSE-NEXT:    movaps {{.*#+}} xmm8 = [-0.000000e+00,-0.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm8 = [-0,-0]
 ; SSE-NEXT:    andps %xmm8, %xmm4
 ; SSE-NEXT:    orps %xmm4, %xmm2
 ; SSE-NEXT:    movhlps {{.*#+}} xmm0 = xmm0[1,1]
 ; SSE-NEXT:    andps %xmm7, %xmm0
 ; SSE-NEXT:    xorps %xmm4, %xmm4
 ; SSE-NEXT:    cvtss2sd %xmm5, %xmm4
 ; SSE-NEXT:    andps %xmm8, %xmm4
 ; SSE-NEXT:    orps %xmm0, %xmm4

 ; SSE-NEXT:    movaps %xmm2, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_vec_fcopysign_fpext_sgn:
 ; AVX:       # %bb.0:
 ; AVX-NEXT:    vbroadcastsd {{.*}}(%rip), %ymm2
 ; AVX-NEXT:    vandps %ymm2, %ymm0, %ymm0
 ; AVX-NEXT:    vcvtps2pd %xmm1, %ymm1
-; AVX-NEXT:    vbroadcastsd {{.*}}(%rip), %ymm2
+; AVX-NEXT:    vbroadcastsd {{.*#+}} ymm2 = [-0,-0,-0,-0]
 ; AVX-NEXT:    vandps %ymm2, %ymm1, %ymm1
 ; AVX-NEXT:    vorps %ymm1, %ymm0, %ymm0
 ; AVX-NEXT:    retq
   %1 = fpext <4 x float> %y to <4 x double>
   %2 = call <4 x double> @llvm.copysign.v4f64(<4 x double> %x, <4 x double> %1)
   ret <4 x double> %2
 }
 
 ; copysign(x, fp_round(y)) -> copysign(x, y)
 define <4 x float> @combine_vec_fcopysign_fptrunc_sgn(<4 x float> %x, <4 x double> %y) {
 ; SSE-LABEL: combine_vec_fcopysign_fptrunc_sgn:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movaps %xmm0, %xmm3
 ; SSE-NEXT:    movaps {{.*#+}} xmm5
 ; SSE-NEXT:    andps %xmm5, %xmm0
 ; SSE-NEXT:    cvtsd2ss %xmm1, %xmm6
-; SSE-NEXT:    movaps {{.*#+}} xmm4 = [-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm4 = [-0,-0,-0,-0]
 ; SSE-NEXT:    andps %xmm4, %xmm6
 ; SSE-NEXT:    orps %xmm6, %xmm0
 ; SSE-NEXT:    movshdup {{.*#+}} xmm6 = xmm3[1,1,3,3]
 ; SSE-NEXT:    andps %xmm5, %xmm6
 ; SSE-NEXT:    movhlps {{.*#+}} xmm1 = xmm1[1,1]
 ; SSE-NEXT:    cvtsd2ss %xmm1, %xmm1
 ; SSE-NEXT:    andps %xmm4, %xmm1
 ; SSE-NEXT:    orps %xmm6, %xmm1

 ; SSE-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[0]
 ; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_vec_fcopysign_fptrunc_sgn:
 ; AVX:       # %bb.0:
 ; AVX-NEXT:    vbroadcastss {{.*}}(%rip), %xmm2
 ; AVX-NEXT:    vandpd %xmm2, %xmm0, %xmm0
 ; AVX-NEXT:    vcvtpd2ps %ymm1, %xmm1
-; AVX-NEXT:    vbroadcastss {{.*}}(%rip), %xmm2
+; AVX-NEXT:    vbroadcastss {{.*#+}} xmm2 = [-0,-0,-0,-0]
 ; AVX-NEXT:    vandpd %xmm2, %xmm1, %xmm1
 ; AVX-NEXT:    vorpd %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    retq
   %1 = fptrunc <4 x double> %y to <4 x float>
   %2 = call <4 x float> @llvm.copysign.v4f32(<4 x float> %x, <4 x float> %1)
   ret <4 x float> %2
 }
 
 declare <4 x float> @llvm.fabs.v4f32(<4 x float> %p)
 declare <4 x float> @llvm.copysign.v4f32(<4 x float> %Mag, <4 x float> %Sgn)
 declare <4 x double> @llvm.copysign.v4f64(<4 x double> %Mag, <4 x double> %Sgn)

llvm/trunk/test/CodeGen/X86/cvtv2f32.ll

 }
 
 define <2 x float> @uitofp_2i32_buildvector_cvt(i32 %x, i32 %y, <2 x float> %v) {
 ; X32-LABEL: uitofp_2i32_buildvector_cvt:
 ; X32:       # %bb.0:
 ; X32-NEXT:    movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
 ; X32-NEXT:    movss {{.*#+}} xmm2 = mem[0],zero,zero,zero
 ; X32-NEXT:    unpcklpd {{.*#+}} xmm2 = xmm2[0],xmm1[0]
-; X32-NEXT:    movapd {{.*#+}} xmm1 = [4.503600e+15,4.503600e+15]
+; X32-NEXT:    movapd {{.*#+}} xmm1 = [4503599627370496,4503599627370496]
 ; X32-NEXT:    orpd %xmm1, %xmm2
 ; X32-NEXT:    subpd %xmm1, %xmm2
 ; X32-NEXT:    cvtpd2ps %xmm2, %xmm1
 ; X32-NEXT:    mulps %xmm1, %xmm0
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: uitofp_2i32_buildvector_cvt:
 ; X64:       # %bb.0:
 ; X64-NEXT:    movd %esi, %xmm1
 ; X64-NEXT:    movd %edi, %xmm2
 ; X64-NEXT:    punpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm1[0]
-; X64-NEXT:    movdqa {{.*#+}} xmm1 = [4.503600e+15,4.503600e+15]
+; X64-NEXT:    movdqa {{.*#+}} xmm1 = [4503599627370496,4503599627370496]
 ; X64-NEXT:    por %xmm1, %xmm2
 ; X64-NEXT:    subpd %xmm1, %xmm2
 ; X64-NEXT:    cvtpd2ps %xmm2, %xmm1
 ; X64-NEXT:    mulps %xmm1, %xmm0
 ; X64-NEXT:    retq
   %t1 = insertelement <2 x i32> undef, i32 %x, i32 0
   %t2 = insertelement <2 x i32> %t1, i32 %y, i32 1
   %t3 = uitofp <2 x i32> %t2 to <2 x float>
   %t4 = fmul <2 x float> %v, %t3
   ret <2 x float> %t4
 }
 
 define <2 x float> @uitofp_2i32_legalized(<2 x i32> %in, <2 x float> %v) {
 ; X32-LABEL: uitofp_2i32_legalized:
 ; X32:       # %bb.0:
 ; X32-NEXT:    xorps %xmm2, %xmm2
 ; X32-NEXT:    blendps {{.*#+}} xmm2 = xmm0[0],xmm2[1],xmm0[2],xmm2[3]
-; X32-NEXT:    movaps {{.*#+}} xmm0 = [4.503600e+15,4.503600e+15]
+; X32-NEXT:    movaps {{.*#+}} xmm0 = [4503599627370496,4503599627370496]
 ; X32-NEXT:    orps %xmm0, %xmm2
 ; X32-NEXT:    subpd %xmm0, %xmm2
 ; X32-NEXT:    cvtpd2ps %xmm2, %xmm0
 ; X32-NEXT:    mulps %xmm1, %xmm0
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: uitofp_2i32_legalized:
 ; X64:       # %bb.0:
 ; X64-NEXT:    xorps %xmm2, %xmm2
 ; X64-NEXT:    blendps {{.*#+}} xmm2 = xmm0[0],xmm2[1],xmm0[2],xmm2[3]
-; X64-NEXT:    movaps {{.*#+}} xmm0 = [4.503600e+15,4.503600e+15]
+; X64-NEXT:    movaps {{.*#+}} xmm0 = [4503599627370496,4503599627370496]
 ; X64-NEXT:    orps %xmm0, %xmm2
 ; X64-NEXT:    subpd %xmm0, %xmm2
 ; X64-NEXT:    cvtpd2ps %xmm2, %xmm0
 ; X64-NEXT:    mulps %xmm1, %xmm0
 ; X64-NEXT:    retq
   %t1 = uitofp <2 x i32> %in to <2 x float>
   %t2 = fmul <2 x float> %v, %t1
   ret <2 x float> %t2
 }

llvm/trunk/test/CodeGen/X86/fma-intrinsics-fast-isel.ll

entry:
   %3 = tail call double @llvm.fma.f64(double %0, double %1, double %2) #2
   %4 = insertelement <2 x double> %a, double %3, i64 0
   ret <2 x double> %4
 }
 
 define <4 x float> @test_mm_fnmsub_ps(<4 x float> %a, <4 x float> %b, <4 x float> %c) {
 ; CHECK-LABEL: test_mm_fnmsub_ps:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    vmovaps {{.*#+}} xmm3 = [-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00]
+; CHECK-NEXT:    vmovaps {{.*#+}} xmm3 = [-0,-0,-0,-0]
 ; CHECK-NEXT:    vxorps %xmm3, %xmm0, %xmm4
 ; CHECK-NEXT:    vxorps %xmm3, %xmm2, %xmm0
 ; CHECK-NEXT:    vfmadd231ps {{.*#+}} xmm0 = (xmm1 * xmm4) + xmm0
 ; CHECK-NEXT:    retq
 entry:
   %sub.i = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %a
   %sub1.i = fsub <4 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %c
   %0 = tail call <4 x float> @llvm.fma.v4f32(<4 x float> %sub.i, <4 x float> %b, <4 x float> %sub1.i) #2
   ret <4 x float> %0
 }
 
 define <2 x double> @test_mm_fnmsub_pd(<2 x double> %a, <2 x double> %b, <2 x double> %c) {
 ; CHECK-LABEL: test_mm_fnmsub_pd:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    vmovapd {{.*#+}} xmm3 = [-0.000000e+00,-0.000000e+00]
+; CHECK-NEXT:    vmovapd {{.*#+}} xmm3 = [-0,-0]
 ; CHECK-NEXT:    vxorpd %xmm3, %xmm0, %xmm4
 ; CHECK-NEXT:    vxorpd %xmm3, %xmm2, %xmm0
 ; CHECK-NEXT:    vfmadd231pd {{.*#+}} xmm0 = (xmm1 * xmm4) + xmm0
 ; CHECK-NEXT:    retq
 entry:
   %sub.i = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, %a
   %sub1.i = fsub <2 x double> <double -0.000000e+00, double -0.000000e+00>, %c
   %0 = tail call <2 x double> @llvm.fma.v2f64(<2 x double> %sub.i, <2 x double> %b, <2 x double> %sub1.i) #2
entry:
   %sub.i = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, %a
   %0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %sub.i, <4 x double> %b, <4 x double> %c) #2
   ret <4 x double> %0
 }
 
 define <8 x float> @test_mm256_fnmsub_ps(<8 x float> %a, <8 x float> %b, <8 x float> %c) {
 ; CHECK-LABEL: test_mm256_fnmsub_ps:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    vmovaps {{.*#+}} ymm3 = [-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00]
+; CHECK-NEXT:    vmovaps {{.*#+}} ymm3 = [-0,-0,-0,-0,-0,-0,-0,-0]
 ; CHECK-NEXT:    vxorps %ymm3, %ymm0, %ymm4
 ; CHECK-NEXT:    vxorps %ymm3, %ymm2, %ymm0
 ; CHECK-NEXT:    vfmadd231ps {{.*#+}} ymm0 = (ymm1 * ymm4) + ymm0
 ; CHECK-NEXT:    retq
 entry:
   %sub.i = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %a
   %sub1.i = fsub <8 x float> <float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00, float -0.000000e+00>, %c
   %0 = tail call <8 x float> @llvm.fma.v8f32(<8 x float> %sub.i, <8 x float> %b, <8 x float> %sub1.i) #2
   ret <8 x float> %0
 }
 
 define <4 x double> @test_mm256_fnmsub_pd(<4 x double> %a, <4 x double> %b, <4 x double> %c) {
 ; CHECK-LABEL: test_mm256_fnmsub_pd:
 ; CHECK:       # %bb.0: # %entry
-; CHECK-NEXT:    vmovapd {{.*#+}} ymm3 = [-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00]
+; CHECK-NEXT:    vmovapd {{.*#+}} ymm3 = [-0,-0,-0,-0]
 ; CHECK-NEXT:    vxorpd %ymm3, %ymm0, %ymm4
 ; CHECK-NEXT:    vxorpd %ymm3, %ymm2, %ymm0
 ; CHECK-NEXT:    vfmadd231pd {{.*#+}} ymm0 = (ymm1 * ymm4) + ymm0
 ; CHECK-NEXT:    retq
 entry:
   %sub.i = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, %a
   %sub1.i = fsub <4 x double> <double -0.000000e+00, double -0.000000e+00, double -0.000000e+00, double -0.000000e+00>, %c
   %0 = tail call <4 x double> @llvm.fma.v4f64(<4 x double> %sub.i, <4 x double> %b, <4 x double> %sub1.i) #2

llvm/trunk/test/CodeGen/X86/fma_patterns.ll

; AVX512-NOINFS-NEXT:    retq
   %a = fadd <4 x float> %x, <float -1.0, float -1.0, float -1.0, float -1.0>
   %m = fmul <4 x float> %y, %a
   ret <4 x float> %m
 }
 
 define <4 x float> @test_v4f32_mul_sub_one_x_y(<4 x float> %x, <4 x float> %y) {
 ; FMA-INFS-LABEL: test_v4f32_mul_sub_one_x_y:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [1,1,1,1]
 ; FMA-INFS-NEXT:    vsubps %xmm0, %xmm2, %xmm0
 ; FMA-INFS-NEXT:    vmulps %xmm1, %xmm0, %xmm0
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v4f32_mul_sub_one_x_y:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA4-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [1,1,1,1]
 ; FMA4-INFS-NEXT:    vsubps %xmm0, %xmm2, %xmm0
 ; FMA4-INFS-NEXT:    vmulps %xmm1, %xmm0, %xmm0
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v4f32_mul_sub_one_x_y:
 ; AVX512-INFS:       # %bb.0:
 ; AVX512-INFS-NEXT:    vbroadcastss {{.*#+}} xmm2 = [1,1,1,1]
 ; AVX512-INFS-NEXT:    vsubps %xmm0, %xmm2, %xmm0
; AVX512-NOINFS-NEXT:    retq
   %s = fsub <4 x float> <float 1.0, float 1.0, float 1.0, float 1.0>, %x
   %m = fmul <4 x float> %s, %y
   ret <4 x float> %m
 }
 
 define <4 x float> @test_v4f32_mul_y_sub_one_x(<4 x float> %x, <4 x float> %y) {
 ; FMA-INFS-LABEL: test_v4f32_mul_y_sub_one_x:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [1,1,1,1]
 ; FMA-INFS-NEXT:    vsubps %xmm0, %xmm2, %xmm0
 ; FMA-INFS-NEXT:    vmulps %xmm0, %xmm1, %xmm0
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v4f32_mul_y_sub_one_x:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA4-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [1,1,1,1]
 ; FMA4-INFS-NEXT:    vsubps %xmm0, %xmm2, %xmm0
 ; FMA4-INFS-NEXT:    vmulps %xmm0, %xmm1, %xmm0
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v4f32_mul_y_sub_one_x:
 ; AVX512-INFS:       # %bb.0:
 ; AVX512-INFS-NEXT:    vbroadcastss {{.*#+}} xmm2 = [1,1,1,1]
 ; AVX512-INFS-NEXT:    vsubps %xmm0, %xmm2, %xmm0
; AVX512-NOINFS-NEXT:    retq
   %s = fsub <4 x float> <float 1.0, float 1.0, float 1.0, float 1.0>, %x
   %m = fmul <4 x float> %y, %s
   ret <4 x float> %m
 }
 
 define <4 x float> @test_v4f32_mul_sub_negone_x_y(<4 x float> %x, <4 x float> %y) {
 ; FMA-INFS-LABEL: test_v4f32_mul_sub_negone_x_y:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [-1,-1,-1,-1]
 ; FMA-INFS-NEXT:    vsubps %xmm0, %xmm2, %xmm0
 ; FMA-INFS-NEXT:    vmulps %xmm1, %xmm0, %xmm0
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v4f32_mul_sub_negone_x_y:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA4-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [-1,-1,-1,-1]
 ; FMA4-INFS-NEXT:    vsubps %xmm0, %xmm2, %xmm0
 ; FMA4-INFS-NEXT:    vmulps %xmm1, %xmm0, %xmm0
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v4f32_mul_sub_negone_x_y:
 ; AVX512-INFS:       # %bb.0:
 ; AVX512-INFS-NEXT:    vbroadcastss {{.*#+}} xmm2 = [-1,-1,-1,-1]
 ; AVX512-INFS-NEXT:    vsubps %xmm0, %xmm2, %xmm0
; AVX512-NOINFS-NEXT:    retq
   %s = fsub <4 x float> <float -1.0, float -1.0, float -1.0, float -1.0>, %x
   %m = fmul <4 x float> %s, %y
   ret <4 x float> %m
 }
 
 define <4 x float> @test_v4f32_mul_y_sub_negone_x(<4 x float> %x, <4 x float> %y) {
 ; FMA-INFS-LABEL: test_v4f32_mul_y_sub_negone_x:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [-1,-1,-1,-1]
 ; FMA-INFS-NEXT:    vsubps %xmm0, %xmm2, %xmm0
 ; FMA-INFS-NEXT:    vmulps %xmm0, %xmm1, %xmm0
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v4f32_mul_y_sub_negone_x:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA4-INFS-NEXT:    vmovaps {{.*#+}} xmm2 = [-1,-1,-1,-1]
 ; FMA4-INFS-NEXT:    vsubps %xmm0, %xmm2, %xmm0
 ; FMA4-INFS-NEXT:    vmulps %xmm0, %xmm1, %xmm0
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v4f32_mul_y_sub_negone_x:
 ; AVX512-INFS:       # %bb.0:
 ; AVX512-INFS-NEXT:    vbroadcastss {{.*#+}} xmm2 = [-1,-1,-1,-1]
 ; AVX512-INFS-NEXT:    vsubps %xmm0, %xmm2, %xmm0
; AVX512-NOINFS-NEXT:    retq
   %ty = fmul float %y, %t1
   %r = fadd float %tx, %ty
   ret float %r
 }
 
 define <4 x float> @test_v4f32_interp(<4 x float> %x, <4 x float> %y, <4 x float> %t) {
 ; FMA-INFS-LABEL: test_v4f32_interp:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-INFS-NEXT:    vmovaps {{.*#+}} xmm3 = [1,1,1,1]
 ; FMA-INFS-NEXT:    vsubps %xmm2, %xmm3, %xmm3
 ; FMA-INFS-NEXT:    vmulps %xmm3, %xmm1, %xmm1
 ; FMA-INFS-NEXT:    vfmadd213ps {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v4f32_interp:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA4-INFS-NEXT:    vmovaps {{.*#+}} xmm3 = [1,1,1,1]
 ; FMA4-INFS-NEXT:    vsubps %xmm2, %xmm3, %xmm3
 ; FMA4-INFS-NEXT:    vmulps %xmm3, %xmm1, %xmm1
 ; FMA4-INFS-NEXT:    vfmaddps %xmm1, %xmm2, %xmm0, %xmm0
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v4f32_interp:
 ; AVX512-INFS:       # %bb.0:
 ; AVX512-INFS-NEXT:    vbroadcastss {{.*#+}} xmm3 = [1,1,1,1]
; AVX512-NOINFS-NEXT:    retq
   %ty = fmul <4 x float> %y, %t1
   %r = fadd <4 x float> %tx, %ty
   ret <4 x float> %r
 }
 
 define <8 x float> @test_v8f32_interp(<8 x float> %x, <8 x float> %y, <8 x float> %t) {
 ; FMA-INFS-LABEL: test_v8f32_interp:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
 ; FMA-INFS-NEXT:    vsubps %ymm2, %ymm3, %ymm3
 ; FMA-INFS-NEXT:    vmulps %ymm3, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    vfmadd213ps {{.*#+}} ymm0 = (ymm2 * ymm0) + ymm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v8f32_interp:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
 ; FMA4-INFS-NEXT:    vsubps %ymm2, %ymm3, %ymm3
 ; FMA4-INFS-NEXT:    vmulps %ymm3, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    vfmaddps %ymm1, %ymm2, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v8f32_interp:
 ; AVX512-INFS:       # %bb.0:
 ; AVX512-INFS-NEXT:    vbroadcastss {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
; AVX512-NOINFS-NEXT:    retq
   %ty = fmul double %y, %t1
   %r = fadd double %tx, %ty
   ret double %r
 }
 
 define <2 x double> @test_v2f64_interp(<2 x double> %x, <2 x double> %y, <2 x double> %t) {
 ; FMA-INFS-LABEL: test_v2f64_interp:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovapd {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00]
+; FMA-INFS-NEXT:    vmovapd {{.*#+}} xmm3 = [1,1]
 ; FMA-INFS-NEXT:    vsubpd %xmm2, %xmm3, %xmm3
 ; FMA-INFS-NEXT:    vmulpd %xmm3, %xmm1, %xmm1
 ; FMA-INFS-NEXT:    vfmadd213pd {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v2f64_interp:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovapd {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00]
+; FMA4-INFS-NEXT:    vmovapd {{.*#+}} xmm3 = [1,1]
 ; FMA4-INFS-NEXT:    vsubpd %xmm2, %xmm3, %xmm3
 ; FMA4-INFS-NEXT:    vmulpd %xmm3, %xmm1, %xmm1
 ; FMA4-INFS-NEXT:    vfmaddpd %xmm1, %xmm2, %xmm0, %xmm0
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v2f64_interp:
 ; AVX512-INFS:       # %bb.0:
-; AVX512-INFS-NEXT:    vmovapd {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00]
+; AVX512-INFS-NEXT:    vmovapd {{.*#+}} xmm3 = [1,1]
 ; AVX512-INFS-NEXT:    vsubpd %xmm2, %xmm3, %xmm3
 ; AVX512-INFS-NEXT:    vmulpd %xmm3, %xmm1, %xmm1
 ; AVX512-INFS-NEXT:    vfmadd213pd {{.*#+}} xmm0 = (xmm2 * xmm0) + xmm1
 ; AVX512-INFS-NEXT:    retq
 ;
 ; FMA-NOINFS-LABEL: test_v2f64_interp:
 ; FMA-NOINFS:       # %bb.0:
 ; FMA-NOINFS-NEXT:    vfnmadd213pd {{.*#+}} xmm1 = -(xmm2 * xmm1) + xmm1
; AVX512-NOINFS-NEXT:    retq
   %ty = fmul <2 x double> %y, %t1
   %r = fadd <2 x double> %tx, %ty
   ret <2 x double> %r
 }
 
 define <4 x double> @test_v4f64_interp(<4 x double> %x, <4 x double> %y, <4 x double> %t) {
 ; FMA-INFS-LABEL: test_v4f64_interp:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm3 = [1,1,1,1]
 ; FMA-INFS-NEXT:    vsubpd %ymm2, %ymm3, %ymm3
 ; FMA-INFS-NEXT:    vmulpd %ymm3, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    vfmadd213pd {{.*#+}} ymm0 = (ymm2 * ymm0) + ymm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v4f64_interp:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm3 = [1,1,1,1]
 ; FMA4-INFS-NEXT:    vsubpd %ymm2, %ymm3, %ymm3
 ; FMA4-INFS-NEXT:    vmulpd %ymm3, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    vfmaddpd %ymm1, %ymm2, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v4f64_interp:
 ; AVX512-INFS:       # %bb.0:
 ; AVX512-INFS-NEXT:    vbroadcastsd {{.*#+}} ymm3 = [1,1,1,1]

llvm/trunk/test/CodeGen/X86/fma_patterns_wide.ll

 
 ;
 ; Patterns (+ fneg variants): mul(add(1.0,x),y), mul(sub(1.0,x),y), mul(sub(x,1.0),y)
 ;
 
 define <16 x float> @test_v16f32_mul_add_x_one_y(<16 x float> %x, <16 x float> %y) {
 ; FMA-INFS-LABEL: test_v16f32_mul_add_x_one_y:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
 ; FMA-INFS-NEXT:    vaddps %ymm4, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    vaddps %ymm4, %ymm0, %ymm0
 ; FMA-INFS-NEXT:    vmulps %ymm2, %ymm0, %ymm0
 ; FMA-INFS-NEXT:    vmulps %ymm3, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v16f32_mul_add_x_one_y:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
 ; FMA4-INFS-NEXT:    vaddps %ymm4, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    vaddps %ymm4, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vmulps %ymm2, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vmulps %ymm3, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v16f32_mul_add_x_one_y:
 ; AVX512-INFS:       # %bb.0:
; AVX512-NOINFS-NEXT:    retq
   %a = fadd <16 x float> %x, <float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>
   %m = fmul <16 x float> %a, %y
   ret <16 x float> %m
 }
 
 define <8 x double> @test_v8f64_mul_y_add_x_one(<8 x double> %x, <8 x double> %y) {
 ; FMA-INFS-LABEL: test_v8f64_mul_y_add_x_one:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [1,1,1,1]
 ; FMA-INFS-NEXT:    vaddpd %ymm4, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    vaddpd %ymm4, %ymm0, %ymm0
 ; FMA-INFS-NEXT:    vmulpd %ymm0, %ymm2, %ymm0
 ; FMA-INFS-NEXT:    vmulpd %ymm1, %ymm3, %ymm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v8f64_mul_y_add_x_one:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [1,1,1,1]
 ; FMA4-INFS-NEXT:    vaddpd %ymm4, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    vaddpd %ymm4, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vmulpd %ymm0, %ymm2, %ymm0
 ; FMA4-INFS-NEXT:    vmulpd %ymm1, %ymm3, %ymm1
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v8f64_mul_y_add_x_one:
 ; AVX512-INFS:       # %bb.0:
; AVX512-NOINFS-NEXT:    retq
   %a = fadd <8 x double> %x, <double 1.0, double 1.0, double 1.0, double 1.0, double 1.0, double 1.0, double 1.0, double 1.0>
   %m = fmul <8 x double> %y, %a
   ret <8 x double> %m
 }
 
 define <16 x float> @test_v16f32_mul_add_x_negone_y(<16 x float> %x, <16 x float> %y) {
 ; FMA-INFS-LABEL: test_v16f32_mul_add_x_negone_y:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [-1,-1,-1,-1,-1,-1,-1,-1]
 ; FMA-INFS-NEXT:    vaddps %ymm4, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    vaddps %ymm4, %ymm0, %ymm0
 ; FMA-INFS-NEXT:    vmulps %ymm2, %ymm0, %ymm0
 ; FMA-INFS-NEXT:    vmulps %ymm3, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v16f32_mul_add_x_negone_y:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [-1,-1,-1,-1,-1,-1,-1,-1]
 ; FMA4-INFS-NEXT:    vaddps %ymm4, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    vaddps %ymm4, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vmulps %ymm2, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vmulps %ymm3, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v16f32_mul_add_x_negone_y:
 ; AVX512-INFS:       # %bb.0:
; AVX512-NOINFS-NEXT:    retq
   %a = fadd <16 x float> %x, <float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0>
   %m = fmul <16 x float> %a, %y
   ret <16 x float> %m
 }
 
 define <8 x double> @test_v8f64_mul_y_add_x_negone(<8 x double> %x, <8 x double> %y) {
 ; FMA-INFS-LABEL: test_v8f64_mul_y_add_x_negone:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [-1,-1,-1,-1]
 ; FMA-INFS-NEXT:    vaddpd %ymm4, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    vaddpd %ymm4, %ymm0, %ymm0
 ; FMA-INFS-NEXT:    vmulpd %ymm0, %ymm2, %ymm0
 ; FMA-INFS-NEXT:    vmulpd %ymm1, %ymm3, %ymm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v8f64_mul_y_add_x_negone:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [-1,-1,-1,-1]
 ; FMA4-INFS-NEXT:    vaddpd %ymm4, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    vaddpd %ymm4, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vmulpd %ymm0, %ymm2, %ymm0
 ; FMA4-INFS-NEXT:    vmulpd %ymm1, %ymm3, %ymm1
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v8f64_mul_y_add_x_negone:
 ; AVX512-INFS:       # %bb.0:
; AVX512-NOINFS-NEXT:    retq
   %a = fadd <8 x double> %x, <double -1.0, double -1.0, double -1.0, double -1.0, double -1.0, double -1.0, double -1.0, double -1.0>
   %m = fmul <8 x double> %y, %a
   ret <8 x double> %m
 }
 
 define <16 x float> @test_v16f32_mul_sub_one_x_y(<16 x float> %x, <16 x float> %y) {
 ; FMA-INFS-LABEL: test_v16f32_mul_sub_one_x_y:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
 ; FMA-INFS-NEXT:    vsubps %ymm1, %ymm4, %ymm1
 ; FMA-INFS-NEXT:    vsubps %ymm0, %ymm4, %ymm0
 ; FMA-INFS-NEXT:    vmulps %ymm2, %ymm0, %ymm0
 ; FMA-INFS-NEXT:    vmulps %ymm3, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v16f32_mul_sub_one_x_y:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
 ; FMA4-INFS-NEXT:    vsubps %ymm1, %ymm4, %ymm1
 ; FMA4-INFS-NEXT:    vsubps %ymm0, %ymm4, %ymm0
 ; FMA4-INFS-NEXT:    vmulps %ymm2, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vmulps %ymm3, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v16f32_mul_sub_one_x_y:
 ; AVX512-INFS:       # %bb.0:
; AVX512-NOINFS-NEXT:    retq
   %s = fsub <16 x float> <float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>, %x
   %m = fmul <16 x float> %s, %y
   ret <16 x float> %m
 }
 
 define <8 x double> @test_v8f64_mul_y_sub_one_x(<8 x double> %x, <8 x double> %y) {
 ; FMA-INFS-LABEL: test_v8f64_mul_y_sub_one_x:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [1,1,1,1]
 ; FMA-INFS-NEXT:    vsubpd %ymm1, %ymm4, %ymm1
 ; FMA-INFS-NEXT:    vsubpd %ymm0, %ymm4, %ymm0
 ; FMA-INFS-NEXT:    vmulpd %ymm0, %ymm2, %ymm0
 ; FMA-INFS-NEXT:    vmulpd %ymm1, %ymm3, %ymm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v8f64_mul_y_sub_one_x:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [1,1,1,1]
 ; FMA4-INFS-NEXT:    vsubpd %ymm1, %ymm4, %ymm1
 ; FMA4-INFS-NEXT:    vsubpd %ymm0, %ymm4, %ymm0
 ; FMA4-INFS-NEXT:    vmulpd %ymm0, %ymm2, %ymm0
 ; FMA4-INFS-NEXT:    vmulpd %ymm1, %ymm3, %ymm1
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v8f64_mul_y_sub_one_x:
 ; AVX512-INFS:       # %bb.0:
; AVX512-NOINFS-NEXT:    retq
   %s = fsub <8 x double> <double 1.0, double 1.0, double 1.0, double 1.0, double 1.0, double 1.0, double 1.0, double 1.0>, %x
   %m = fmul <8 x double> %y, %s
   ret <8 x double> %m
 }
 
 define <16 x float> @test_v16f32_mul_sub_negone_x_y(<16 x float> %x, <16 x float> %y) {
 ; FMA-INFS-LABEL: test_v16f32_mul_sub_negone_x_y:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [-1,-1,-1,-1,-1,-1,-1,-1]
 ; FMA-INFS-NEXT:    vsubps %ymm1, %ymm4, %ymm1
 ; FMA-INFS-NEXT:    vsubps %ymm0, %ymm4, %ymm0
 ; FMA-INFS-NEXT:    vmulps %ymm2, %ymm0, %ymm0
 ; FMA-INFS-NEXT:    vmulps %ymm3, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v16f32_mul_sub_negone_x_y:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [-1,-1,-1,-1,-1,-1,-1,-1]
 ; FMA4-INFS-NEXT:    vsubps %ymm1, %ymm4, %ymm1
 ; FMA4-INFS-NEXT:    vsubps %ymm0, %ymm4, %ymm0
 ; FMA4-INFS-NEXT:    vmulps %ymm2, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vmulps %ymm3, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v16f32_mul_sub_negone_x_y:
 ; AVX512-INFS:       # %bb.0:
; AVX512-NOINFS-NEXT:    retq
   %s = fsub <16 x float> <float -1.0, float -1.0, float -1.0, float -1.0,float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0>, %x
   %m = fmul <16 x float> %s, %y
   ret <16 x float> %m
 }
 
 define <8 x double> @test_v8f64_mul_y_sub_negone_x(<8 x double> %x, <8 x double> %y) {
 ; FMA-INFS-LABEL: test_v8f64_mul_y_sub_negone_x:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [-1,-1,-1,-1]
 ; FMA-INFS-NEXT:    vsubpd %ymm1, %ymm4, %ymm1
 ; FMA-INFS-NEXT:    vsubpd %ymm0, %ymm4, %ymm0
 ; FMA-INFS-NEXT:    vmulpd %ymm0, %ymm2, %ymm0
 ; FMA-INFS-NEXT:    vmulpd %ymm1, %ymm3, %ymm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v8f64_mul_y_sub_negone_x:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [-1,-1,-1,-1]
 ; FMA4-INFS-NEXT:    vsubpd %ymm1, %ymm4, %ymm1
 ; FMA4-INFS-NEXT:    vsubpd %ymm0, %ymm4, %ymm0
 ; FMA4-INFS-NEXT:    vmulpd %ymm0, %ymm2, %ymm0
 ; FMA4-INFS-NEXT:    vmulpd %ymm1, %ymm3, %ymm1
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v8f64_mul_y_sub_negone_x:
 ; AVX512-INFS:       # %bb.0:
; AVX512-NOINFS-NEXT:    retq
   %s = fsub <8 x double> <double -1.0, double -1.0, double -1.0, double -1.0, double -1.0, double -1.0, double -1.0, double -1.0>, %x
   %m = fmul <8 x double> %y, %s
   ret <8 x double> %m
 }
 
 define <16 x float> @test_v16f32_mul_sub_x_one_y(<16 x float> %x, <16 x float> %y) {
 ; FMA-INFS-LABEL: test_v16f32_mul_sub_x_one_y:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
 ; FMA-INFS-NEXT:    vsubps %ymm4, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    vsubps %ymm4, %ymm0, %ymm0
 ; FMA-INFS-NEXT:    vmulps %ymm2, %ymm0, %ymm0
 ; FMA-INFS-NEXT:    vmulps %ymm3, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v16f32_mul_sub_x_one_y:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
 ; FMA4-INFS-NEXT:    vsubps %ymm4, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    vsubps %ymm4, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vmulps %ymm2, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vmulps %ymm3, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v16f32_mul_sub_x_one_y:
 ; AVX512-INFS:       # %bb.0:
; AVX512-NOINFS-NEXT:    retq
   %s = fsub <16 x float> %x, <float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>
   %m = fmul <16 x float> %s, %y
   ret <16 x float> %m
 }
 
 define <8 x double> @test_v8f64_mul_y_sub_x_one(<8 x double> %x, <8 x double> %y) {
 ; FMA-INFS-LABEL: test_v8f64_mul_y_sub_x_one:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [1,1,1,1]
 ; FMA-INFS-NEXT:    vsubpd %ymm4, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    vsubpd %ymm4, %ymm0, %ymm0
 ; FMA-INFS-NEXT:    vmulpd %ymm0, %ymm2, %ymm0
 ; FMA-INFS-NEXT:    vmulpd %ymm1, %ymm3, %ymm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v8f64_mul_y_sub_x_one:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [1,1,1,1]
 ; FMA4-INFS-NEXT:    vsubpd %ymm4, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    vsubpd %ymm4, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vmulpd %ymm0, %ymm2, %ymm0
 ; FMA4-INFS-NEXT:    vmulpd %ymm1, %ymm3, %ymm1
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v8f64_mul_y_sub_x_one:
 ; AVX512-INFS:       # %bb.0:
; AVX512-NOINFS-NEXT:    retq
   %s = fsub <8 x double> %x, <double 1.0, double 1.0, double 1.0, double 1.0, double 1.0, double 1.0, double 1.0, double 1.0>
   %m = fmul <8 x double> %y, %s
   ret <8 x double> %m
 }
 
 define <16 x float> @test_v16f32_mul_sub_x_negone_y(<16 x float> %x, <16 x float> %y) {
 ; FMA-INFS-LABEL: test_v16f32_mul_sub_x_negone_y:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [-1,-1,-1,-1,-1,-1,-1,-1]
 ; FMA-INFS-NEXT:    vsubps %ymm4, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    vsubps %ymm4, %ymm0, %ymm0
 ; FMA-INFS-NEXT:    vmulps %ymm2, %ymm0, %ymm0
 ; FMA-INFS-NEXT:    vmulps %ymm3, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v16f32_mul_sub_x_negone_y:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm4 = [-1,-1,-1,-1,-1,-1,-1,-1]
 ; FMA4-INFS-NEXT:    vsubps %ymm4, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    vsubps %ymm4, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vmulps %ymm2, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vmulps %ymm3, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v16f32_mul_sub_x_negone_y:
 ; AVX512-INFS:       # %bb.0:
; AVX512-NOINFS-NEXT:    retq
   %s = fsub <16 x float> %x, <float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0, float -1.0>
   %m = fmul <16 x float> %s, %y
   ret <16 x float> %m
 }
 
 define <8 x double> @test_v8f64_mul_y_sub_x_negone(<8 x double> %x, <8 x double> %y) {
 ; FMA-INFS-LABEL: test_v8f64_mul_y_sub_x_negone:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [-1,-1,-1,-1]
 ; FMA-INFS-NEXT:    vsubpd %ymm4, %ymm1, %ymm1
 ; FMA-INFS-NEXT:    vsubpd %ymm4, %ymm0, %ymm0
 ; FMA-INFS-NEXT:    vmulpd %ymm0, %ymm2, %ymm0
 ; FMA-INFS-NEXT:    vmulpd %ymm1, %ymm3, %ymm1
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v8f64_mul_y_sub_x_negone:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [-1.000000e+00,-1.000000e+00,-1.000000e+00,-1.000000e+00]
+; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm4 = [-1,-1,-1,-1]
 ; FMA4-INFS-NEXT:    vsubpd %ymm4, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    vsubpd %ymm4, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vmulpd %ymm0, %ymm2, %ymm0
 ; FMA4-INFS-NEXT:    vmulpd %ymm1, %ymm3, %ymm1
 ; FMA4-INFS-NEXT:    retq
 ;
 ; AVX512-INFS-LABEL: test_v8f64_mul_y_sub_x_negone:
 ; AVX512-INFS:       # %bb.0:

 
 ;
 ; Interpolation Patterns: add(mul(x,t),mul(sub(1.0,t),y))
 ;
 
 define <16 x float> @test_v16f32_interp(<16 x float> %x, <16 x float> %y, <16 x float> %t) {
 ; FMA-INFS-LABEL: test_v16f32_interp:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm6 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-INFS-NEXT:    vmovaps {{.*#+}} ymm6 = [1,1,1,1,1,1,1,1]
 ; FMA-INFS-NEXT:    vsubps %ymm4, %ymm6, %ymm7
 ; FMA-INFS-NEXT:    vsubps %ymm5, %ymm6, %ymm6
 ; FMA-INFS-NEXT:    vmulps %ymm6, %ymm3, %ymm3
 ; FMA-INFS-NEXT:    vmulps %ymm7, %ymm2, %ymm2
 ; FMA-INFS-NEXT:    vfmadd213ps {{.*#+}} ymm0 = (ymm4 * ymm0) + ymm2
 ; FMA-INFS-NEXT:    vfmadd213ps {{.*#+}} ymm1 = (ymm5 * ymm1) + ymm3
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v16f32_interp:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm6 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA4-INFS-NEXT:    vmovaps {{.*#+}} ymm6 = [1,1,1,1,1,1,1,1]
 ; FMA4-INFS-NEXT:    vsubps %ymm4, %ymm6, %ymm7
 ; FMA4-INFS-NEXT:    vsubps %ymm5, %ymm6, %ymm6
 ; FMA4-INFS-NEXT:    vmulps %ymm6, %ymm3, %ymm3
 ; FMA4-INFS-NEXT:    vmulps %ymm7, %ymm2, %ymm2
 ; FMA4-INFS-NEXT:    vfmaddps %ymm2, %ymm4, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vfmaddps %ymm3, %ymm5, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    retq
 ;
; AVX512-NOINFS-NEXT:    retq
   %ty = fmul <16 x float> %y, %t1
   %r = fadd <16 x float> %tx, %ty
   ret <16 x float> %r
 }
 
 define <8 x double> @test_v8f64_interp(<8 x double> %x, <8 x double> %y, <8 x double> %t) {
 ; FMA-INFS-LABEL: test_v8f64_interp:
 ; FMA-INFS:       # %bb.0:
-; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm6 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-INFS-NEXT:    vmovapd {{.*#+}} ymm6 = [1,1,1,1]
 ; FMA-INFS-NEXT:    vsubpd %ymm4, %ymm6, %ymm7
 ; FMA-INFS-NEXT:    vsubpd %ymm5, %ymm6, %ymm6
 ; FMA-INFS-NEXT:    vmulpd %ymm6, %ymm3, %ymm3
 ; FMA-INFS-NEXT:    vmulpd %ymm7, %ymm2, %ymm2
 ; FMA-INFS-NEXT:    vfmadd213pd {{.*#+}} ymm0 = (ymm4 * ymm0) + ymm2
 ; FMA-INFS-NEXT:    vfmadd213pd {{.*#+}} ymm1 = (ymm5 * ymm1) + ymm3
 ; FMA-INFS-NEXT:    retq
 ;
 ; FMA4-INFS-LABEL: test_v8f64_interp:
 ; FMA4-INFS:       # %bb.0:
-; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm6 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA4-INFS-NEXT:    vmovapd {{.*#+}} ymm6 = [1,1,1,1]
 ; FMA4-INFS-NEXT:    vsubpd %ymm4, %ymm6, %ymm7
 ; FMA4-INFS-NEXT:    vsubpd %ymm5, %ymm6, %ymm6
 ; FMA4-INFS-NEXT:    vmulpd %ymm6, %ymm3, %ymm3
 ; FMA4-INFS-NEXT:    vmulpd %ymm7, %ymm2, %ymm2
 ; FMA4-INFS-NEXT:    vfmaddpd %ymm2, %ymm4, %ymm0, %ymm0
 ; FMA4-INFS-NEXT:    vfmaddpd %ymm3, %ymm5, %ymm1, %ymm1
 ; FMA4-INFS-NEXT:    retq
 ;
; AVX512-NEXT:    retq
   ret <8 x double> %n
 }
 
 define <8 x double> @test_v8f64_fneg_fmul_no_nsz(<8 x double> %x, <8 x double> %y) #0 {
 ; FMA-LABEL: test_v8f64_fneg_fmul_no_nsz:
 ; FMA:       # %bb.0:
 ; FMA-NEXT:    vmulpd %ymm3, %ymm1, %ymm1
 ; FMA-NEXT:    vmulpd %ymm2, %ymm0, %ymm0
-; FMA-NEXT:    vmovapd {{.*#+}} ymm2 = [-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00]
+; FMA-NEXT:    vmovapd {{.*#+}} ymm2 = [-0,-0,-0,-0]
 ; FMA-NEXT:    vxorpd %ymm2, %ymm0, %ymm0
 ; FMA-NEXT:    vxorpd %ymm2, %ymm1, %ymm1
 ; FMA-NEXT:    retq
 ;
 ; FMA4-LABEL: test_v8f64_fneg_fmul_no_nsz:
 ; FMA4:       # %bb.0:
 ; FMA4-NEXT:    vmulpd %ymm3, %ymm1, %ymm1
 ; FMA4-NEXT:    vmulpd %ymm2, %ymm0, %ymm0
-; FMA4-NEXT:    vmovapd {{.*#+}} ymm2 = [-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00]
+; FMA4-NEXT:    vmovapd {{.*#+}} ymm2 = [-0,-0,-0,-0]
 ; FMA4-NEXT:    vxorpd %ymm2, %ymm0, %ymm0
 ; FMA4-NEXT:    vxorpd %ymm2, %ymm1, %ymm1
 ; FMA4-NEXT:    retq
 ;
 ; AVX512-LABEL: test_v8f64_fneg_fmul_no_nsz:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vmulpd %zmm1, %zmm0, %zmm0
 ; AVX512-NEXT:    vxorpd {{.*}}(%rip){1to8}, %zmm0, %zmm0
 ; AVX512-NEXT:    retq
   %m = fmul <8 x double> %x, %y
   %n = fsub <8 x double> <double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0, double -0.0>, %m
   ret <8 x double> %n
 }
 
 attributes #0 = { "unsafe-fp-math"="true" }

llvm/trunk/test/CodeGen/X86/fmul-combines.ll

 ; CHECK-NEXT:    retq
   %y = fmul <4 x float> %x, <float 2.0, float 2.0, float 2.0, float 2.0>
   ret <4 x float> %y
 }
 
 define <4 x float> @constant_fold_fmul_v4f32(<4 x float> %x) {
 ; CHECK-LABEL: constant_fold_fmul_v4f32:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    movaps {{.*#+}} xmm0 = [8.000000e+00,8.000000e+00,8.000000e+00,8.000000e+00]
+; CHECK-NEXT:    movaps {{.*#+}} xmm0 = [8,8,8,8]
 ; CHECK-NEXT:    retq
   %y = fmul <4 x float> <float 4.0, float 4.0, float 4.0, float 4.0>, <float 2.0, float 2.0, float 2.0, float 2.0>
   ret <4 x float> %y
 }
 
 define <4 x float> @fmul0_v4f32(<4 x float> %x) #0 {
 ; CHECK-LABEL: fmul0_v4f32:
 ; CHECK:       # %bb.0:

llvm/trunk/test/CodeGen/X86/fold-vector-trunc-sitofp.ll

+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=i686-unknown -mattr=+avx | FileCheck %s
 ; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx | FileCheck %s
 
 ; Check that constant integers are correctly being truncated before float conversion
 
 define <4 x float> @test1() {
-; CHECK-LABEL: test1
-; CHECK: movaps {{.*#+}} xmm0 = [-1.000000e+00,0.000000e+00,-1.000000e+00,0.000000e+00]
-; CHECK-NEXT: ret
+; CHECK-LABEL: test1:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vmovaps {{.*#+}} xmm0 = [-1,0,-1,0]
+; CHECK-NEXT:    ret{{[l|q]}}
   %1 = trunc <4 x i3> <i3 -1, i3 -22, i3 7, i3 8> to <4 x i1>
   %2 = sitofp <4 x i1> %1 to <4 x float>
   ret <4 x float> %2
 }

llvm/trunk/test/CodeGen/X86/ftrunc.ll

 ; SSE2-NEXT:    cvttsd2si %xmm1, %rax
 ; SSE2-NEXT:    xorq %rcx, %rax
 ; SSE2-NEXT:    cvttsd2si %xmm0, %rcx
 ; SSE2-NEXT:    ucomisd %xmm2, %xmm0
 ; SSE2-NEXT:    cmovaeq %rax, %rcx
 ; SSE2-NEXT:    movq %rcx, %xmm1
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [1127219200,1160773632,0,0]
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSE2-NEXT:    movapd {{.*#+}} xmm3 = [4.503600e+15,1.934281e+25]
+; SSE2-NEXT:    movapd {{.*#+}} xmm3 = [4503599627370496,1.9342813113834067E+25]
 ; SSE2-NEXT:    subpd %xmm3, %xmm1
 ; SSE2-NEXT:    movapd %xmm1, %xmm0
 ; SSE2-NEXT:    unpckhpd {{.*#+}} xmm0 = xmm0[1],xmm1[1]
 ; SSE2-NEXT:    addpd %xmm1, %xmm0
 ; SSE2-NEXT:    movq %rdx, %xmm1
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
 ; SSE2-NEXT:    subpd %xmm3, %xmm1
 ; SSE2-NEXT:    movapd %xmm1, %xmm2

 ; SSE2-NEXT:    cvttsd2si %xmm1, %rsi
 ; SSE2-NEXT:    xorq %rdx, %rsi
 ; SSE2-NEXT:    cvttsd2si %xmm0, %rdx
 ; SSE2-NEXT:    ucomisd %xmm2, %xmm0
 ; SSE2-NEXT:    cmovaeq %rsi, %rdx
 ; SSE2-NEXT:    movq %rdx, %xmm1
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [1127219200,1160773632,0,0]
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSE2-NEXT:    movapd {{.*#+}} xmm3 = [4.503600e+15,1.934281e+25]
+; SSE2-NEXT:    movapd {{.*#+}} xmm3 = [4503599627370496,1.9342813113834067E+25]
 ; SSE2-NEXT:    subpd %xmm3, %xmm1
 ; SSE2-NEXT:    movapd %xmm1, %xmm0
 ; SSE2-NEXT:    unpckhpd {{.*#+}} xmm0 = xmm0[1],xmm1[1]
 ; SSE2-NEXT:    addpd %xmm1, %xmm0
 ; SSE2-NEXT:    movq %rdi, %xmm1
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
 ; SSE2-NEXT:    subpd %xmm3, %xmm1
 ; SSE2-NEXT:    movapd %xmm1, %xmm4

llvm/trunk/test/CodeGen/X86/insert-into-constant-vector.ll

 
 define <8 x float> @elt6_v8f32(float %x) {
 ; X32SSE2-LABEL: elt6_v8f32:
 ; X32SSE2:       # %bb.0:
 ; X32SSE2-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; X32SSE2-NEXT:    movaps {{.*#+}} xmm1 = <4,5,u,7>
 ; X32SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[3,0]
 ; X32SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,1],xmm0[0,2]
-; X32SSE2-NEXT:    movaps {{.*#+}} xmm0 = [4.200000e+01,1.000000e+00,2.000000e+00,3.000000e+00]
+; X32SSE2-NEXT:    movaps {{.*#+}} xmm0 = [42,1,2,3]
 ; X32SSE2-NEXT:    retl
 ;
 ; X64SSE2-LABEL: elt6_v8f32:
 ; X64SSE2:       # %bb.0:
 ; X64SSE2-NEXT:    movaps {{.*#+}} xmm1 = <4,5,u,7>
 ; X64SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[3,0]
 ; X64SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,1],xmm0[0,2]
-; X64SSE2-NEXT:    movaps {{.*#+}} xmm0 = [4.200000e+01,1.000000e+00,2.000000e+00,3.000000e+00]
+; X64SSE2-NEXT:    movaps {{.*#+}} xmm0 = [42,1,2,3]
 ; X64SSE2-NEXT:    retq
 ;
 ; X32SSE4-LABEL: elt6_v8f32:
 ; X32SSE4:       # %bb.0:
 ; X32SSE4-NEXT:    movaps {{.*#+}} xmm1 = <4,5,u,7>
 ; X32SSE4-NEXT:    insertps {{.*#+}} xmm1 = xmm1[0,1],mem[0],xmm1[3]
-; X32SSE4-NEXT:    movaps {{.*#+}} xmm0 = [4.200000e+01,1.000000e+00,2.000000e+00,3.000000e+00]
+; X32SSE4-NEXT:    movaps {{.*#+}} xmm0 = [42,1,2,3]
 ; X32SSE4-NEXT:    retl
 ;
 ; X64SSE4-LABEL: elt6_v8f32:
 ; X64SSE4:       # %bb.0:
 ; X64SSE4-NEXT:    movaps {{.*#+}} xmm1 = <4,5,u,7>
 ; X64SSE4-NEXT:    insertps {{.*#+}} xmm1 = xmm1[0,1],xmm0[0],xmm1[3]
-; X64SSE4-NEXT:    movaps {{.*#+}} xmm0 = [4.200000e+01,1.000000e+00,2.000000e+00,3.000000e+00]
+; X64SSE4-NEXT:    movaps {{.*#+}} xmm0 = [42,1,2,3]
 ; X64SSE4-NEXT:    retq
 ;
 ; X32AVX-LABEL: elt6_v8f32:
 ; X32AVX:       # %bb.0:
 ; X32AVX-NEXT:    vmovaps {{.*#+}} ymm0 = <42,1,2,3,4,5,u,7>
 ; X32AVX-NEXT:    vextractf128 $1, %ymm0, %xmm1
 ; X32AVX-NEXT:    vinsertps {{.*#+}} xmm1 = xmm1[0,1],mem[0],xmm1[3]
 ; X32AVX-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
; X64AVX512F-NEXT:    retq
    ret <8 x i64> %ins
 }
 
 define <8 x double> @elt1_v8f64(double %x) {
 ; X32SSE-LABEL: elt1_v8f64:
 ; X32SSE:       # %bb.0:
 ; X32SSE-NEXT:    movapd {{.*#+}} xmm0 = <42,u>
 ; X32SSE-NEXT:    movhpd {{.*#+}} xmm0 = xmm0[0],mem[0]
-; X32SSE-NEXT:    movaps {{.*#+}} xmm1 = [2.000000e+00,3.000000e+00]
-; X32SSE-NEXT:    movaps {{.*#+}} xmm2 = [4.000000e+00,5.000000e+00]
-; X32SSE-NEXT:    movaps {{.*#+}} xmm3 = [6.000000e+00,7.000000e+00]
+; X32SSE-NEXT:    movaps {{.*#+}} xmm1 = [2,3]
+; X32SSE-NEXT:    movaps {{.*#+}} xmm2 = [4,5]
+; X32SSE-NEXT:    movaps {{.*#+}} xmm3 = [6,7]
 ; X32SSE-NEXT:    retl
 ;
 ; X64SSE-LABEL: elt1_v8f64:
 ; X64SSE:       # %bb.0:
 ; X64SSE-NEXT:    movaps {{.*#+}} xmm4 = <42,u>
 ; X64SSE-NEXT:    movlhps {{.*#+}} xmm4 = xmm4[0],xmm0[0]
-; X64SSE-NEXT:    movaps {{.*#+}} xmm1 = [2.000000e+00,3.000000e+00]
-; X64SSE-NEXT:    movaps {{.*#+}} xmm2 = [4.000000e+00,5.000000e+00]
-; X64SSE-NEXT:    movaps {{.*#+}} xmm3 = [6.000000e+00,7.000000e+00]
+; X64SSE-NEXT:    movaps {{.*#+}} xmm1 = [2,3]
+; X64SSE-NEXT:    movaps {{.*#+}} xmm2 = [4,5]
+; X64SSE-NEXT:    movaps {{.*#+}} xmm3 = [6,7]
 ; X64SSE-NEXT:    movaps %xmm4, %xmm0
 ; X64SSE-NEXT:    retq
 ;
 ; X32AVX2-LABEL: elt1_v8f64:
 ; X32AVX2:       # %bb.0:
 ; X32AVX2-NEXT:    vmovapd {{.*#+}} ymm0 = <42,u,2,3>
 ; X32AVX2-NEXT:    vmovhpd {{.*#+}} xmm1 = xmm0[0],mem[0]
 ; X32AVX2-NEXT:    vblendpd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3]
-; X32AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = [4.000000e+00,5.000000e+00,6.000000e+00,7.000000e+00]
+; X32AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = [4,5,6,7]
 ; X32AVX2-NEXT:    retl
 ;
 ; X64AVX2-LABEL: elt1_v8f64:
 ; X64AVX2:       # %bb.0:
 ; X64AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = <42,u,2,3>
 ; X64AVX2-NEXT:    vmovlhps {{.*#+}} xmm0 = xmm1[0],xmm0[0]
 ; X64AVX2-NEXT:    vblendps {{.*#+}} ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
-; X64AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = [4.000000e+00,5.000000e+00,6.000000e+00,7.000000e+00]
+; X64AVX2-NEXT:    vmovaps {{.*#+}} ymm1 = [4,5,6,7]
 ; X64AVX2-NEXT:    retq
 ;
 ; X32AVX512F-LABEL: elt1_v8f64:
 ; X32AVX512F:       # %bb.0:
 ; X32AVX512F-NEXT:    vmovapd {{.*#+}} zmm0 = <42,u,2,3,4,5,6,7>
 ; X32AVX512F-NEXT:    vmovhpd {{.*#+}} xmm1 = xmm0[0],mem[0]
 ; X32AVX512F-NEXT:    vinsertf32x4 $0, %xmm1, %zmm0, %zmm0
 ; X32AVX512F-NEXT:    retl

llvm/trunk/test/CodeGen/X86/packss.ll

 ; X86-SSE-NEXT:    movdqa %xmm1, %xmm3
 ; X86-SSE-NEXT:    movsd {{.*#+}} xmm3 = xmm2[0],xmm3[1]
 ; X86-SSE-NEXT:    movdqa %xmm0, %xmm2
 ; X86-SSE-NEXT:    psllq $63, %xmm2
 ; X86-SSE-NEXT:    movdqa %xmm0, %xmm4
 ; X86-SSE-NEXT:    movsd {{.*#+}} xmm4 = xmm2[0],xmm4[1]
 ; X86-SSE-NEXT:    psrlq $63, %xmm4
 ; X86-SSE-NEXT:    movsd {{.*#+}} xmm0 = xmm4[0],xmm0[1]
-; X86-SSE-NEXT:    movapd {{.*#+}} xmm2 = [4.940656e-324,-0.000000e+00]
+; X86-SSE-NEXT:    movapd {{.*#+}} xmm2 = [4.9406564584124654E-324,-0]
 ; X86-SSE-NEXT:    xorpd %xmm2, %xmm0
 ; X86-SSE-NEXT:    psubq %xmm2, %xmm0
 ; X86-SSE-NEXT:    psrlq $63, %xmm3
 ; X86-SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm3[0],xmm1[1]
 ; X86-SSE-NEXT:    xorpd %xmm2, %xmm1
 ; X86-SSE-NEXT:    psubq %xmm2, %xmm1
 ; X86-SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,0,0]
 ; X86-SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,0,0,0]

llvm/trunk/test/CodeGen/X86/pow.ll

 }
 
 define <4 x float> @pow_v4f32_one_fourth_fmf(<4 x float> %x) nounwind {
 ; CHECK-LABEL: pow_v4f32_one_fourth_fmf:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    rsqrtps %xmm0, %xmm1
 ; CHECK-NEXT:    movaps %xmm0, %xmm2
 ; CHECK-NEXT:    mulps %xmm1, %xmm2
-; CHECK-NEXT:    movaps {{.*#+}} xmm3 = [-5.000000e-01,-5.000000e-01,-5.000000e-01,-5.000000e-01]
+; CHECK-NEXT:    movaps {{.*#+}} xmm3 = [-0.5,-0.5,-0.5,-0.5]
 ; CHECK-NEXT:    movaps %xmm2, %xmm4
 ; CHECK-NEXT:    mulps %xmm3, %xmm4
 ; CHECK-NEXT:    mulps %xmm1, %xmm2
-; CHECK-NEXT:    movaps {{.*#+}} xmm1 = [-3.000000e+00,-3.000000e+00,-3.000000e+00,-3.000000e+00]
+; CHECK-NEXT:    movaps {{.*#+}} xmm1 = [-3,-3,-3,-3]
 ; CHECK-NEXT:    addps %xmm1, %xmm2
 ; CHECK-NEXT:    mulps %xmm4, %xmm2
 ; CHECK-NEXT:    xorps %xmm4, %xmm4
 ; CHECK-NEXT:    cmpneqps %xmm4, %xmm0
 ; CHECK-NEXT:    andps %xmm2, %xmm0
 ; CHECK-NEXT:    rsqrtps %xmm0, %xmm2
 ; CHECK-NEXT:    movaps %xmm0, %xmm5
 ; CHECK-NEXT:    mulps %xmm2, %xmm5

llvm/trunk/test/CodeGen/X86/pr2656.ll

 ; but we do not. It has more than one use, so it gets loaded into a register.
 
 define void @foo(%struct.anon* byval %p) nounwind {
 ; CHECK-LABEL: foo:
 ; CHECK:       ## %bb.0: ## %entry
 ; CHECK-NEXT:    subl $28, %esp
 ; CHECK-NEXT:    movss {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; CHECK-NEXT:    movss {{.*#+}} xmm1 = mem[0],zero,zero,zero
-; CHECK-NEXT:    movaps {{.*#+}} xmm2 = [-0.000000e+00,-0.000000e+00,-0.000000e+00,-0.000000e+00]
+; CHECK-NEXT:    movaps {{.*#+}} xmm2 = [-0,-0,-0,-0]
 ; CHECK-NEXT:    xorps %xmm2, %xmm0
 ; CHECK-NEXT:    cvtss2sd %xmm0, %xmm0
 ; CHECK-NEXT:    xorps %xmm2, %xmm1
 ; CHECK-NEXT:    cvtss2sd %xmm1, %xmm1
 ; CHECK-NEXT:    movsd %xmm1, {{[0-9]+}}(%esp)
 ; CHECK-NEXT:    movsd %xmm0, {{[0-9]+}}(%esp)
 ; CHECK-NEXT:    movl $_.str, (%esp)
 ; CHECK-NEXT:    calll _printf

llvm/trunk/test/CodeGen/X86/pr38639.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=x86_64-unknown -mcpu=btver2 | FileCheck %s
 
 define <8 x double> @test(<4 x double> %a, <4 x double> %b) {
 ; CHECK-LABEL: test:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    vmovaps {{.*#+}} ymm1 = <u,0.82071743224100002,0.82071743224100002,0.82071743224100002>
 ; CHECK-NEXT:    vblendps {{.*#+}} ymm2 = ymm0[0,1,2,3],ymm1[4,5,6,7]
 ; CHECK-NEXT:    vblendps {{.*#+}} ymm1 = ymm0[0,1],ymm1[2,3],ymm0[4,5,6,7]
 ; CHECK-NEXT:    vunpckhpd {{.*#+}} ymm1 = ymm1[1],ymm2[1],ymm1[3],ymm2[3]
-; CHECK-NEXT:    vmovaps {{.*#+}} xmm2 = [8.207174e-01,8.207174e-01]
+; CHECK-NEXT:    vmovaps {{.*#+}} xmm2 = [0.82071743224100002,0.82071743224100002]
 ; CHECK-NEXT:    vblendps {{.*#+}} ymm0 = ymm2[0,1,2,3],ymm0[4,5,6,7]
 ; CHECK-NEXT:    retq
   %1 = shufflevector <4 x double> %a, <4 x double> <double undef, double 0x3FEA435134576E1C, double 0x3FEA435134576E1C, double 0x3FEA435134576E1C>, <8 x i32> <i32 6, i32 5, i32 2, i32 3, i32 5, i32 1, i32 3, i32 7>
   ret <8 x double> %1
 }
 

llvm/trunk/test/CodeGen/X86/recip-fastmath.ll

 ; SKX-NEXT:    retq # sched: [7:1.00]
   %div = fdiv fast float 1.0, %x
   ret float %div
 }
 
 define <4 x float> @v4f32_no_estimate(<4 x float> %x) #0 {
 ; SSE-LABEL: v4f32_no_estimate:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1,1,1,1]
 ; SSE-NEXT:    divps %xmm0, %xmm1
 ; SSE-NEXT:    movaps %xmm1, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v4f32_no_estimate:
 ; AVX-RECIP:       # %bb.0:
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} xmm1 = [1,1,1,1]
 ; AVX-RECIP-NEXT:    vdivps %xmm0, %xmm1, %xmm0
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v4f32_no_estimate:
 ; FMA-RECIP:       # %bb.0:
-; FMA-RECIP-NEXT:    vmovaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT:    vmovaps {{.*#+}} xmm1 = [1,1,1,1]
 ; FMA-RECIP-NEXT:    vdivps %xmm0, %xmm1, %xmm0
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v4f32_no_estimate:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} xmm1 = [1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vdivps %xmm0, %xmm1, %xmm0 # sched: [19:19.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v4f32_no_estimate:
 ; SANDY:       # %bb.0:
-; SANDY-NEXT:    vmovaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [6:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} xmm1 = [1,1,1,1] sched: [6:0.50]
 ; SANDY-NEXT:    vdivps %xmm0, %xmm1, %xmm0 # sched: [14:14.00]
 ; SANDY-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: v4f32_no_estimate:
 ; HASWELL:       # %bb.0:
 ; HASWELL-NEXT:    vbroadcastss {{.*#+}} xmm1 = [1,1,1,1] sched: [6:0.50]
 ; HASWELL-NEXT:    vdivps %xmm0, %xmm1, %xmm0 # sched: [13:7.00]
 ; HASWELL-NEXT:    retq # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
   ret <4 x float> %div
 }
 
 define <4 x float> @v4f32_one_step(<4 x float> %x) #1 {
 ; SSE-LABEL: v4f32_one_step:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    rcpps %xmm0, %xmm2
 ; SSE-NEXT:    mulps %xmm2, %xmm0
-; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1,1,1,1]
 ; SSE-NEXT:    subps %xmm0, %xmm1
 ; SSE-NEXT:    mulps %xmm2, %xmm1
 ; SSE-NEXT:    addps %xmm2, %xmm1
 ; SSE-NEXT:    movaps %xmm1, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v4f32_one_step:
 ; AVX-RECIP:       # %bb.0:
 ; AVX-RECIP-NEXT:    vrcpps %xmm0, %xmm1
 ; AVX-RECIP-NEXT:    vmulps %xmm1, %xmm0, %xmm0
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} xmm2 = [1,1,1,1]
 ; AVX-RECIP-NEXT:    vsubps %xmm0, %xmm2, %xmm0
 ; AVX-RECIP-NEXT:    vmulps %xmm0, %xmm1, %xmm0
 ; AVX-RECIP-NEXT:    vaddps %xmm0, %xmm1, %xmm0
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v4f32_one_step:
 ; FMA-RECIP:       # %bb.0:
 ; FMA-RECIP-NEXT:    vrcpps %xmm0, %xmm1
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + mem
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v4f32_one_step:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vrcpps %xmm0, %xmm1 # sched: [2:1.00]
 ; BTVER2-NEXT:    vmulps %xmm1, %xmm0, %xmm0 # sched: [2:1.00]
 ; BTVER2-NEXT:    vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
 ; BTVER2-NEXT:    vmulps %xmm0, %xmm1, %xmm0 # sched: [2:1.00]
 ; BTVER2-NEXT:    vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v4f32_one_step:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    vrcpps %xmm0, %xmm1 # sched: [5:1.00]
 ; SANDY-NEXT:    vmulps %xmm1, %xmm0, %xmm0 # sched: [5:1.00]
-; SANDY-NEXT:    vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [6:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
 ; SANDY-NEXT:    vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %xmm0, %xmm1, %xmm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
 ; SANDY-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: v4f32_one_step:
 ; HASWELL:       # %bb.0:
 ; HASWELL-NEXT:    vrcpps %xmm0, %xmm1 # sched: [5:1.00]

 }
 
 define <4 x float> @v4f32_two_step(<4 x float> %x) #2 {
 ; SSE-LABEL: v4f32_two_step:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    rcpps %xmm0, %xmm2
 ; SSE-NEXT:    movaps %xmm0, %xmm3
 ; SSE-NEXT:    mulps %xmm2, %xmm3
-; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm1, %xmm4
 ; SSE-NEXT:    subps %xmm3, %xmm4
 ; SSE-NEXT:    mulps %xmm2, %xmm4
 ; SSE-NEXT:    addps %xmm2, %xmm4
 ; SSE-NEXT:    mulps %xmm4, %xmm0
 ; SSE-NEXT:    subps %xmm0, %xmm1
 ; SSE-NEXT:    mulps %xmm4, %xmm1
 ; SSE-NEXT:    addps %xmm4, %xmm1
 ; SSE-NEXT:    movaps %xmm1, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v4f32_two_step:
 ; AVX-RECIP:       # %bb.0:
 ; AVX-RECIP-NEXT:    vrcpps %xmm0, %xmm1
 ; AVX-RECIP-NEXT:    vmulps %xmm1, %xmm0, %xmm2
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} xmm3 = [1,1,1,1]
 ; AVX-RECIP-NEXT:    vsubps %xmm2, %xmm3, %xmm2
 ; AVX-RECIP-NEXT:    vmulps %xmm2, %xmm1, %xmm2
 ; AVX-RECIP-NEXT:    vaddps %xmm2, %xmm1, %xmm1
 ; AVX-RECIP-NEXT:    vmulps %xmm1, %xmm0, %xmm0
 ; AVX-RECIP-NEXT:    vsubps %xmm0, %xmm3, %xmm0
 ; AVX-RECIP-NEXT:    vmulps %xmm0, %xmm1, %xmm0
 ; AVX-RECIP-NEXT:    vaddps %xmm0, %xmm1, %xmm0
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v4f32_two_step:
 ; FMA-RECIP:       # %bb.0:
 ; FMA-RECIP-NEXT:    vrcpps %xmm0, %xmm1
-; FMA-RECIP-NEXT:    vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT:    vmovaps {{.*#+}} xmm2 = [1,1,1,1]
 ; FMA-RECIP-NEXT:    vmovaps %xmm1, %xmm3
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} xmm3 = -(xmm0 * xmm3) + xmm2
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} xmm3 = (xmm3 * xmm1) + xmm1
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm2
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm3) + xmm3
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v4f32_two_step:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} xmm3 = [1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vrcpps %xmm0, %xmm1 # sched: [2:1.00]
 ; BTVER2-NEXT:    vmulps %xmm1, %xmm0, %xmm2 # sched: [2:1.00]
 ; BTVER2-NEXT:    vsubps %xmm2, %xmm3, %xmm2 # sched: [3:1.00]
 ; BTVER2-NEXT:    vmulps %xmm2, %xmm1, %xmm2 # sched: [2:1.00]
 ; BTVER2-NEXT:    vaddps %xmm2, %xmm1, %xmm1 # sched: [3:1.00]
 ; BTVER2-NEXT:    vmulps %xmm1, %xmm0, %xmm0 # sched: [2:1.00]
 ; BTVER2-NEXT:    vsubps %xmm0, %xmm3, %xmm0 # sched: [3:1.00]
 ; BTVER2-NEXT:    vmulps %xmm0, %xmm1, %xmm0 # sched: [2:1.00]
 ; BTVER2-NEXT:    vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v4f32_two_step:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    vrcpps %xmm0, %xmm1 # sched: [5:1.00]
 ; SANDY-NEXT:    vmulps %xmm1, %xmm0, %xmm2 # sched: [5:1.00]
-; SANDY-NEXT:    vmovaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [6:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} xmm3 = [1,1,1,1] sched: [6:0.50]
 ; SANDY-NEXT:    vsubps %xmm2, %xmm3, %xmm2 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %xmm2, %xmm1, %xmm2 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %xmm2, %xmm1, %xmm1 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %xmm1, %xmm0, %xmm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vsubps %xmm0, %xmm3, %xmm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %xmm0, %xmm1, %xmm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
 ; SANDY-NEXT:    retq # sched: [1:1.00]

 ; SKX-NEXT:    retq # sched: [7:1.00]
   %div = fdiv fast <4 x float> <float 1.0, float 1.0, float 1.0, float 1.0>, %x
   ret <4 x float> %div
 }
 
 define <8 x float> @v8f32_no_estimate(<8 x float> %x) #0 {
 ; SSE-LABEL: v8f32_no_estimate:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm2 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm2, %xmm3
 ; SSE-NEXT:    divps %xmm0, %xmm3
 ; SSE-NEXT:    divps %xmm1, %xmm2
 ; SSE-NEXT:    movaps %xmm3, %xmm0
 ; SSE-NEXT:    movaps %xmm2, %xmm1
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v8f32_no_estimate:
 ; AVX-RECIP:       # %bb.0:
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
 ; AVX-RECIP-NEXT:    vdivps %ymm0, %ymm1, %ymm0
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v8f32_no_estimate:
 ; FMA-RECIP:       # %bb.0:
-; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
 ; FMA-RECIP-NEXT:    vdivps %ymm0, %ymm1, %ymm0
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v8f32_no_estimate:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} ymm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vdivps %ymm0, %ymm1, %ymm0 # sched: [38:38.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v8f32_no_estimate:
 ; SANDY:       # %bb.0:
-; SANDY-NEXT:    vmovaps {{.*#+}} ymm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
 ; SANDY-NEXT:    vdivps %ymm0, %ymm1, %ymm0 # sched: [29:28.00]
 ; SANDY-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: v8f32_no_estimate:
 ; HASWELL:       # %bb.0:
 ; HASWELL-NEXT:    vbroadcastss {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
 ; HASWELL-NEXT:    vdivps %ymm0, %ymm1, %ymm0 # sched: [21:14.00]
 ; HASWELL-NEXT:    retq # sched: [7:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
   ret <8 x float> %div
 }
 
 define <8 x float> @v8f32_one_step(<8 x float> %x) #1 {
 ; SSE-LABEL: v8f32_one_step:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    rcpps %xmm0, %xmm4
 ; SSE-NEXT:    mulps %xmm4, %xmm0
-; SSE-NEXT:    movaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm2 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm2, %xmm3
 ; SSE-NEXT:    subps %xmm0, %xmm3
 ; SSE-NEXT:    mulps %xmm4, %xmm3
 ; SSE-NEXT:    addps %xmm4, %xmm3
 ; SSE-NEXT:    rcpps %xmm1, %xmm0
 ; SSE-NEXT:    mulps %xmm0, %xmm1
 ; SSE-NEXT:    subps %xmm1, %xmm2
 ; SSE-NEXT:    mulps %xmm0, %xmm2
 ; SSE-NEXT:    addps %xmm0, %xmm2
 ; SSE-NEXT:    movaps %xmm3, %xmm0
 ; SSE-NEXT:    movaps %xmm2, %xmm1
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v8f32_one_step:
 ; AVX-RECIP:       # %bb.0:
 ; AVX-RECIP-NEXT:    vrcpps %ymm0, %ymm1
 ; AVX-RECIP-NEXT:    vmulps %ymm1, %ymm0, %ymm0
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
 ; AVX-RECIP-NEXT:    vsubps %ymm0, %ymm2, %ymm0
 ; AVX-RECIP-NEXT:    vmulps %ymm0, %ymm1, %ymm0
 ; AVX-RECIP-NEXT:    vaddps %ymm0, %ymm1, %ymm0
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v8f32_one_step:
 ; FMA-RECIP:       # %bb.0:
 ; FMA-RECIP-NEXT:    vrcpps %ymm0, %ymm1
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + mem
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v8f32_one_step:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vrcpps %ymm0, %ymm1 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm1, %ymm0, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm0, %ymm2, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm0, %ymm1, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v8f32_one_step:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    vrcpps %ymm0, %ymm1 # sched: [7:2.00]
 ; SANDY-NEXT:    vmulps %ymm1, %ymm0, %ymm0 # sched: [5:1.00]
-; SANDY-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
 ; SANDY-NEXT:    vsubps %ymm0, %ymm2, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm0, %ymm1, %ymm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: v8f32_one_step:
 ; HASWELL:       # %bb.0:
 ; HASWELL-NEXT:    vrcpps %ymm0, %ymm1 # sched: [11:2.00]

 
 define <8 x float> @v8f32_two_step(<8 x float> %x) #2 {
 ; SSE-LABEL: v8f32_two_step:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movaps %xmm1, %xmm2
 ; SSE-NEXT:    rcpps %xmm0, %xmm3
 ; SSE-NEXT:    movaps %xmm0, %xmm4
 ; SSE-NEXT:    mulps %xmm3, %xmm4
-; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm1, %xmm5
 ; SSE-NEXT:    subps %xmm4, %xmm5
 ; SSE-NEXT:    mulps %xmm3, %xmm5
 ; SSE-NEXT:    addps %xmm3, %xmm5
 ; SSE-NEXT:    mulps %xmm5, %xmm0
 ; SSE-NEXT:    movaps %xmm1, %xmm3
 ; SSE-NEXT:    subps %xmm0, %xmm3
 ; SSE-NEXT:    mulps %xmm5, %xmm3

 ; SSE-NEXT:    addps %xmm5, %xmm1
 ; SSE-NEXT:    movaps %xmm3, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v8f32_two_step:
 ; AVX-RECIP:       # %bb.0:
 ; AVX-RECIP-NEXT:    vrcpps %ymm0, %ymm1
 ; AVX-RECIP-NEXT:    vmulps %ymm1, %ymm0, %ymm2
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
 ; AVX-RECIP-NEXT:    vsubps %ymm2, %ymm3, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm1, %ymm2
 ; AVX-RECIP-NEXT:    vaddps %ymm2, %ymm1, %ymm1
 ; AVX-RECIP-NEXT:    vmulps %ymm1, %ymm0, %ymm0
 ; AVX-RECIP-NEXT:    vsubps %ymm0, %ymm3, %ymm0
 ; AVX-RECIP-NEXT:    vmulps %ymm0, %ymm1, %ymm0
 ; AVX-RECIP-NEXT:    vaddps %ymm0, %ymm1, %ymm0
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v8f32_two_step:
 ; FMA-RECIP:       # %bb.0:
 ; FMA-RECIP-NEXT:    vrcpps %ymm0, %ymm1
-; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
 ; FMA-RECIP-NEXT:    vmovaps %ymm1, %ymm3
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm3 = -(ymm0 * ymm3) + ymm2
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm3 = (ymm3 * ymm1) + ymm1
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm0 = -(ymm3 * ymm0) + ymm2
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm3) + ymm3
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v8f32_two_step:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vrcpps %ymm0, %ymm1 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm1, %ymm0, %ymm2 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm2, %ymm3, %ymm2 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm2, %ymm1, %ymm2 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm2, %ymm1, %ymm1 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm1, %ymm0, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm0, %ymm3, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm0, %ymm1, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v8f32_two_step:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    vrcpps %ymm0, %ymm1 # sched: [7:2.00]
 ; SANDY-NEXT:    vmulps %ymm1, %ymm0, %ymm2 # sched: [5:1.00]
-; SANDY-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
 ; SANDY-NEXT:    vsubps %ymm2, %ymm3, %ymm2 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm2, %ymm1, %ymm2 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm2, %ymm1, %ymm1 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm1, %ymm0, %ymm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vsubps %ymm0, %ymm3, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm0, %ymm1, %ymm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    retq # sched: [1:1.00]

 ; SKX-NEXT:    retq # sched: [7:1.00]
   %div = fdiv fast <8 x float> <float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>, %x
   ret <8 x float> %div
 }
 
 define <16 x float> @v16f32_no_estimate(<16 x float> %x) #0 {
 ; SSE-LABEL: v16f32_no_estimate:
 ; SSE:       # %bb.0:
-; SSE-NEXT:    movaps {{.*#+}} xmm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm4 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm4, %xmm5
 ; SSE-NEXT:    divps %xmm0, %xmm5
 ; SSE-NEXT:    movaps %xmm4, %xmm6
 ; SSE-NEXT:    divps %xmm1, %xmm6
 ; SSE-NEXT:    movaps %xmm4, %xmm7
 ; SSE-NEXT:    divps %xmm2, %xmm7
 ; SSE-NEXT:    divps %xmm3, %xmm4
 ; SSE-NEXT:    movaps %xmm5, %xmm0
 ; SSE-NEXT:    movaps %xmm6, %xmm1
 ; SSE-NEXT:    movaps %xmm7, %xmm2
 ; SSE-NEXT:    movaps %xmm4, %xmm3
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v16f32_no_estimate:
 ; AVX-RECIP:       # %bb.0:
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
 ; AVX-RECIP-NEXT:    vdivps %ymm0, %ymm2, %ymm0
 ; AVX-RECIP-NEXT:    vdivps %ymm1, %ymm2, %ymm1
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v16f32_no_estimate:
 ; FMA-RECIP:       # %bb.0:
-; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
 ; FMA-RECIP-NEXT:    vdivps %ymm0, %ymm2, %ymm0
 ; FMA-RECIP-NEXT:    vdivps %ymm1, %ymm2, %ymm1
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v16f32_no_estimate:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vdivps %ymm0, %ymm2, %ymm0 # sched: [38:38.00]
 ; BTVER2-NEXT:    vdivps %ymm1, %ymm2, %ymm1 # sched: [38:38.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v16f32_no_estimate:
 ; SANDY:       # %bb.0:
-; SANDY-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
 ; SANDY-NEXT:    vdivps %ymm0, %ymm2, %ymm0 # sched: [29:28.00]
 ; SANDY-NEXT:    vdivps %ymm1, %ymm2, %ymm1 # sched: [29:28.00]
 ; SANDY-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: v16f32_no_estimate:
 ; HASWELL:       # %bb.0:
 ; HASWELL-NEXT:    vbroadcastss {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
 ; HASWELL-NEXT:    vdivps %ymm0, %ymm2, %ymm0 # sched: [21:14.00]

 
 define <16 x float> @v16f32_one_step(<16 x float> %x) #1 {
 ; SSE-LABEL: v16f32_one_step:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movaps %xmm3, %xmm4
 ; SSE-NEXT:    movaps %xmm0, %xmm5
 ; SSE-NEXT:    rcpps %xmm0, %xmm6
 ; SSE-NEXT:    mulps %xmm6, %xmm5
-; SSE-NEXT:    movaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm3 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm3, %xmm0
 ; SSE-NEXT:    subps %xmm5, %xmm0
 ; SSE-NEXT:    mulps %xmm6, %xmm0
 ; SSE-NEXT:    addps %xmm6, %xmm0
 ; SSE-NEXT:    rcpps %xmm1, %xmm6
 ; SSE-NEXT:    mulps %xmm6, %xmm1
 ; SSE-NEXT:    movaps %xmm3, %xmm5
 ; SSE-NEXT:    subps %xmm1, %xmm5

 ; SSE-NEXT:    movaps %xmm5, %xmm1
 ; SSE-NEXT:    movaps %xmm6, %xmm2
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v16f32_one_step:
 ; AVX-RECIP:       # %bb.0:
 ; AVX-RECIP-NEXT:    vrcpps %ymm0, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm0, %ymm0
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
 ; AVX-RECIP-NEXT:    vsubps %ymm0, %ymm3, %ymm0
 ; AVX-RECIP-NEXT:    vmulps %ymm0, %ymm2, %ymm0
 ; AVX-RECIP-NEXT:    vaddps %ymm0, %ymm2, %ymm0
 ; AVX-RECIP-NEXT:    vrcpps %ymm1, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm1, %ymm1
 ; AVX-RECIP-NEXT:    vsubps %ymm1, %ymm3, %ymm1
 ; AVX-RECIP-NEXT:    vmulps %ymm1, %ymm2, %ymm1
 ; AVX-RECIP-NEXT:    vaddps %ymm1, %ymm2, %ymm1
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v16f32_one_step:
 ; FMA-RECIP:       # %bb.0:
 ; FMA-RECIP-NEXT:    vrcpps %ymm0, %ymm2
-; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm0 = -(ymm2 * ymm0) + ymm3
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm2) + ymm2
 ; FMA-RECIP-NEXT:    vrcpps %ymm1, %ymm2
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm1 = -(ymm2 * ymm1) + ymm3
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm1 = (ymm1 * ymm2) + ymm2
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v16f32_one_step:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vrcpps %ymm0, %ymm2 # sched: [2:2.00]
 ; BTVER2-NEXT:    vrcpps %ymm1, %ymm4 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm2, %ymm0, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm4, %ymm1, %ymm1 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm0, %ymm3, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vsubps %ymm1, %ymm3, %ymm1 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm0, %ymm2, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm1, %ymm4, %ymm1 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm0, %ymm2, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vaddps %ymm1, %ymm4, %ymm1 # sched: [3:2.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v16f32_one_step:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    vrcpps %ymm0, %ymm2 # sched: [7:2.00]
 ; SANDY-NEXT:    vmulps %ymm2, %ymm0, %ymm0 # sched: [5:1.00]
-; SANDY-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
 ; SANDY-NEXT:    vsubps %ymm0, %ymm3, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm0, %ymm2, %ymm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm0, %ymm2, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vrcpps %ymm1, %ymm2 # sched: [7:2.00]
 ; SANDY-NEXT:    vmulps %ymm2, %ymm1, %ymm1 # sched: [5:1.00]
 ; SANDY-NEXT:    vsubps %ymm1, %ymm3, %ymm1 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm1, %ymm2, %ymm1 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm1, %ymm2, %ymm1 # sched: [3:1.00]

 ; SSE-LABEL: v16f32_two_step:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movaps %xmm3, %xmm4
 ; SSE-NEXT:    movaps %xmm1, %xmm5
 ; SSE-NEXT:    movaps %xmm0, %xmm1
 ; SSE-NEXT:    rcpps %xmm0, %xmm0
 ; SSE-NEXT:    movaps %xmm1, %xmm6
 ; SSE-NEXT:    mulps %xmm0, %xmm6
-; SSE-NEXT:    movaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm3 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm3, %xmm7
 ; SSE-NEXT:    subps %xmm6, %xmm7
 ; SSE-NEXT:    mulps %xmm0, %xmm7
 ; SSE-NEXT:    addps %xmm0, %xmm7
 ; SSE-NEXT:    mulps %xmm7, %xmm1
 ; SSE-NEXT:    movaps %xmm3, %xmm0
 ; SSE-NEXT:    subps %xmm1, %xmm0
 ; SSE-NEXT:    mulps %xmm7, %xmm0

 ; SSE-NEXT:    addps %xmm7, %xmm3
 ; SSE-NEXT:    movaps %xmm5, %xmm2
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v16f32_two_step:
 ; AVX-RECIP:       # %bb.0:
 ; AVX-RECIP-NEXT:    vrcpps %ymm0, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm0, %ymm3
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
 ; AVX-RECIP-NEXT:    vsubps %ymm3, %ymm4, %ymm3
 ; AVX-RECIP-NEXT:    vmulps %ymm3, %ymm2, %ymm3
 ; AVX-RECIP-NEXT:    vaddps %ymm3, %ymm2, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm0, %ymm0
 ; AVX-RECIP-NEXT:    vsubps %ymm0, %ymm4, %ymm0
 ; AVX-RECIP-NEXT:    vmulps %ymm0, %ymm2, %ymm0
 ; AVX-RECIP-NEXT:    vaddps %ymm0, %ymm2, %ymm0
 ; AVX-RECIP-NEXT:    vrcpps %ymm1, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm1, %ymm3
 ; AVX-RECIP-NEXT:    vsubps %ymm3, %ymm4, %ymm3
 ; AVX-RECIP-NEXT:    vmulps %ymm3, %ymm2, %ymm3
 ; AVX-RECIP-NEXT:    vaddps %ymm3, %ymm2, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm1, %ymm1
 ; AVX-RECIP-NEXT:    vsubps %ymm1, %ymm4, %ymm1
 ; AVX-RECIP-NEXT:    vmulps %ymm1, %ymm2, %ymm1
 ; AVX-RECIP-NEXT:    vaddps %ymm1, %ymm2, %ymm1
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v16f32_two_step:
 ; FMA-RECIP:       # %bb.0:
 ; FMA-RECIP-NEXT:    vrcpps %ymm0, %ymm2
-; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
 ; FMA-RECIP-NEXT:    vmovaps %ymm2, %ymm4
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm4 = -(ymm0 * ymm4) + ymm3
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm4 = (ymm4 * ymm2) + ymm2
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm0 = -(ymm4 * ymm0) + ymm3
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm4) + ymm4
 ; FMA-RECIP-NEXT:    vrcpps %ymm1, %ymm2
 ; FMA-RECIP-NEXT:    vmovaps %ymm2, %ymm4
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm4 = -(ymm1 * ymm4) + ymm3
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm4 = (ymm4 * ymm2) + ymm2
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm1 = -(ymm4 * ymm1) + ymm3
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm1 = (ymm1 * ymm4) + ymm4
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v16f32_two_step:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vrcpps %ymm0, %ymm2 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm2, %ymm0, %ymm3 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm3, %ymm4, %ymm3 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm3, %ymm2, %ymm3 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm3, %ymm2, %ymm2 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm2, %ymm0, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm0, %ymm4, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm0, %ymm2, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm0, %ymm2, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vrcpps %ymm1, %ymm2 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm2, %ymm1, %ymm3 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm3, %ymm4, %ymm3 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm3, %ymm2, %ymm3 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm3, %ymm2, %ymm2 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm2, %ymm1, %ymm1 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm1, %ymm4, %ymm1 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm1, %ymm2, %ymm1 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm1, %ymm2, %ymm1 # sched: [3:2.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v16f32_two_step:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    vrcpps %ymm0, %ymm2 # sched: [7:2.00]
 ; SANDY-NEXT:    vmulps %ymm2, %ymm0, %ymm3 # sched: [5:1.00]
-; SANDY-NEXT:    vmovaps {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
 ; SANDY-NEXT:    vsubps %ymm3, %ymm4, %ymm3 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm3, %ymm2, %ymm3 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm3, %ymm2, %ymm2 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm2, %ymm0, %ymm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vsubps %ymm0, %ymm4, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm0, %ymm2, %ymm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm0, %ymm2, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vrcpps %ymm1, %ymm2 # sched: [7:2.00]

llvm/trunk/test/CodeGen/X86/recip-fastmath2.ll

; SKX-NEXT:    retq # sched: [7:1.00]
   ret float %div
 }
 
 define <4 x float> @v4f32_one_step2(<4 x float> %x) #1 {
 ; SSE-LABEL: v4f32_one_step2:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    rcpps %xmm0, %xmm2
 ; SSE-NEXT:    mulps %xmm2, %xmm0
-; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1,1,1,1]
 ; SSE-NEXT:    subps %xmm0, %xmm1
 ; SSE-NEXT:    mulps %xmm2, %xmm1
 ; SSE-NEXT:    addps %xmm2, %xmm1
 ; SSE-NEXT:    mulps {{.*}}(%rip), %xmm1
 ; SSE-NEXT:    movaps %xmm1, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v4f32_one_step2:
 ; AVX-RECIP:       # %bb.0:
 ; AVX-RECIP-NEXT:    vrcpps %xmm0, %xmm1
 ; AVX-RECIP-NEXT:    vmulps %xmm1, %xmm0, %xmm0
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} xmm2 = [1,1,1,1]
 ; AVX-RECIP-NEXT:    vsubps %xmm0, %xmm2, %xmm0
 ; AVX-RECIP-NEXT:    vmulps %xmm0, %xmm1, %xmm0
 ; AVX-RECIP-NEXT:    vaddps %xmm0, %xmm1, %xmm0
 ; AVX-RECIP-NEXT:    vmulps {{.*}}(%rip), %xmm0, %xmm0
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v4f32_one_step2:
 ; FMA-RECIP:       # %bb.0:
 ; FMA-RECIP-NEXT:    vrcpps %xmm0, %xmm1
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + mem
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1
 ; FMA-RECIP-NEXT:    vmulps {{.*}}(%rip), %xmm0, %xmm0
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v4f32_one_step2:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vrcpps %xmm0, %xmm1 # sched: [2:1.00]
 ; BTVER2-NEXT:    vmulps %xmm1, %xmm0, %xmm0 # sched: [2:1.00]
 ; BTVER2-NEXT:    vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
 ; BTVER2-NEXT:    vmulps %xmm0, %xmm1, %xmm0 # sched: [2:1.00]
 ; BTVER2-NEXT:    vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
 ; BTVER2-NEXT:    vmulps {{.*}}(%rip), %xmm0, %xmm0 # sched: [7:1.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v4f32_one_step2:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    vrcpps %xmm0, %xmm1 # sched: [5:1.00]
 ; SANDY-NEXT:    vmulps %xmm1, %xmm0, %xmm0 # sched: [5:1.00]
-; SANDY-NEXT:    vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [6:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
 ; SANDY-NEXT:    vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %xmm0, %xmm1, %xmm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps {{.*}}(%rip), %xmm0, %xmm0 # sched: [11:1.00]
 ; SANDY-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: v4f32_one_step2:
 ; HASWELL:       # %bb.0:
; SKX-NEXT:    retq # sched: [7:1.00]
   ret <4 x float> %div
 }
 
 define <4 x float> @v4f32_one_step_2_divs(<4 x float> %x) #1 {
 ; SSE-LABEL: v4f32_one_step_2_divs:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    rcpps %xmm0, %xmm1
 ; SSE-NEXT:    mulps %xmm1, %xmm0
-; SSE-NEXT:    movaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm2 = [1,1,1,1]
 ; SSE-NEXT:    subps %xmm0, %xmm2
 ; SSE-NEXT:    mulps %xmm1, %xmm2
 ; SSE-NEXT:    addps %xmm1, %xmm2
-; SSE-NEXT:    movaps {{.*#+}} xmm0 = [1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [1,2,3,4]
 ; SSE-NEXT:    mulps %xmm2, %xmm0
 ; SSE-NEXT:    mulps %xmm2, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v4f32_one_step_2_divs:
 ; AVX-RECIP:       # %bb.0:
 ; AVX-RECIP-NEXT:    vrcpps %xmm0, %xmm1
 ; AVX-RECIP-NEXT:    vmulps %xmm1, %xmm0, %xmm0
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} xmm2 = [1,1,1,1]
 ; AVX-RECIP-NEXT:    vsubps %xmm0, %xmm2, %xmm0
 ; AVX-RECIP-NEXT:    vmulps %xmm0, %xmm1, %xmm0
 ; AVX-RECIP-NEXT:    vaddps %xmm0, %xmm1, %xmm0
 ; AVX-RECIP-NEXT:    vmulps {{.*}}(%rip), %xmm0, %xmm1
 ; AVX-RECIP-NEXT:    vmulps %xmm0, %xmm1, %xmm0
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v4f32_one_step_2_divs:
 ; FMA-RECIP:       # %bb.0:
 ; FMA-RECIP-NEXT:    vrcpps %xmm0, %xmm1
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} xmm0 = -(xmm1 * xmm0) + mem
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm1) + xmm1
 ; FMA-RECIP-NEXT:    vmulps {{.*}}(%rip), %xmm0, %xmm1
 ; FMA-RECIP-NEXT:    vmulps %xmm0, %xmm1, %xmm0
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v4f32_one_step_2_divs:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vrcpps %xmm0, %xmm1 # sched: [2:1.00]
 ; BTVER2-NEXT:    vmulps %xmm1, %xmm0, %xmm0 # sched: [2:1.00]
 ; BTVER2-NEXT:    vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
 ; BTVER2-NEXT:    vmulps %xmm0, %xmm1, %xmm0 # sched: [2:1.00]
 ; BTVER2-NEXT:    vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
 ; BTVER2-NEXT:    vmulps {{.*}}(%rip), %xmm0, %xmm1 # sched: [7:1.00]
 ; BTVER2-NEXT:    vmulps %xmm0, %xmm1, %xmm0 # sched: [2:1.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v4f32_one_step_2_divs:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    vrcpps %xmm0, %xmm1 # sched: [5:1.00]
 ; SANDY-NEXT:    vmulps %xmm1, %xmm0, %xmm0 # sched: [5:1.00]
-; SANDY-NEXT:    vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [6:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} xmm2 = [1,1,1,1] sched: [6:0.50]
 ; SANDY-NEXT:    vsubps %xmm0, %xmm2, %xmm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %xmm0, %xmm1, %xmm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps {{.*}}(%rip), %xmm0, %xmm1 # sched: [11:1.00]
 ; SANDY-NEXT:    vmulps %xmm0, %xmm1, %xmm0 # sched: [5:1.00]
 ; SANDY-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: v4f32_one_step_2_divs:

 }
 
 define <4 x float> @v4f32_two_step2(<4 x float> %x) #2 {
 ; SSE-LABEL: v4f32_two_step2:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    rcpps %xmm0, %xmm2
 ; SSE-NEXT:    movaps %xmm0, %xmm3
 ; SSE-NEXT:    mulps %xmm2, %xmm3
-; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm1, %xmm4
 ; SSE-NEXT:    subps %xmm3, %xmm4
 ; SSE-NEXT:    mulps %xmm2, %xmm4
 ; SSE-NEXT:    addps %xmm2, %xmm4
 ; SSE-NEXT:    mulps %xmm4, %xmm0
 ; SSE-NEXT:    subps %xmm0, %xmm1
 ; SSE-NEXT:    mulps %xmm4, %xmm1
 ; SSE-NEXT:    addps %xmm4, %xmm1
 ; SSE-NEXT:    mulps {{.*}}(%rip), %xmm1
 ; SSE-NEXT:    movaps %xmm1, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v4f32_two_step2:
 ; AVX-RECIP:       # %bb.0:
 ; AVX-RECIP-NEXT:    vrcpps %xmm0, %xmm1
 ; AVX-RECIP-NEXT:    vmulps %xmm1, %xmm0, %xmm2
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} xmm3 = [1,1,1,1]
 ; AVX-RECIP-NEXT:    vsubps %xmm2, %xmm3, %xmm2
 ; AVX-RECIP-NEXT:    vmulps %xmm2, %xmm1, %xmm2
 ; AVX-RECIP-NEXT:    vaddps %xmm2, %xmm1, %xmm1
 ; AVX-RECIP-NEXT:    vmulps %xmm1, %xmm0, %xmm0
 ; AVX-RECIP-NEXT:    vsubps %xmm0, %xmm3, %xmm0
 ; AVX-RECIP-NEXT:    vmulps %xmm0, %xmm1, %xmm0
 ; AVX-RECIP-NEXT:    vaddps %xmm0, %xmm1, %xmm0
 ; AVX-RECIP-NEXT:    vmulps {{.*}}(%rip), %xmm0, %xmm0
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v4f32_two_step2:
 ; FMA-RECIP:       # %bb.0:
 ; FMA-RECIP-NEXT:    vrcpps %xmm0, %xmm1
-; FMA-RECIP-NEXT:    vmovaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT:    vmovaps {{.*#+}} xmm2 = [1,1,1,1]
 ; FMA-RECIP-NEXT:    vmovaps %xmm1, %xmm3
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} xmm3 = -(xmm0 * xmm3) + xmm2
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} xmm3 = (xmm3 * xmm1) + xmm1
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} xmm0 = -(xmm3 * xmm0) + xmm2
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} xmm0 = (xmm0 * xmm3) + xmm3
 ; FMA-RECIP-NEXT:    vmulps {{.*}}(%rip), %xmm0, %xmm0
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v4f32_two_step2:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} xmm3 = [1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vrcpps %xmm0, %xmm1 # sched: [2:1.00]
 ; BTVER2-NEXT:    vmulps %xmm1, %xmm0, %xmm2 # sched: [2:1.00]
 ; BTVER2-NEXT:    vsubps %xmm2, %xmm3, %xmm2 # sched: [3:1.00]
 ; BTVER2-NEXT:    vmulps %xmm2, %xmm1, %xmm2 # sched: [2:1.00]
 ; BTVER2-NEXT:    vaddps %xmm2, %xmm1, %xmm1 # sched: [3:1.00]
 ; BTVER2-NEXT:    vmulps %xmm1, %xmm0, %xmm0 # sched: [2:1.00]
 ; BTVER2-NEXT:    vsubps %xmm0, %xmm3, %xmm0 # sched: [3:1.00]
 ; BTVER2-NEXT:    vmulps %xmm0, %xmm1, %xmm0 # sched: [2:1.00]
 ; BTVER2-NEXT:    vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
 ; BTVER2-NEXT:    vmulps {{.*}}(%rip), %xmm0, %xmm0 # sched: [7:1.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v4f32_two_step2:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    vrcpps %xmm0, %xmm1 # sched: [5:1.00]
 ; SANDY-NEXT:    vmulps %xmm1, %xmm0, %xmm2 # sched: [5:1.00]
-; SANDY-NEXT:    vmovaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [6:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} xmm3 = [1,1,1,1] sched: [6:0.50]
 ; SANDY-NEXT:    vsubps %xmm2, %xmm3, %xmm2 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %xmm2, %xmm1, %xmm2 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %xmm2, %xmm1, %xmm1 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %xmm1, %xmm0, %xmm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vsubps %xmm0, %xmm3, %xmm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %xmm0, %xmm1, %xmm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %xmm0, %xmm1, %xmm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps {{.*}}(%rip), %xmm0, %xmm0 # sched: [11:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
   ret <4 x float> %div
 }
 
 define <8 x float> @v8f32_one_step2(<8 x float> %x) #1 {
 ; SSE-LABEL: v8f32_one_step2:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    rcpps %xmm1, %xmm4
 ; SSE-NEXT:    mulps %xmm4, %xmm1
-; SSE-NEXT:    movaps {{.*#+}} xmm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm2 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm2, %xmm3
 ; SSE-NEXT:    subps %xmm1, %xmm3
 ; SSE-NEXT:    mulps %xmm4, %xmm3
 ; SSE-NEXT:    addps %xmm4, %xmm3
 ; SSE-NEXT:    rcpps %xmm0, %xmm1
 ; SSE-NEXT:    mulps %xmm1, %xmm0
 ; SSE-NEXT:    subps %xmm0, %xmm2
 ; SSE-NEXT:    mulps %xmm1, %xmm2
 ; SSE-NEXT:    addps %xmm1, %xmm2
 ; SSE-NEXT:    mulps {{.*}}(%rip), %xmm2
 ; SSE-NEXT:    mulps {{.*}}(%rip), %xmm3
 ; SSE-NEXT:    movaps %xmm2, %xmm0
 ; SSE-NEXT:    movaps %xmm3, %xmm1
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v8f32_one_step2:
 ; AVX-RECIP:       # %bb.0:
 ; AVX-RECIP-NEXT:    vrcpps %ymm0, %ymm1
 ; AVX-RECIP-NEXT:    vmulps %ymm1, %ymm0, %ymm0
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
 ; AVX-RECIP-NEXT:    vsubps %ymm0, %ymm2, %ymm0
 ; AVX-RECIP-NEXT:    vmulps %ymm0, %ymm1, %ymm0
 ; AVX-RECIP-NEXT:    vaddps %ymm0, %ymm1, %ymm0
 ; AVX-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm0
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v8f32_one_step2:
 ; FMA-RECIP:       # %bb.0:
 ; FMA-RECIP-NEXT:    vrcpps %ymm0, %ymm1
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + mem
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1
 ; FMA-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm0
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v8f32_one_step2:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vrcpps %ymm0, %ymm1 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm1, %ymm0, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm0, %ymm2, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm0, %ymm1, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm0 # sched: [7:2.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v8f32_one_step2:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    vrcpps %ymm0, %ymm1 # sched: [7:2.00]
 ; SANDY-NEXT:    vmulps %ymm1, %ymm0, %ymm0 # sched: [5:1.00]
-; SANDY-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
 ; SANDY-NEXT:    vsubps %ymm0, %ymm2, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm0, %ymm1, %ymm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm0 # sched: [12:1.00]
 ; SANDY-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: v8f32_one_step2:
 ; HASWELL:       # %bb.0:
; SKX-NEXT:    retq # sched: [7:1.00]
   ret <8 x float> %div
 }
 
 define <8 x float> @v8f32_one_step_2_divs(<8 x float> %x) #1 {
 ; SSE-LABEL: v8f32_one_step_2_divs:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    rcpps %xmm0, %xmm2
 ; SSE-NEXT:    mulps %xmm2, %xmm0
-; SSE-NEXT:    movaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm3 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm3, %xmm4
 ; SSE-NEXT:    subps %xmm0, %xmm4
 ; SSE-NEXT:    mulps %xmm2, %xmm4
 ; SSE-NEXT:    addps %xmm2, %xmm4
 ; SSE-NEXT:    rcpps %xmm1, %xmm0
 ; SSE-NEXT:    mulps %xmm0, %xmm1
 ; SSE-NEXT:    subps %xmm1, %xmm3
 ; SSE-NEXT:    mulps %xmm0, %xmm3
 ; SSE-NEXT:    addps %xmm0, %xmm3
-; SSE-NEXT:    movaps {{.*#+}} xmm1 = [5.000000e+00,6.000000e+00,7.000000e+00,8.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [5,6,7,8]
 ; SSE-NEXT:    mulps %xmm3, %xmm1
-; SSE-NEXT:    movaps {{.*#+}} xmm0 = [1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [1,2,3,4]
 ; SSE-NEXT:    mulps %xmm4, %xmm0
 ; SSE-NEXT:    mulps %xmm4, %xmm0
 ; SSE-NEXT:    mulps %xmm3, %xmm1
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v8f32_one_step_2_divs:
 ; AVX-RECIP:       # %bb.0:
 ; AVX-RECIP-NEXT:    vrcpps %ymm0, %ymm1
 ; AVX-RECIP-NEXT:    vmulps %ymm1, %ymm0, %ymm0
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
 ; AVX-RECIP-NEXT:    vsubps %ymm0, %ymm2, %ymm0
 ; AVX-RECIP-NEXT:    vmulps %ymm0, %ymm1, %ymm0
 ; AVX-RECIP-NEXT:    vaddps %ymm0, %ymm1, %ymm0
 ; AVX-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm1
 ; AVX-RECIP-NEXT:    vmulps %ymm0, %ymm1, %ymm0
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v8f32_one_step_2_divs:
 ; FMA-RECIP:       # %bb.0:
 ; FMA-RECIP-NEXT:    vrcpps %ymm0, %ymm1
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm0 = -(ymm1 * ymm0) + mem
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm1) + ymm1
 ; FMA-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm1
 ; FMA-RECIP-NEXT:    vmulps %ymm0, %ymm1, %ymm0
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v8f32_one_step_2_divs:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vrcpps %ymm0, %ymm1 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm1, %ymm0, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm0, %ymm2, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm0, %ymm1, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm1 # sched: [7:2.00]
 ; BTVER2-NEXT:    vmulps %ymm0, %ymm1, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v8f32_one_step_2_divs:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    vrcpps %ymm0, %ymm1 # sched: [7:2.00]
 ; SANDY-NEXT:    vmulps %ymm1, %ymm0, %ymm0 # sched: [5:1.00]
-; SANDY-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
 ; SANDY-NEXT:    vsubps %ymm0, %ymm2, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm0, %ymm1, %ymm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm1 # sched: [12:1.00]
 ; SANDY-NEXT:    vmulps %ymm0, %ymm1, %ymm0 # sched: [5:1.00]
 ; SANDY-NEXT:    retq # sched: [1:1.00]
 ;
 ; HASWELL-LABEL: v8f32_one_step_2_divs:

 
 define <8 x float> @v8f32_two_step2(<8 x float> %x) #2 {
 ; SSE-LABEL: v8f32_two_step2:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movaps %xmm0, %xmm2
 ; SSE-NEXT:    rcpps %xmm1, %xmm3
 ; SSE-NEXT:    movaps %xmm1, %xmm4
 ; SSE-NEXT:    mulps %xmm3, %xmm4
-; SSE-NEXT:    movaps {{.*#+}} xmm0 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm0, %xmm5
 ; SSE-NEXT:    subps %xmm4, %xmm5
 ; SSE-NEXT:    mulps %xmm3, %xmm5
 ; SSE-NEXT:    addps %xmm3, %xmm5
 ; SSE-NEXT:    mulps %xmm5, %xmm1
 ; SSE-NEXT:    movaps %xmm0, %xmm3
 ; SSE-NEXT:    subps %xmm1, %xmm3
 ; SSE-NEXT:    mulps %xmm5, %xmm3

 ; SSE-NEXT:    mulps {{.*}}(%rip), %xmm3
 ; SSE-NEXT:    movaps %xmm3, %xmm1
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v8f32_two_step2:
 ; AVX-RECIP:       # %bb.0:
 ; AVX-RECIP-NEXT:    vrcpps %ymm0, %ymm1
 ; AVX-RECIP-NEXT:    vmulps %ymm1, %ymm0, %ymm2
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
 ; AVX-RECIP-NEXT:    vsubps %ymm2, %ymm3, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm1, %ymm2
 ; AVX-RECIP-NEXT:    vaddps %ymm2, %ymm1, %ymm1
 ; AVX-RECIP-NEXT:    vmulps %ymm1, %ymm0, %ymm0
 ; AVX-RECIP-NEXT:    vsubps %ymm0, %ymm3, %ymm0
 ; AVX-RECIP-NEXT:    vmulps %ymm0, %ymm1, %ymm0
 ; AVX-RECIP-NEXT:    vaddps %ymm0, %ymm1, %ymm0
 ; AVX-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm0
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v8f32_two_step2:
 ; FMA-RECIP:       # %bb.0:
 ; FMA-RECIP-NEXT:    vrcpps %ymm0, %ymm1
-; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
 ; FMA-RECIP-NEXT:    vmovaps %ymm1, %ymm3
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm3 = -(ymm0 * ymm3) + ymm2
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm3 = (ymm3 * ymm1) + ymm1
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm0 = -(ymm3 * ymm0) + ymm2
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm3) + ymm3
 ; FMA-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm0
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v8f32_two_step2:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vrcpps %ymm0, %ymm1 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm1, %ymm0, %ymm2 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm2, %ymm3, %ymm2 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm2, %ymm1, %ymm2 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm2, %ymm1, %ymm1 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm1, %ymm0, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm0, %ymm3, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm0, %ymm1, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm0 # sched: [7:2.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v8f32_two_step2:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    vrcpps %ymm0, %ymm1 # sched: [7:2.00]
 ; SANDY-NEXT:    vmulps %ymm1, %ymm0, %ymm2 # sched: [5:1.00]
-; SANDY-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
 ; SANDY-NEXT:    vsubps %ymm2, %ymm3, %ymm2 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm2, %ymm1, %ymm2 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm2, %ymm1, %ymm1 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm1, %ymm0, %ymm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vsubps %ymm0, %ymm3, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm0, %ymm1, %ymm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm0 # sched: [12:1.00]

 define <16 x float> @v16f32_one_step2(<16 x float> %x) #1 {
 ; SSE-LABEL: v16f32_one_step2:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movaps %xmm3, %xmm4
 ; SSE-NEXT:    movaps %xmm2, %xmm5
 ; SSE-NEXT:    movaps %xmm0, %xmm6
 ; SSE-NEXT:    rcpps %xmm3, %xmm2
 ; SSE-NEXT:    mulps %xmm2, %xmm4
-; SSE-NEXT:    movaps {{.*#+}} xmm0 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm0, %xmm3
 ; SSE-NEXT:    subps %xmm4, %xmm3
 ; SSE-NEXT:    mulps %xmm2, %xmm3
 ; SSE-NEXT:    addps %xmm2, %xmm3
 ; SSE-NEXT:    rcpps %xmm5, %xmm4
 ; SSE-NEXT:    mulps %xmm4, %xmm5
 ; SSE-NEXT:    movaps %xmm0, %xmm2
 ; SSE-NEXT:    subps %xmm5, %xmm2

 ; SSE-NEXT:    mulps {{.*}}(%rip), %xmm3
 ; SSE-NEXT:    movaps %xmm4, %xmm1
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v16f32_one_step2:
 ; AVX-RECIP:       # %bb.0:
 ; AVX-RECIP-NEXT:    vrcpps %ymm1, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm1, %ymm1
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
 ; AVX-RECIP-NEXT:    vsubps %ymm1, %ymm3, %ymm1
 ; AVX-RECIP-NEXT:    vmulps %ymm1, %ymm2, %ymm1
 ; AVX-RECIP-NEXT:    vaddps %ymm1, %ymm2, %ymm1
 ; AVX-RECIP-NEXT:    vrcpps %ymm0, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm0, %ymm0
 ; AVX-RECIP-NEXT:    vsubps %ymm0, %ymm3, %ymm0
 ; AVX-RECIP-NEXT:    vmulps %ymm0, %ymm2, %ymm0
 ; AVX-RECIP-NEXT:    vaddps %ymm0, %ymm2, %ymm0
 ; AVX-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm0
 ; AVX-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm1, %ymm1
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v16f32_one_step2:
 ; FMA-RECIP:       # %bb.0:
 ; FMA-RECIP-NEXT:    vrcpps %ymm1, %ymm2
-; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm1 = -(ymm2 * ymm1) + ymm3
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm1 = (ymm1 * ymm2) + ymm2
 ; FMA-RECIP-NEXT:    vrcpps %ymm0, %ymm2
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm0 = -(ymm2 * ymm0) + ymm3
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm2) + ymm2
 ; FMA-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm0
 ; FMA-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm1, %ymm1
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v16f32_one_step2:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vrcpps %ymm1, %ymm2 # sched: [2:2.00]
 ; BTVER2-NEXT:    vrcpps %ymm0, %ymm4 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm2, %ymm1, %ymm1 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm4, %ymm0, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm1, %ymm3, %ymm1 # sched: [3:2.00]
 ; BTVER2-NEXT:    vsubps %ymm0, %ymm3, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm1, %ymm2, %ymm1 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm0, %ymm4, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm1, %ymm2, %ymm1 # sched: [3:2.00]
 ; BTVER2-NEXT:    vaddps %ymm0, %ymm4, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm0 # sched: [7:2.00]
 ; BTVER2-NEXT:    vmulps {{.*}}(%rip), %ymm1, %ymm1 # sched: [7:2.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v16f32_one_step2:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    vrcpps %ymm1, %ymm2 # sched: [7:2.00]
 ; SANDY-NEXT:    vmulps %ymm2, %ymm1, %ymm1 # sched: [5:1.00]
-; SANDY-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
 ; SANDY-NEXT:    vsubps %ymm1, %ymm3, %ymm1 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm1, %ymm2, %ymm1 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm1, %ymm2, %ymm1 # sched: [3:1.00]
 ; SANDY-NEXT:    vrcpps %ymm0, %ymm2 # sched: [7:2.00]
 ; SANDY-NEXT:    vmulps %ymm2, %ymm0, %ymm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vsubps %ymm0, %ymm3, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm0, %ymm2, %ymm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm0, %ymm2, %ymm0 # sched: [3:1.00]
; SKX-NEXT:    retq # sched: [7:1.00]
   ret <16 x float> %div
 }
 
 define <16 x float> @v16f32_one_step_2_divs(<16 x float> %x) #1 {
 ; SSE-LABEL: v16f32_one_step_2_divs:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    rcpps %xmm0, %xmm6
 ; SSE-NEXT:    mulps %xmm6, %xmm0
-; SSE-NEXT:    movaps {{.*#+}} xmm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm4 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm4, %xmm5
 ; SSE-NEXT:    subps %xmm0, %xmm5
 ; SSE-NEXT:    mulps %xmm6, %xmm5
 ; SSE-NEXT:    addps %xmm6, %xmm5
 ; SSE-NEXT:    rcpps %xmm1, %xmm0
 ; SSE-NEXT:    mulps %xmm0, %xmm1
 ; SSE-NEXT:    movaps %xmm4, %xmm6
 ; SSE-NEXT:    subps %xmm1, %xmm6
 ; SSE-NEXT:    mulps %xmm0, %xmm6
 ; SSE-NEXT:    addps %xmm0, %xmm6
 ; SSE-NEXT:    rcpps %xmm2, %xmm0
 ; SSE-NEXT:    mulps %xmm0, %xmm2
 ; SSE-NEXT:    movaps %xmm4, %xmm7
 ; SSE-NEXT:    subps %xmm2, %xmm7
 ; SSE-NEXT:    mulps %xmm0, %xmm7
 ; SSE-NEXT:    addps %xmm0, %xmm7
 ; SSE-NEXT:    rcpps %xmm3, %xmm0
 ; SSE-NEXT:    mulps %xmm0, %xmm3
 ; SSE-NEXT:    subps %xmm3, %xmm4
 ; SSE-NEXT:    mulps %xmm0, %xmm4
 ; SSE-NEXT:    addps %xmm0, %xmm4
-; SSE-NEXT:    movaps {{.*#+}} xmm3 = [1.300000e+01,1.400000e+01,1.500000e+01,1.600000e+01]
+; SSE-NEXT:    movaps {{.*#+}} xmm3 = [13,14,15,16]
 ; SSE-NEXT:    mulps %xmm4, %xmm3
-; SSE-NEXT:    movaps {{.*#+}} xmm2 = [9.000000e+00,1.000000e+01,1.100000e+01,1.200000e+01]
+; SSE-NEXT:    movaps {{.*#+}} xmm2 = [9,10,11,12]
 ; SSE-NEXT:    mulps %xmm7, %xmm2
-; SSE-NEXT:    movaps {{.*#+}} xmm1 = [5.000000e+00,6.000000e+00,7.000000e+00,8.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [5,6,7,8]
 ; SSE-NEXT:    mulps %xmm6, %xmm1
-; SSE-NEXT:    movaps {{.*#+}} xmm0 = [1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [1,2,3,4]
 ; SSE-NEXT:    mulps %xmm5, %xmm0
 ; SSE-NEXT:    mulps %xmm5, %xmm0
 ; SSE-NEXT:    mulps %xmm6, %xmm1
 ; SSE-NEXT:    mulps %xmm7, %xmm2
 ; SSE-NEXT:    mulps %xmm4, %xmm3
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v16f32_one_step_2_divs:
 ; AVX-RECIP:       # %bb.0:
 ; AVX-RECIP-NEXT:    vrcpps %ymm0, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm0, %ymm0
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
 ; AVX-RECIP-NEXT:    vsubps %ymm0, %ymm3, %ymm0
 ; AVX-RECIP-NEXT:    vmulps %ymm0, %ymm2, %ymm0
 ; AVX-RECIP-NEXT:    vaddps %ymm0, %ymm2, %ymm0
 ; AVX-RECIP-NEXT:    vrcpps %ymm1, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm1, %ymm1
 ; AVX-RECIP-NEXT:    vsubps %ymm1, %ymm3, %ymm1
 ; AVX-RECIP-NEXT:    vmulps %ymm1, %ymm2, %ymm1
 ; AVX-RECIP-NEXT:    vaddps %ymm1, %ymm2, %ymm1
 ; AVX-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm1, %ymm2
 ; AVX-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm3
 ; AVX-RECIP-NEXT:    vmulps %ymm0, %ymm3, %ymm0
 ; AVX-RECIP-NEXT:    vmulps %ymm1, %ymm2, %ymm1
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v16f32_one_step_2_divs:
 ; FMA-RECIP:       # %bb.0:
 ; FMA-RECIP-NEXT:    vrcpps %ymm0, %ymm2
-; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm0 = -(ymm2 * ymm0) + ymm3
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm2) + ymm2
 ; FMA-RECIP-NEXT:    vrcpps %ymm1, %ymm2
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm1 = -(ymm2 * ymm1) + ymm3
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm1 = (ymm1 * ymm2) + ymm2
 ; FMA-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm1, %ymm2
 ; FMA-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm3
 ; FMA-RECIP-NEXT:    vmulps %ymm0, %ymm3, %ymm0
 ; FMA-RECIP-NEXT:    vmulps %ymm1, %ymm2, %ymm1
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v16f32_one_step_2_divs:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vrcpps %ymm0, %ymm2 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm2, %ymm0, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm0, %ymm3, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm0, %ymm2, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm0, %ymm2, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vrcpps %ymm1, %ymm2 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm2, %ymm1, %ymm1 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm1, %ymm3, %ymm1 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm3 # sched: [7:2.00]
 ; BTVER2-NEXT:    vmulps %ymm1, %ymm2, %ymm1 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm1, %ymm2, %ymm1 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps {{.*}}(%rip), %ymm1, %ymm2 # sched: [7:2.00]
 ; BTVER2-NEXT:    vmulps %ymm0, %ymm3, %ymm0 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm1, %ymm2, %ymm1 # sched: [2:2.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v16f32_one_step_2_divs:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    vrcpps %ymm0, %ymm2 # sched: [7:2.00]
 ; SANDY-NEXT:    vmulps %ymm2, %ymm0, %ymm0 # sched: [5:1.00]
-; SANDY-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
 ; SANDY-NEXT:    vsubps %ymm0, %ymm3, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vrcpps %ymm1, %ymm4 # sched: [7:2.00]
 ; SANDY-NEXT:    vmulps %ymm0, %ymm2, %ymm0 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm0, %ymm2, %ymm0 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm4, %ymm1, %ymm1 # sched: [5:1.00]
 ; SANDY-NEXT:    vsubps %ymm1, %ymm3, %ymm1 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm1, %ymm4, %ymm1 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm1, %ymm4, %ymm1 # sched: [3:1.00]

 define <16 x float> @v16f32_two_step2(<16 x float> %x) #2 {
 ; SSE-LABEL: v16f32_two_step2:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movaps %xmm3, %xmm6
 ; SSE-NEXT:    movaps %xmm2, %xmm5
 ; SSE-NEXT:    movaps %xmm0, %xmm4
 ; SSE-NEXT:    rcpps %xmm3, %xmm2
 ; SSE-NEXT:    mulps %xmm2, %xmm3
-; SSE-NEXT:    movaps {{.*#+}} xmm0 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm0, %xmm7
 ; SSE-NEXT:    subps %xmm3, %xmm7
 ; SSE-NEXT:    mulps %xmm2, %xmm7
 ; SSE-NEXT:    addps %xmm2, %xmm7
 ; SSE-NEXT:    mulps %xmm7, %xmm6
 ; SSE-NEXT:    movaps %xmm0, %xmm3
 ; SSE-NEXT:    subps %xmm6, %xmm3
 ; SSE-NEXT:    mulps %xmm7, %xmm3

 ; SSE-NEXT:    mulps {{.*}}(%rip), %xmm3
 ; SSE-NEXT:    movaps %xmm5, %xmm1
 ; SSE-NEXT:    retq
 ;
 ; AVX-RECIP-LABEL: v16f32_two_step2:
 ; AVX-RECIP:       # %bb.0:
 ; AVX-RECIP-NEXT:    vrcpps %ymm1, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm1, %ymm3
-; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX-RECIP-NEXT:    vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1]
 ; AVX-RECIP-NEXT:    vsubps %ymm3, %ymm4, %ymm3
 ; AVX-RECIP-NEXT:    vmulps %ymm3, %ymm2, %ymm3
 ; AVX-RECIP-NEXT:    vaddps %ymm3, %ymm2, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm1, %ymm1
 ; AVX-RECIP-NEXT:    vsubps %ymm1, %ymm4, %ymm1
 ; AVX-RECIP-NEXT:    vmulps %ymm1, %ymm2, %ymm1
 ; AVX-RECIP-NEXT:    vaddps %ymm1, %ymm2, %ymm1
 ; AVX-RECIP-NEXT:    vrcpps %ymm0, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm0, %ymm3
 ; AVX-RECIP-NEXT:    vsubps %ymm3, %ymm4, %ymm3
 ; AVX-RECIP-NEXT:    vmulps %ymm3, %ymm2, %ymm3
 ; AVX-RECIP-NEXT:    vaddps %ymm3, %ymm2, %ymm2
 ; AVX-RECIP-NEXT:    vmulps %ymm2, %ymm0, %ymm0
 ; AVX-RECIP-NEXT:    vsubps %ymm0, %ymm4, %ymm0
 ; AVX-RECIP-NEXT:    vmulps %ymm0, %ymm2, %ymm0
 ; AVX-RECIP-NEXT:    vaddps %ymm0, %ymm2, %ymm0
 ; AVX-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm0
 ; AVX-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm1, %ymm1
 ; AVX-RECIP-NEXT:    retq
 ;
 ; FMA-RECIP-LABEL: v16f32_two_step2:
 ; FMA-RECIP:       # %bb.0:
 ; FMA-RECIP-NEXT:    vrcpps %ymm1, %ymm2
-; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; FMA-RECIP-NEXT:    vmovaps {{.*#+}} ymm3 = [1,1,1,1,1,1,1,1]
 ; FMA-RECIP-NEXT:    vmovaps %ymm2, %ymm4
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm4 = -(ymm1 * ymm4) + ymm3
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm4 = (ymm4 * ymm2) + ymm2
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm1 = -(ymm4 * ymm1) + ymm3
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm1 = (ymm1 * ymm4) + ymm4
 ; FMA-RECIP-NEXT:    vrcpps %ymm0, %ymm2
 ; FMA-RECIP-NEXT:    vmovaps %ymm2, %ymm4
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm4 = -(ymm0 * ymm4) + ymm3
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm4 = (ymm4 * ymm2) + ymm2
 ; FMA-RECIP-NEXT:    vfnmadd213ps {{.*#+}} ymm0 = -(ymm4 * ymm0) + ymm3
 ; FMA-RECIP-NEXT:    vfmadd132ps {{.*#+}} ymm0 = (ymm0 * ymm4) + ymm4
 ; FMA-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm0
 ; FMA-RECIP-NEXT:    vmulps {{.*}}(%rip), %ymm1, %ymm1
 ; FMA-RECIP-NEXT:    retq
 ;
 ; BTVER2-LABEL: v16f32_two_step2:
 ; BTVER2:       # %bb.0:
-; BTVER2-NEXT:    vmovaps {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [5:1.00]
+; BTVER2-NEXT:    vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1] sched: [5:1.00]
 ; BTVER2-NEXT:    vrcpps %ymm1, %ymm2 # sched: [2:2.00]
 ; BTVER2-NEXT:    vmulps %ymm2, %ymm1, %ymm3 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm3, %ymm4, %ymm3 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm3, %ymm2, %ymm3 # sched: [2:2.00]
 ; BTVER2-NEXT:    vaddps %ymm3, %ymm2, %ymm2 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm2, %ymm1, %ymm1 # sched: [2:2.00]
 ; BTVER2-NEXT:    vsubps %ymm1, %ymm4, %ymm1 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps %ymm1, %ymm2, %ymm1 # sched: [2:2.00]

 ; BTVER2-NEXT:    vaddps %ymm0, %ymm2, %ymm0 # sched: [3:2.00]
 ; BTVER2-NEXT:    vmulps {{.*}}(%rip), %ymm0, %ymm0 # sched: [7:2.00]
 ; BTVER2-NEXT:    retq # sched: [4:1.00]
 ;
 ; SANDY-LABEL: v16f32_two_step2:
 ; SANDY:       # %bb.0:
 ; SANDY-NEXT:    vrcpps %ymm1, %ymm2 # sched: [7:2.00]
 ; SANDY-NEXT:    vmulps %ymm2, %ymm1, %ymm3 # sched: [5:1.00]
-; SANDY-NEXT:    vmovaps {{.*#+}} ymm4 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00] sched: [7:0.50]
+; SANDY-NEXT:    vmovaps {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1] sched: [7:0.50]
 ; SANDY-NEXT:    vsubps %ymm3, %ymm4, %ymm3 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm3, %ymm2, %ymm3 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm3, %ymm2, %ymm2 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm2, %ymm1, %ymm1 # sched: [5:1.00]
 ; SANDY-NEXT:    vsubps %ymm1, %ymm4, %ymm1 # sched: [3:1.00]
 ; SANDY-NEXT:    vmulps %ymm1, %ymm2, %ymm1 # sched: [5:1.00]
 ; SANDY-NEXT:    vaddps %ymm1, %ymm2, %ymm1 # sched: [3:1.00]
 ; SANDY-NEXT:    vrcpps %ymm0, %ymm2 # sched: [7:2.00]

llvm/trunk/test/CodeGen/X86/select_const.ll

 }
 
 define <2 x double> @sel_constants_fmul_constant_vec(i1 %cond) {
 ; CHECK-LABEL: sel_constants_fmul_constant_vec:
 ; CHECK:       # %bb.0:
 ; CHECK-NEXT:    testb $1, %dil
 ; CHECK-NEXT:    jne .LBB37_1
 ; CHECK-NEXT:  # %bb.2:
-; CHECK-NEXT:    movaps {{.*#+}} xmm0 = [1.188300e+02,3.454000e+01]
+; CHECK-NEXT:    movaps {{.*#+}} xmm0 = [118.83,34.539999999999999]
 ; CHECK-NEXT:    retq
 ; CHECK-NEXT:  .LBB37_1:
-; CHECK-NEXT:    movaps {{.*#+}} xmm0 = [-2.040000e+01,3.768000e+01]
+; CHECK-NEXT:    movaps {{.*#+}} xmm0 = [-20.399999999999999,37.68]
 ; CHECK-NEXT:    retq
   %sel = select i1 %cond, <2 x double> <double -4.0, double 12.0>, <2 x double> <double 23.3, double 11.0>
   %bo = fmul <2 x double> %sel, <double 5.1, double 3.14>
   ret <2 x double> %bo
 }
 
 ; 4294967297 = 0x100000001.
 ; This becomes an opaque constant via ConstantHoisting, so we don't fold it into the select.

llvm/trunk/test/CodeGen/X86/sqrt-fastmath.ll

 }
 
 define <4 x float> @sqrt_v4f32_check_denorms(<4 x float> %x) #3 {
 ; SSE-LABEL: sqrt_v4f32_check_denorms:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    rsqrtps %xmm0, %xmm2
 ; SSE-NEXT:    movaps %xmm0, %xmm1
 ; SSE-NEXT:    mulps %xmm2, %xmm1
-; SSE-NEXT:    movaps {{.*#+}} xmm3 = [-5.000000e-01,-5.000000e-01,-5.000000e-01,-5.000000e-01]
+; SSE-NEXT:    movaps {{.*#+}} xmm3 = [-0.5,-0.5,-0.5,-0.5]
 ; SSE-NEXT:    mulps %xmm1, %xmm3
 ; SSE-NEXT:    mulps %xmm2, %xmm1
 ; SSE-NEXT:    addps {{.*}}(%rip), %xmm1
 ; SSE-NEXT:    mulps %xmm3, %xmm1
 ; SSE-NEXT:    andps {{.*}}(%rip), %xmm0
-; SSE-NEXT:    movaps {{.*#+}} xmm2 = [1.175494e-38,1.175494e-38,1.175494e-38,1.175494e-38]
+; SSE-NEXT:    movaps {{.*#+}} xmm2 = [1.17549435E-38,1.17549435E-38,1.17549435E-38,1.17549435E-38]
 ; SSE-NEXT:    cmpleps %xmm0, %xmm2
 ; SSE-NEXT:    andps %xmm2, %xmm1
 ; SSE-NEXT:    movaps %xmm1, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: sqrt_v4f32_check_denorms:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vrsqrtps %xmm0, %xmm1
 ; AVX1-NEXT:    vmulps %xmm1, %xmm0, %xmm2
 ; AVX1-NEXT:    vmulps {{.*}}(%rip), %xmm2, %xmm3
 ; AVX1-NEXT:    vmulps %xmm1, %xmm2, %xmm1
 ; AVX1-NEXT:    vaddps {{.*}}(%rip), %xmm1, %xmm1
 ; AVX1-NEXT:    vmulps %xmm1, %xmm3, %xmm1
 ; AVX1-NEXT:    vandps {{.*}}(%rip), %xmm0, %xmm0
-; AVX1-NEXT:    vmovaps {{.*#+}} xmm2 = [1.175494e-38,1.175494e-38,1.175494e-38,1.175494e-38]
+; AVX1-NEXT:    vmovaps {{.*#+}} xmm2 = [1.17549435E-38,1.17549435E-38,1.17549435E-38,1.17549435E-38]
 ; AVX1-NEXT:    vcmpleps %xmm0, %xmm2, %xmm0
 ; AVX1-NEXT:    vandps %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX512-LABEL: sqrt_v4f32_check_denorms:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vrsqrtps %xmm0, %xmm1
 ; AVX512-NEXT:    vmulps %xmm1, %xmm0, %xmm2
; AVX512-NEXT:    retq
   %div = fdiv fast float 1.0, %sqrt
   ret float %div
 }
 
 define <4 x float> @v4f32_no_estimate(<4 x float> %x) #0 {
 ; SSE-LABEL: v4f32_no_estimate:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    sqrtps %xmm0, %xmm1
-; SSE-NEXT:    movaps {{.*#+}} xmm0 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [1,1,1,1]
 ; SSE-NEXT:    divps %xmm1, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: v4f32_no_estimate:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vsqrtps %xmm0, %xmm0
-; AVX1-NEXT:    vmovaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX1-NEXT:    vmovaps {{.*#+}} xmm1 = [1,1,1,1]
 ; AVX1-NEXT:    vdivps %xmm0, %xmm1, %xmm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX512-LABEL: v4f32_no_estimate:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vsqrtps %xmm0, %xmm0
 ; AVX512-NEXT:    vbroadcastss {{.*#+}} xmm1 = [1,1,1,1]
 ; AVX512-NEXT:    vdivps %xmm0, %xmm1, %xmm0
; AVX512-NEXT:    retq
   ret <4 x float> %div
 }
 
 define <8 x float> @v8f32_no_estimate(<8 x float> %x) #0 {
 ; SSE-LABEL: v8f32_no_estimate:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    sqrtps %xmm1, %xmm2
 ; SSE-NEXT:    sqrtps %xmm0, %xmm3
-; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm1 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm1, %xmm0
 ; SSE-NEXT:    divps %xmm3, %xmm0
 ; SSE-NEXT:    divps %xmm2, %xmm1
 ; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: v8f32_no_estimate:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vsqrtps %ymm0, %ymm0
-; AVX1-NEXT:    vmovaps {{.*#+}} ymm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX1-NEXT:    vmovaps {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
 ; AVX1-NEXT:    vdivps %ymm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX512-LABEL: v8f32_no_estimate:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vsqrtps %ymm0, %ymm0
 ; AVX512-NEXT:    vbroadcastss {{.*#+}} ymm1 = [1,1,1,1,1,1,1,1]
 ; AVX512-NEXT:    vdivps %ymm0, %ymm1, %ymm0
 ; AVX512-NEXT:    retq
   %sqrt = tail call <8 x float> @llvm.sqrt.v8f32(<8 x float> %x)
   %div = fdiv fast <8 x float> <float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>, %sqrt
   ret <8 x float> %div
 }
 
 define <8 x float> @v8f32_estimate(<8 x float> %x) #1 {
 ; SSE-LABEL: v8f32_estimate:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    rsqrtps %xmm0, %xmm3
-; SSE-NEXT:    movaps {{.*#+}} xmm4 = [-5.000000e-01,-5.000000e-01,-5.000000e-01,-5.000000e-01]
+; SSE-NEXT:    movaps {{.*#+}} xmm4 = [-0.5,-0.5,-0.5,-0.5]
 ; SSE-NEXT:    movaps %xmm3, %xmm2
 ; SSE-NEXT:    mulps %xmm3, %xmm2
 ; SSE-NEXT:    mulps %xmm0, %xmm2
-; SSE-NEXT:    movaps {{.*#+}} xmm0 = [-3.000000e+00,-3.000000e+00,-3.000000e+00,-3.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm0 = [-3,-3,-3,-3]
 ; SSE-NEXT:    addps %xmm0, %xmm2
 ; SSE-NEXT:    mulps %xmm4, %xmm2
 ; SSE-NEXT:    mulps %xmm3, %xmm2
 ; SSE-NEXT:    rsqrtps %xmm1, %xmm5
 ; SSE-NEXT:    movaps %xmm5, %xmm3
 ; SSE-NEXT:    mulps %xmm5, %xmm3
 ; SSE-NEXT:    mulps %xmm1, %xmm3
 ; SSE-NEXT:    addps %xmm0, %xmm3

 
 define <16 x float> @v16f32_no_estimate(<16 x float> %x) #0 {
 ; SSE-LABEL: v16f32_no_estimate:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    sqrtps %xmm3, %xmm4
 ; SSE-NEXT:    sqrtps %xmm2, %xmm5
 ; SSE-NEXT:    sqrtps %xmm1, %xmm2
 ; SSE-NEXT:    sqrtps %xmm0, %xmm1
-; SSE-NEXT:    movaps {{.*#+}} xmm3 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm3 = [1,1,1,1]
 ; SSE-NEXT:    movaps %xmm3, %xmm0
 ; SSE-NEXT:    divps %xmm1, %xmm0
 ; SSE-NEXT:    movaps %xmm3, %xmm1
 ; SSE-NEXT:    divps %xmm2, %xmm1
 ; SSE-NEXT:    movaps %xmm3, %xmm2
 ; SSE-NEXT:    divps %xmm5, %xmm2
 ; SSE-NEXT:    divps %xmm4, %xmm3
 ; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: v16f32_no_estimate:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vsqrtps %ymm1, %ymm1
 ; AVX1-NEXT:    vsqrtps %ymm0, %ymm0
-; AVX1-NEXT:    vmovaps {{.*#+}} ymm2 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; AVX1-NEXT:    vmovaps {{.*#+}} ymm2 = [1,1,1,1,1,1,1,1]
 ; AVX1-NEXT:    vdivps %ymm0, %ymm2, %ymm0
 ; AVX1-NEXT:    vdivps %ymm1, %ymm2, %ymm1
 ; AVX1-NEXT:    retq
 ;
 ; AVX512-LABEL: v16f32_no_estimate:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vsqrtps %zmm0, %zmm0
 ; AVX512-NEXT:    vbroadcastss {{.*#+}} zmm1 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
 ; AVX512-NEXT:    vdivps %zmm0, %zmm1, %zmm0
 ; AVX512-NEXT:    retq
   %sqrt = tail call <16 x float> @llvm.sqrt.v16f32(<16 x float> %x)
   %div = fdiv fast <16 x float> <float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0, float 1.0>, %sqrt
   ret <16 x float> %div
 }
 
 define <16 x float> @v16f32_estimate(<16 x float> %x) #1 {
 ; SSE-LABEL: v16f32_estimate:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movaps %xmm1, %xmm4
 ; SSE-NEXT:    movaps %xmm0, %xmm1
 ; SSE-NEXT:    rsqrtps %xmm0, %xmm5
-; SSE-NEXT:    movaps {{.*#+}} xmm6 = [-5.000000e-01,-5.000000e-01,-5.000000e-01,-5.000000e-01]
+; SSE-NEXT:    movaps {{.*#+}} xmm6 = [-0.5,-0.5,-0.5,-0.5]
 ; SSE-NEXT:    movaps %xmm5, %xmm0
 ; SSE-NEXT:    mulps %xmm5, %xmm0
 ; SSE-NEXT:    mulps %xmm1, %xmm0
-; SSE-NEXT:    movaps {{.*#+}} xmm7 = [-3.000000e+00,-3.000000e+00,-3.000000e+00,-3.000000e+00]
+; SSE-NEXT:    movaps {{.*#+}} xmm7 = [-3,-3,-3,-3]
 ; SSE-NEXT:    addps %xmm7, %xmm0
 ; SSE-NEXT:    mulps %xmm6, %xmm0
 ; SSE-NEXT:    mulps %xmm5, %xmm0
 ; SSE-NEXT:    rsqrtps %xmm4, %xmm5
 ; SSE-NEXT:    movaps %xmm5, %xmm1
 ; SSE-NEXT:    mulps %xmm5, %xmm1
 ; SSE-NEXT:    mulps %xmm4, %xmm1
 ; SSE-NEXT:    addps %xmm7, %xmm1

 ; SSE-NEXT:    mulps %xmm2, %xmm5
 ; SSE-NEXT:    movaps %xmm4, %xmm2
 ; SSE-NEXT:    movaps %xmm5, %xmm3
 ; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: v16f32_estimate:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vrsqrtps %ymm0, %ymm2
-; AVX1-NEXT:    vmovaps {{.*#+}} ymm3 = [-5.000000e-01,-5.000000e-01,-5.000000e-01,-5.000000e-01,-5.000000e-01,-5.000000e-01,-5.000000e-01,-5.000000e-01]
+; AVX1-NEXT:    vmovaps {{.*#+}} ymm3 = [-0.5,-0.5,-0.5,-0.5,-0.5,-0.5,-0.5,-0.5]
 ; AVX1-NEXT:    vmulps %ymm2, %ymm2, %ymm4
 ; AVX1-NEXT:    vmulps %ymm4, %ymm0, %ymm0
-; AVX1-NEXT:    vmovaps {{.*#+}} ymm4 = [-3.000000e+00,-3.000000e+00,-3.000000e+00,-3.000000e+00,-3.000000e+00,-3.000000e+00,-3.000000e+00,-3.000000e+00]
+; AVX1-NEXT:    vmovaps {{.*#+}} ymm4 = [-3,-3,-3,-3,-3,-3,-3,-3]
 ; AVX1-NEXT:    vaddps %ymm4, %ymm0, %ymm0
 ; AVX1-NEXT:    vmulps %ymm0, %ymm3, %ymm0
 ; AVX1-NEXT:    vmulps %ymm0, %ymm2, %ymm0
 ; AVX1-NEXT:    vrsqrtps %ymm1, %ymm2
 ; AVX1-NEXT:    vmulps %ymm2, %ymm2, %ymm5
 ; AVX1-NEXT:    vmulps %ymm5, %ymm1, %ymm1
 ; AVX1-NEXT:    vaddps %ymm4, %ymm1, %ymm1
 ; AVX1-NEXT:    vmulps %ymm1, %ymm3, %ymm1

llvm/trunk/test/CodeGen/X86/sse2.ll

; X64-AVX-NEXT:    retq
   %tmp7 = insertelement <2 x double> %tmp, double %b, i32 1    ; <<2 x double>> [#uses=1]
   ret <2 x double> %tmp7
 }
 
 define void @test12() nounwind {
 ; SSE-LABEL: test12:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movapd 0, %xmm0
-; SSE-NEXT:    movapd {{.*#+}} xmm1 = [1.000000e+00,1.000000e+00,1.000000e+00,1.000000e+00]
+; SSE-NEXT:    movapd {{.*#+}} xmm1 = [1,1,1,1]
 ; SSE-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
 ; SSE-NEXT:    xorps %xmm2, %xmm2
 ; SSE-NEXT:    movhlps {{.*#+}} xmm2 = xmm0[1],xmm2[1]
 ; SSE-NEXT:    addps %xmm1, %xmm2
 ; SSE-NEXT:    movaps %xmm2, 0
 ; SSE-NEXT:    ret{{[l|q]}}
 ;
 ; AVX1-LABEL: test12:

llvm/trunk/test/CodeGen/X86/subvector-broadcast.ll

 
 
 @ga2 = global <4 x double> zeroinitializer, align 8
 @gb2 = global <8 x double> zeroinitializer, align 8
 
 define void @fallback_broadcast_v4f64_to_v8f64(<4 x double> %a, <8 x double> %b) {
 ; X32-AVX-LABEL: fallback_broadcast_v4f64_to_v8f64:
 ; X32-AVX:       # %bb.0: # %entry
-; X32-AVX-NEXT:    vmovapd {{.*#+}} ymm3 = [1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00]
+; X32-AVX-NEXT:    vmovapd {{.*#+}} ymm3 = [1,2,3,4]
 ; X32-AVX-NEXT:    vaddpd %ymm3, %ymm0, %ymm0
 ; X32-AVX-NEXT:    vaddpd %ymm3, %ymm2, %ymm2
 ; X32-AVX-NEXT:    vaddpd %ymm3, %ymm1, %ymm1
 ; X32-AVX-NEXT:    vdivpd %ymm3, %ymm1, %ymm1
 ; X32-AVX-NEXT:    vdivpd %ymm3, %ymm2, %ymm2
 ; X32-AVX-NEXT:    vmovupd %ymm0, ga2
 ; X32-AVX-NEXT:    vmovupd %ymm2, gb2+32
 ; X32-AVX-NEXT:    vmovupd %ymm1, gb2
 ; X32-AVX-NEXT:    vzeroupper
 ; X32-AVX-NEXT:    retl
 ;
 ; X32-AVX512-LABEL: fallback_broadcast_v4f64_to_v8f64:
 ; X32-AVX512:       # %bb.0: # %entry
-; X32-AVX512-NEXT:    vmovapd {{.*#+}} ymm2 = [1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00]
+; X32-AVX512-NEXT:    vmovapd {{.*#+}} ymm2 = [1,2,3,4]
 ; X32-AVX512-NEXT:    vaddpd %ymm2, %ymm0, %ymm0
 ; X32-AVX512-NEXT:    vinsertf64x4 $1, %ymm2, %zmm2, %zmm2
 ; X32-AVX512-NEXT:    vaddpd %zmm2, %zmm1, %zmm1
 ; X32-AVX512-NEXT:    vdivpd %zmm2, %zmm1, %zmm1
 ; X32-AVX512-NEXT:    vmovupd %ymm0, ga2
 ; X32-AVX512-NEXT:    vmovupd %zmm1, gb2
 ; X32-AVX512-NEXT:    vzeroupper
 ; X32-AVX512-NEXT:    retl
 ;
 ; X64-AVX-LABEL: fallback_broadcast_v4f64_to_v8f64:
 ; X64-AVX:       # %bb.0: # %entry
-; X64-AVX-NEXT:    vmovapd {{.*#+}} ymm3 = [1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00]
+; X64-AVX-NEXT:    vmovapd {{.*#+}} ymm3 = [1,2,3,4]
 ; X64-AVX-NEXT:    vaddpd %ymm3, %ymm0, %ymm0
 ; X64-AVX-NEXT:    vaddpd %ymm3, %ymm2, %ymm2
 ; X64-AVX-NEXT:    vaddpd %ymm3, %ymm1, %ymm1
 ; X64-AVX-NEXT:    vdivpd %ymm3, %ymm1, %ymm1
 ; X64-AVX-NEXT:    vdivpd %ymm3, %ymm2, %ymm2
 ; X64-AVX-NEXT:    vmovupd %ymm0, {{.*}}(%rip)
 ; X64-AVX-NEXT:    vmovupd %ymm2, gb2+{{.*}}(%rip)
 ; X64-AVX-NEXT:    vmovupd %ymm1, {{.*}}(%rip)
 ; X64-AVX-NEXT:    vzeroupper
 ; X64-AVX-NEXT:    retq
 ;
 ; X64-AVX512-LABEL: fallback_broadcast_v4f64_to_v8f64:
 ; X64-AVX512:       # %bb.0: # %entry
-; X64-AVX512-NEXT:    vmovapd {{.*#+}} ymm2 = [1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00]
+; X64-AVX512-NEXT:    vmovapd {{.*#+}} ymm2 = [1,2,3,4]
 ; X64-AVX512-NEXT:    vaddpd %ymm2, %ymm0, %ymm0
 ; X64-AVX512-NEXT:    vinsertf64x4 $1, %ymm2, %zmm2, %zmm2
 ; X64-AVX512-NEXT:    vaddpd %zmm2, %zmm1, %zmm1
 ; X64-AVX512-NEXT:    vdivpd %zmm2, %zmm1, %zmm1
 ; X64-AVX512-NEXT:    vmovupd %ymm0, {{.*}}(%rip)
 ; X64-AVX512-NEXT:    vmovupd %zmm1, {{.*}}(%rip)
 ; X64-AVX512-NEXT:    vzeroupper
 ; X64-AVX512-NEXT:    retq

llvm/trunk/test/CodeGen/X86/unfold-masked-merge-vector-variablemask-const.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+sse,-sse2 < %s | FileCheck %s --check-prefixes=CHECK,CHECK-SSE,CHECK-SSE1
 ; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+sse,+sse2 < %s | FileCheck %s --check-prefixes=CHECK,CHECK-SSE,CHECK-SSE2
 ; RUN: llc -mtriple=x86_64-unknown-linux-gnu -mattr=+xop < %s | FileCheck %s --check-prefixes=CHECK,CHECK-XOP
 
 ; ============================================================================ ;
 ; Various cases with %x and/or %y being a constant
 ; ============================================================================ ;
 
 define <4 x i32> @out_constant_varx_mone(<4 x i32> *%px, <4 x i32> *%py, <4 x i32> *%pmask) {
 ; CHECK-SSE1-LABEL: out_constant_varx_mone:
 ; CHECK-SSE1:       # %bb.0:
 ; CHECK-SSE1-NEXT:    movq %rdi, %rax
 ; CHECK-SSE1-NEXT:    movaps (%rcx), %xmm0
-; CHECK-SSE1-NEXT:    movaps {{.*#+}} xmm1 = [nan,nan,nan,nan]
+; CHECK-SSE1-NEXT:    movaps {{.*#+}} xmm1 = [NaN,NaN,NaN,NaN]
 ; CHECK-SSE1-NEXT:    xorps %xmm0, %xmm1
 ; CHECK-SSE1-NEXT:    andps (%rsi), %xmm0
 ; CHECK-SSE1-NEXT:    orps %xmm1, %xmm0
 ; CHECK-SSE1-NEXT:    movaps %xmm0, (%rdi)
 ; CHECK-SSE1-NEXT:    retq
 ;
 ; CHECK-SSE2-LABEL: out_constant_varx_mone:
 ; CHECK-SSE2:       # %bb.0:

 }
 
 ; This is not a canonical form. Testing for completeness only.
 define <4 x i32> @in_constant_varx_mone_invmask(<4 x i32> *%px, <4 x i32> *%py, <4 x i32> *%pmask) {
 ; CHECK-SSE1-LABEL: in_constant_varx_mone_invmask:
 ; CHECK-SSE1:       # %bb.0:
 ; CHECK-SSE1-NEXT:    movq %rdi, %rax
 ; CHECK-SSE1-NEXT:    movaps (%rsi), %xmm0
-; CHECK-SSE1-NEXT:    movaps {{.*#+}} xmm1 = [nan,nan,nan,nan]
+; CHECK-SSE1-NEXT:    movaps {{.*#+}} xmm1 = [NaN,NaN,NaN,NaN]
 ; CHECK-SSE1-NEXT:    movaps (%rcx), %xmm2
 ; CHECK-SSE1-NEXT:    xorps %xmm1, %xmm2
 ; CHECK-SSE1-NEXT:    andnps %xmm2, %xmm0
 ; CHECK-SSE1-NEXT:    xorps %xmm1, %xmm0
 ; CHECK-SSE1-NEXT:    movaps %xmm0, (%rdi)
 ; CHECK-SSE1-NEXT:    retq
 ;
 ; CHECK-SSE2-LABEL: in_constant_varx_mone_invmask:

 }
 
 ; This is not a canonical form. Testing for completeness only.
 define <4 x i32> @out_constant_mone_vary_invmask(<4 x i32> *%px, <4 x i32> *%py, <4 x i32> *%pmask) {
 ; CHECK-SSE1-LABEL: out_constant_mone_vary_invmask:
 ; CHECK-SSE1:       # %bb.0:
 ; CHECK-SSE1-NEXT:    movq %rdi, %rax
 ; CHECK-SSE1-NEXT:    movaps (%rcx), %xmm0
-; CHECK-SSE1-NEXT:    movaps {{.*#+}} xmm1 = [nan,nan,nan,nan]
+; CHECK-SSE1-NEXT:    movaps {{.*#+}} xmm1 = [NaN,NaN,NaN,NaN]
 ; CHECK-SSE1-NEXT:    xorps %xmm0, %xmm1
 ; CHECK-SSE1-NEXT:    andps (%rdx), %xmm0
 ; CHECK-SSE1-NEXT:    orps %xmm1, %xmm0
 ; CHECK-SSE1-NEXT:    movaps %xmm0, (%rdi)
 ; CHECK-SSE1-NEXT:    retq
 ;
 ; CHECK-SSE2-LABEL: out_constant_mone_vary_invmask:
 ; CHECK-SSE2:       # %bb.0:

 }
 
 ; This is not a canonical form. Testing for completeness only.
 define <4 x i32> @in_constant_mone_vary_invmask(<4 x i32> *%px, <4 x i32> *%py, <4 x i32> *%pmask) {
 ; CHECK-SSE1-LABEL: in_constant_mone_vary_invmask:
 ; CHECK-SSE1:       # %bb.0:
 ; CHECK-SSE1-NEXT:    movq %rdi, %rax
 ; CHECK-SSE1-NEXT:    movaps (%rcx), %xmm0
-; CHECK-SSE1-NEXT:    movaps {{.*#+}} xmm1 = [nan,nan,nan,nan]
+; CHECK-SSE1-NEXT:    movaps {{.*#+}} xmm1 = [NaN,NaN,NaN,NaN]
 ; CHECK-SSE1-NEXT:    xorps %xmm0, %xmm1
 ; CHECK-SSE1-NEXT:    andps (%rdx), %xmm0
 ; CHECK-SSE1-NEXT:    orps %xmm1, %xmm0
 ; CHECK-SSE1-NEXT:    movaps %xmm0, (%rdi)
 ; CHECK-SSE1-NEXT:    retq
 ;
 ; CHECK-SSE2-LABEL: in_constant_mone_vary_invmask:
 ; CHECK-SSE2:       # %bb.0:
; CHECK-XOP-NEXT:    retq
   ret <4 x i32> %r
 }
 
 define <4 x i32> @out_constant_42_vary(<4 x i32> *%px, <4 x i32> *%py, <4 x i32> *%pmask) {
 ; CHECK-SSE1-LABEL: out_constant_42_vary:
 ; CHECK-SSE1:       # %bb.0:
 ; CHECK-SSE1-NEXT:    movq %rdi, %rax
 ; CHECK-SSE1-NEXT:    movaps (%rcx), %xmm0
-; CHECK-SSE1-NEXT:    movaps {{.*#+}} xmm1 = [5.885454e-44,5.885454e-44,5.885454e-44,5.885454e-44]
+; CHECK-SSE1-NEXT:    movaps {{.*#+}} xmm1 = [5.88545355E-44,5.88545355E-44,5.88545355E-44,5.88545355E-44]
 ; CHECK-SSE1-NEXT:    andps %xmm0, %xmm1
 ; CHECK-SSE1-NEXT:    andnps (%rdx), %xmm0
 ; CHECK-SSE1-NEXT:    orps %xmm1, %xmm0
 ; CHECK-SSE1-NEXT:    movaps %xmm0, (%rdi)
 ; CHECK-SSE1-NEXT:    retq
 ;
 ; CHECK-SSE2-LABEL: out_constant_42_vary:
 ; CHECK-SSE2:       # %bb.0:

llvm/trunk/test/CodeGen/X86/v4f32-immediate.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=i686-unknown -mattr=+sse | FileCheck %s --check-prefix=X32
 ; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+sse | FileCheck %s --check-prefix=X64
 
 define <4 x float> @foo() {
 ; X32-LABEL: foo:
 ; X32:       # %bb.0:
-; X32-NEXT:    movaps {{.*#+}} xmm0 = [3.223542e+00,2.300000e+00,1.200000e+00,1.000000e-01]
+; X32-NEXT:    movaps {{.*#+}} xmm0 = [3.22354245,2.29999995,1.20000005,0.100000001]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: foo:
 ; X64:       # %bb.0:
-; X64-NEXT:    movaps {{.*#+}} xmm0 = [3.223542e+00,2.300000e+00,1.200000e+00,1.000000e-01]
+; X64-NEXT:    movaps {{.*#+}} xmm0 = [3.22354245,2.29999995,1.20000005,0.100000001]
 ; X64-NEXT:    retq
   ret <4 x float> <float 0x4009C9D0A0000000, float 0x4002666660000000, float 0x3FF3333340000000, float 0x3FB99999A0000000>
 }

llvm/trunk/test/CodeGen/X86/vec_cast3.ll

; CHECK-WIDE-NEXT:    retl
   ret <2 x float> %res
 }
 
 define <2 x float> @cvt_v2u32_v2f32(<2 x i32> %src) {
 ; CHECK-LABEL: cvt_v2u32_v2f32:
 ; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    vxorps %xmm1, %xmm1, %xmm1
 ; CHECK-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
-; CHECK-NEXT:    vmovaps {{.*#+}} xmm1 = [4.503600e+15,4.503600e+15]
+; CHECK-NEXT:    vmovaps {{.*#+}} xmm1 = [4503599627370496,4503599627370496]
 ; CHECK-NEXT:    vorps %xmm1, %xmm0, %xmm0
 ; CHECK-NEXT:    vsubpd %xmm1, %xmm0, %xmm0
 ; CHECK-NEXT:    vcvtpd2ps %xmm0, %xmm0
 ; CHECK-NEXT:    retl
 ;
 ; CHECK-WIDE-LABEL: cvt_v2u32_v2f32:
 ; CHECK-WIDE:       ## %bb.0:
 ; CHECK-WIDE-NEXT:    vpmovzxdq {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero
-; CHECK-WIDE-NEXT:    vmovdqa {{.*#+}} xmm1 = [4.503600e+15,4.503600e+15]
+; CHECK-WIDE-NEXT:    vmovdqa {{.*#+}} xmm1 = [4503599627370496,4503599627370496]
 ; CHECK-WIDE-NEXT:    vpor %xmm1, %xmm0, %xmm0
 ; CHECK-WIDE-NEXT:    vsubpd %xmm1, %xmm0, %xmm0
 ; CHECK-WIDE-NEXT:    vcvtpd2ps %xmm0, %xmm0
 ; CHECK-WIDE-NEXT:    retl
   %res = uitofp <2 x i32> %src to <2 x float>
   ret <2 x float> %res
 }
 

llvm/trunk/test/CodeGen/X86/vec_fabs.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=X32 --check-prefix=X32_AVX
 ; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx512vl | FileCheck %s --check-prefix=X32 --check-prefix=X32_AVX512VL
 ; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx512dq,+avx512vl | FileCheck %s --check-prefix=X32 --check-prefix=X32_AVX512VLDQ
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx | FileCheck %s --check-prefix=X64 --check-prefix=X64_AVX
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512vl | FileCheck %s --check-prefix=X64 --check-prefix=X64_AVX512VL
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512dq,+avx512vl | FileCheck %s --check-prefix=X64 --check-prefix=X64_AVX512VLDQ
 
-; FIXME: Drop the regex pattern matching of 'nan' once we drop support for MSVC
-; 2013.
-
 define <2 x double> @fabs_v2f64(<2 x double> %p) {
 ; X32-LABEL: fabs_v2f64:
 ; X32:       # %bb.0:
 ; X32-NEXT:    vandps {{\.LCPI.*}}, %xmm0, %xmm0
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: fabs_v2f64:
 ; X64:       # %bb.0:
; X64_AVX512VLDQ-NEXT:    retq
   %t = call <8 x float> @llvm.fabs.v8f32(<8 x float> %p)
   ret <8 x float> %t
 }
 declare <8 x float> @llvm.fabs.v8f32(<8 x float> %p)
 
 define <8 x double> @fabs_v8f64(<8 x double> %p) {
 ; X32_AVX-LABEL: fabs_v8f64:
 ; X32_AVX:       # %bb.0:
-; X32_AVX-NEXT:    vmovaps {{.*#+}} ymm2 = [{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}}]
+; X32_AVX-NEXT:    vmovaps {{.*#+}} ymm2 = [NaN,NaN,NaN,NaN]
 ; X32_AVX-NEXT:    vandps %ymm2, %ymm0, %ymm0
 ; X32_AVX-NEXT:    vandps %ymm2, %ymm1, %ymm1
 ; X32_AVX-NEXT:    retl
 ;
 ; X32_AVX512VL-LABEL: fabs_v8f64:
 ; X32_AVX512VL:       # %bb.0:
 ; X32_AVX512VL-NEXT:    vpandq {{\.LCPI.*}}{1to8}, %zmm0, %zmm0
 ; X32_AVX512VL-NEXT:    retl
 ;
 ; X32_AVX512VLDQ-LABEL: fabs_v8f64:
 ; X32_AVX512VLDQ:       # %bb.0:
 ; X32_AVX512VLDQ-NEXT:    vandpd {{\.LCPI.*}}{1to8}, %zmm0, %zmm0
 ; X32_AVX512VLDQ-NEXT:    retl
 ;
 ; X64_AVX-LABEL: fabs_v8f64:
 ; X64_AVX:       # %bb.0:
-; X64_AVX-NEXT:    vmovaps {{.*#+}} ymm2 = [{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}}]
+; X64_AVX-NEXT:    vmovaps {{.*#+}} ymm2 = [NaN,NaN,NaN,NaN]
 ; X64_AVX-NEXT:    vandps %ymm2, %ymm0, %ymm0
 ; X64_AVX-NEXT:    vandps %ymm2, %ymm1, %ymm1
 ; X64_AVX-NEXT:    retq
 ;
 ; X64_AVX512VL-LABEL: fabs_v8f64:
 ; X64_AVX512VL:       # %bb.0:
 ; X64_AVX512VL-NEXT:    vpandq {{.*}}(%rip){1to8}, %zmm0, %zmm0
 ; X64_AVX512VL-NEXT:    retq
 ;
 ; X64_AVX512VLDQ-LABEL: fabs_v8f64:
 ; X64_AVX512VLDQ:       # %bb.0:
 ; X64_AVX512VLDQ-NEXT:    vandpd {{.*}}(%rip){1to8}, %zmm0, %zmm0
 ; X64_AVX512VLDQ-NEXT:    retq
   %t = call <8 x double> @llvm.fabs.v8f64(<8 x double> %p)
   ret <8 x double> %t
 }
 declare <8 x double> @llvm.fabs.v8f64(<8 x double> %p)
 
 define <16 x float> @fabs_v16f32(<16 x float> %p) {
 ; X32_AVX-LABEL: fabs_v16f32:
 ; X32_AVX:       # %bb.0:
-; X32_AVX-NEXT:    vmovaps {{.*#+}} ymm2 = [{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}}]
+; X32_AVX-NEXT:    vmovaps {{.*#+}} ymm2 = [NaN,NaN,NaN,NaN,NaN,NaN,NaN,NaN]
 ; X32_AVX-NEXT:    vandps %ymm2, %ymm0, %ymm0
 ; X32_AVX-NEXT:    vandps %ymm2, %ymm1, %ymm1
 ; X32_AVX-NEXT:    retl
 ;
 ; X32_AVX512VL-LABEL: fabs_v16f32:
 ; X32_AVX512VL:       # %bb.0:
 ; X32_AVX512VL-NEXT:    vpandd {{\.LCPI.*}}{1to16}, %zmm0, %zmm0
 ; X32_AVX512VL-NEXT:    retl
 ;
 ; X32_AVX512VLDQ-LABEL: fabs_v16f32:
 ; X32_AVX512VLDQ:       # %bb.0:
 ; X32_AVX512VLDQ-NEXT:    vandps {{\.LCPI.*}}{1to16}, %zmm0, %zmm0
 ; X32_AVX512VLDQ-NEXT:    retl
 ;
 ; X64_AVX-LABEL: fabs_v16f32:
 ; X64_AVX:       # %bb.0:
-; X64_AVX-NEXT:    vmovaps {{.*#+}} ymm2 = [{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}},{{(nan|1\.#QNAN0e\+00)}}]
+; X64_AVX-NEXT:    vmovaps {{.*#+}} ymm2 = [NaN,NaN,NaN,NaN,NaN,NaN,NaN,NaN]
 ; X64_AVX-NEXT:    vandps %ymm2, %ymm0, %ymm0
 ; X64_AVX-NEXT:    vandps %ymm2, %ymm1, %ymm1
 ; X64_AVX-NEXT:    retq
 ;
 ; X64_AVX512VL-LABEL: fabs_v16f32:
 ; X64_AVX512VL:       # %bb.0:
 ; X64_AVX512VL-NEXT:    vpandd {{.*}}(%rip){1to16}, %zmm0, %zmm0
 ; X64_AVX512VL-NEXT:    retq

llvm/trunk/test/CodeGen/X86/vec_floor.ll

 
 ;
 ; Constant Folding
 ;
 
 define <2 x double> @const_floor_v2f64() {
 ; SSE41-LABEL: const_floor_v2f64:
 ; SSE41:       ## %bb.0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [-2.000000e+00,2.000000e+00]
+; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [-2,2]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: const_floor_v2f64:
 ; AVX:       ## %bb.0:
-; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [-2.000000e+00,2.000000e+00]
+; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [-2,2]
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: const_floor_v2f64:
 ; AVX512:       ## %bb.0:
-; AVX512-NEXT:    vmovaps {{.*#+}} xmm0 = [-2.000000e+00,2.000000e+00]
+; AVX512-NEXT:    vmovaps {{.*#+}} xmm0 = [-2,2]
 ; AVX512-NEXT:    retq
   %t = call <2 x double> @llvm.floor.v2f64(<2 x double> <double -1.5, double 2.5>)
   ret <2 x double> %t
 }
 
 define <4 x float> @const_floor_v4f32() {
 ; SSE41-LABEL: const_floor_v4f32:
 ; SSE41:       ## %bb.0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [-4.000000e+00,6.000000e+00,-9.000000e+00,2.000000e+00]
+; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [-4,6,-9,2]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: const_floor_v4f32:
 ; AVX:       ## %bb.0:
-; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [-4.000000e+00,6.000000e+00,-9.000000e+00,2.000000e+00]
+; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [-4,6,-9,2]
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: const_floor_v4f32:
 ; AVX512:       ## %bb.0:
-; AVX512-NEXT:    vmovaps {{.*#+}} xmm0 = [-4.000000e+00,6.000000e+00,-9.000000e+00,2.000000e+00]
+; AVX512-NEXT:    vmovaps {{.*#+}} xmm0 = [-4,6,-9,2]
 ; AVX512-NEXT:    retq
   %t = call <4 x float> @llvm.floor.v4f32(<4 x float> <float -3.5, float 6.0, float -9.0, float 2.5>)
   ret <4 x float> %t
 }
 
 define <2 x double> @const_ceil_v2f64() {
 ; SSE41-LABEL: const_ceil_v2f64:
 ; SSE41:       ## %bb.0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [-1.000000e+00,3.000000e+00]
+; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [-1,3]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: const_ceil_v2f64:
 ; AVX:       ## %bb.0:
-; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [-1.000000e+00,3.000000e+00]
+; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [-1,3]
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: const_ceil_v2f64:
 ; AVX512:       ## %bb.0:
-; AVX512-NEXT:    vmovaps {{.*#+}} xmm0 = [-1.000000e+00,3.000000e+00]
+; AVX512-NEXT:    vmovaps {{.*#+}} xmm0 = [-1,3]
 ; AVX512-NEXT:    retq
   %t = call <2 x double> @llvm.ceil.v2f64(<2 x double> <double -1.5, double 2.5>)
   ret <2 x double> %t
 }
 
 define <4 x float> @const_ceil_v4f32() {
 ; SSE41-LABEL: const_ceil_v4f32:
 ; SSE41:       ## %bb.0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [-3.000000e+00,6.000000e+00,-9.000000e+00,3.000000e+00]
+; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [-3,6,-9,3]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: const_ceil_v4f32:
 ; AVX:       ## %bb.0:
-; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [-3.000000e+00,6.000000e+00,-9.000000e+00,3.000000e+00]
+; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [-3,6,-9,3]
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: const_ceil_v4f32:
 ; AVX512:       ## %bb.0:
-; AVX512-NEXT:    vmovaps {{.*#+}} xmm0 = [-3.000000e+00,6.000000e+00,-9.000000e+00,3.000000e+00]
+; AVX512-NEXT:    vmovaps {{.*#+}} xmm0 = [-3,6,-9,3]
 ; AVX512-NEXT:    retq
   %t = call <4 x float> @llvm.ceil.v4f32(<4 x float> <float -3.5, float 6.0, float -9.0, float 2.5>)
   ret <4 x float> %t
 }
 
 define <2 x double> @const_trunc_v2f64() {
 ; SSE41-LABEL: const_trunc_v2f64:
 ; SSE41:       ## %bb.0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [-1.000000e+00,2.000000e+00]
+; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [-1,2]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: const_trunc_v2f64:
 ; AVX:       ## %bb.0:
-; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [-1.000000e+00,2.000000e+00]
+; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [-1,2]
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: const_trunc_v2f64:
 ; AVX512:       ## %bb.0:
-; AVX512-NEXT:    vmovaps {{.*#+}} xmm0 = [-1.000000e+00,2.000000e+00]
+; AVX512-NEXT:    vmovaps {{.*#+}} xmm0 = [-1,2]
 ; AVX512-NEXT:    retq
   %t = call <2 x double> @llvm.trunc.v2f64(<2 x double> <double -1.5, double 2.5>)
   ret <2 x double> %t
 }
 
 define <4 x float> @const_trunc_v4f32() {
 ; SSE41-LABEL: const_trunc_v4f32:
 ; SSE41:       ## %bb.0:
-; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [-3.000000e+00,6.000000e+00,-9.000000e+00,2.000000e+00]
+; SSE41-NEXT:    movaps {{.*#+}} xmm0 = [-3,6,-9,2]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: const_trunc_v4f32:
 ; AVX:       ## %bb.0:
-; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [-3.000000e+00,6.000000e+00,-9.000000e+00,2.000000e+00]
+; AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [-3,6,-9,2]
 ; AVX-NEXT:    retq
 ;
 ; AVX512-LABEL: const_trunc_v4f32:
 ; AVX512:       ## %bb.0:
-; AVX512-NEXT:    vmovaps {{.*#+}} xmm0 = [-3.000000e+00,6.000000e+00,-9.000000e+00,2.000000e+00]
+; AVX512-NEXT:    vmovaps {{.*#+}} xmm0 = [-3,6,-9,2]
 ; AVX512-NEXT:    retq
   %t = call <4 x float> @llvm.trunc.v4f32(<4 x float> <float -3.5, float 6.0, float -9.0, float 2.5>)
   ret <4 x float> %t
 }
 
 ;
 ; Scalar and masked instructions
 ;

llvm/trunk/test/CodeGen/X86/vec_fpext.ll

entry:
   %1 = fpext <8 x float> %0 to <8 x double>
   store <8 x double> %1, <8 x double>* %out, align 1
   ret void
 }
 
 define <2 x double> @fpext_fromconst() {
 ; X32-SSE-LABEL: fpext_fromconst:
 ; X32-SSE:       # %bb.0: # %entry
-; X32-SSE-NEXT:    movaps {{.*#+}} xmm0 = [1.000000e+00,-2.000000e+00]
+; X32-SSE-NEXT:    movaps {{.*#+}} xmm0 = [1,-2]
 ; X32-SSE-NEXT:    # encoding: [0x0f,0x28,0x05,A,A,A,A]
 ; X32-SSE-NEXT:    # fixup A - offset: 3, value: {{\.LCPI.*}}, kind: FK_Data_4
 ; X32-SSE-NEXT:    retl # encoding: [0xc3]
 ;
 ; X32-AVX-LABEL: fpext_fromconst:
 ; X32-AVX:       # %bb.0: # %entry
-; X32-AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [1.000000e+00,-2.000000e+00]
+; X32-AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [1,-2]
 ; X32-AVX-NEXT:    # encoding: [0xc5,0xf8,0x28,0x05,A,A,A,A]
 ; X32-AVX-NEXT:    # fixup A - offset: 4, value: {{\.LCPI.*}}, kind: FK_Data_4
 ; X32-AVX-NEXT:    retl # encoding: [0xc3]
 ;
 ; X32-AVX512VL-LABEL: fpext_fromconst:
 ; X32-AVX512VL:       # %bb.0: # %entry
-; X32-AVX512VL-NEXT:    vmovaps {{\.LCPI.*}}, %xmm0 # EVEX TO VEX Compression xmm0 = [1.000000e+00,-2.000000e+00]
+; X32-AVX512VL-NEXT:    vmovaps {{\.LCPI.*}}, %xmm0 # EVEX TO VEX Compression xmm0 = [1,-2]
 ; X32-AVX512VL-NEXT:    # encoding: [0xc5,0xf8,0x28,0x05,A,A,A,A]
 ; X32-AVX512VL-NEXT:    # fixup A - offset: 4, value: {{\.LCPI.*}}, kind: FK_Data_4
 ; X32-AVX512VL-NEXT:    retl # encoding: [0xc3]
 ;
 ; X64-SSE-LABEL: fpext_fromconst:
 ; X64-SSE:       # %bb.0: # %entry
-; X64-SSE-NEXT:    movaps {{.*#+}} xmm0 = [1.000000e+00,-2.000000e+00]
+; X64-SSE-NEXT:    movaps {{.*#+}} xmm0 = [1,-2]
 ; X64-SSE-NEXT:    # encoding: [0x0f,0x28,0x05,A,A,A,A]
 ; X64-SSE-NEXT:    # fixup A - offset: 3, value: {{\.LCPI.*}}-4, kind: reloc_riprel_4byte
 ; X64-SSE-NEXT:    retq # encoding: [0xc3]
 ;
 ; X64-AVX-LABEL: fpext_fromconst:
 ; X64-AVX:       # %bb.0: # %entry
-; X64-AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [1.000000e+00,-2.000000e+00]
+; X64-AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [1,-2]
 ; X64-AVX-NEXT:    # encoding: [0xc5,0xf8,0x28,0x05,A,A,A,A]
 ; X64-AVX-NEXT:    # fixup A - offset: 4, value: {{\.LCPI.*}}-4, kind: reloc_riprel_4byte
 ; X64-AVX-NEXT:    retq # encoding: [0xc3]
 ;
 ; X64-AVX512VL-LABEL: fpext_fromconst:
 ; X64-AVX512VL:       # %bb.0: # %entry
-; X64-AVX512VL-NEXT:    vmovaps {{.*}}(%rip), %xmm0 # EVEX TO VEX Compression xmm0 = [1.000000e+00,-2.000000e+00]
+; X64-AVX512VL-NEXT:    vmovaps {{.*}}(%rip), %xmm0 # EVEX TO VEX Compression xmm0 = [1,-2]
 ; X64-AVX512VL-NEXT:    # encoding: [0xc5,0xf8,0x28,0x05,A,A,A,A]
 ; X64-AVX512VL-NEXT:    # fixup A - offset: 4, value: {{\.LCPI.*}}-4, kind: reloc_riprel_4byte
 ; X64-AVX512VL-NEXT:    retq # encoding: [0xc3]
 entry:
   %0  = insertelement <2 x float> undef, float 1.0, i32 0
   %1  = insertelement <2 x float> %0, float -2.0, i32 1
   %2  = fpext <2 x float> %1 to <2 x double>
   ret <2 x double> %2
 }

llvm/trunk/test/CodeGen/X86/vec_int_to_fp.ll

 ;
 
 define <2 x double> @uitofp_2i64_to_2f64(<2 x i64> %a) {
 ; SSE-LABEL: uitofp_2i64_to_2f64:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [1127219200,1160773632,0,0]
 ; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
 ; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
-; SSE-NEXT:    movapd {{.*#+}} xmm4 = [4.503600e+15,1.934281e+25]
+; SSE-NEXT:    movapd {{.*#+}} xmm4 = [4503599627370496,1.9342813113834067E+25]
 ; SSE-NEXT:    subpd %xmm4, %xmm0
 ; SSE-NEXT:    movapd %xmm0, %xmm1
 ; SSE-NEXT:    unpckhpd {{.*#+}} xmm1 = xmm1[1],xmm0[1]
 ; SSE-NEXT:    addpd %xmm0, %xmm1
 ; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
 ; SSE-NEXT:    subpd %xmm4, %xmm3
 ; SSE-NEXT:    movapd %xmm3, %xmm0
 ; SSE-NEXT:    unpckhpd {{.*#+}} xmm0 = xmm0[1],xmm3[1]
 ; SSE-NEXT:    addpd %xmm3, %xmm0
 ; SSE-NEXT:    unpcklpd {{.*#+}} xmm1 = xmm1[0],xmm0[0]
 ; SSE-NEXT:    movapd %xmm1, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; VEX-LABEL: uitofp_2i64_to_2f64:
 ; VEX:       # %bb.0:
 ; VEX-NEXT:    vmovapd {{.*#+}} xmm1 = [1127219200,1160773632,0,0]
 ; VEX-NEXT:    vunpcklps {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; VEX-NEXT:    vmovapd {{.*#+}} xmm3 = [4.503600e+15,1.934281e+25]
+; VEX-NEXT:    vmovapd {{.*#+}} xmm3 = [4503599627370496,1.9342813113834067E+25]
 ; VEX-NEXT:    vsubpd %xmm3, %xmm2, %xmm2
 ; VEX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[2,3,0,1]
 ; VEX-NEXT:    vunpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; VEX-NEXT:    vsubpd %xmm3, %xmm0, %xmm0
 ; VEX-NEXT:    vhaddpd %xmm0, %xmm2, %xmm0
 ; VEX-NEXT:    retq
 ;
 ; AVX512F-LABEL: uitofp_2i64_to_2f64:

 
 define <4 x double> @uitofp_4i64_to_4f64(<4 x i64> %a) {
 ; SSE-LABEL: uitofp_4i64_to_4f64:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movdqa %xmm0, %xmm2
 ; SSE-NEXT:    movdqa {{.*#+}} xmm3 = [1127219200,1160773632,0,0]
 ; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm0[2,3,0,1]
 ; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
-; SSE-NEXT:    movapd {{.*#+}} xmm5 = [4.503600e+15,1.934281e+25]
+; SSE-NEXT:    movapd {{.*#+}} xmm5 = [4503599627370496,1.9342813113834067E+25]
 ; SSE-NEXT:    subpd %xmm5, %xmm2
 ; SSE-NEXT:    movapd %xmm2, %xmm0
 ; SSE-NEXT:    unpckhpd {{.*#+}} xmm0 = xmm0[1],xmm2[1]
 ; SSE-NEXT:    addpd %xmm2, %xmm0
 ; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
 ; SSE-NEXT:    subpd %xmm5, %xmm4
 ; SSE-NEXT:    movapd %xmm4, %xmm2
 ; SSE-NEXT:    unpckhpd {{.*#+}} xmm2 = xmm2[1],xmm4[1]

 ; SSE-NEXT:    movapd %xmm2, %xmm1
 ; SSE-NEXT:    retq
 ;
 ; VEX-LABEL: uitofp_4i64_to_4f64:
 ; VEX:       # %bb.0:
 ; VEX-NEXT:    vextractf128 $1, %ymm0, %xmm1
 ; VEX-NEXT:    vmovapd {{.*#+}} xmm2 = [1127219200,1160773632,0,0]
 ; VEX-NEXT:    vunpcklps {{.*#+}} xmm3 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; VEX-NEXT:    vmovapd {{.*#+}} xmm4 = [4.503600e+15,1.934281e+25]
+; VEX-NEXT:    vmovapd {{.*#+}} xmm4 = [4503599627370496,1.9342813113834067E+25]
 ; VEX-NEXT:    vsubpd %xmm4, %xmm3, %xmm3
 ; VEX-NEXT:    vpermilps {{.*#+}} xmm1 = xmm1[2,3,0,1]
 ; VEX-NEXT:    vunpcklps {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
 ; VEX-NEXT:    vsubpd %xmm4, %xmm1, %xmm1
 ; VEX-NEXT:    vhaddpd %xmm1, %xmm3, %xmm1
 ; VEX-NEXT:    vunpcklps {{.*#+}} xmm3 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
 ; VEX-NEXT:    vsubpd %xmm4, %xmm3, %xmm3
 ; VEX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[2,3,0,1]

 }
 
 define <4 x double> @uitofp_4i32_to_4f64(<4 x i32> %a) {
 ; SSE-LABEL: uitofp_4i32_to_4f64:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movdqa %xmm0, %xmm1
 ; SSE-NEXT:    psrld $16, %xmm1
 ; SSE-NEXT:    cvtdq2pd %xmm1, %xmm1
-; SSE-NEXT:    movapd {{.*#+}} xmm2 = [6.553600e+04,6.553600e+04]
+; SSE-NEXT:    movapd {{.*#+}} xmm2 = [65536,65536]
 ; SSE-NEXT:    mulpd %xmm2, %xmm1
 ; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
 ; SSE-NEXT:    pand {{.*}}(%rip), %xmm0
 ; SSE-NEXT:    cvtdq2pd %xmm0, %xmm0
 ; SSE-NEXT:    addpd %xmm1, %xmm0
 ; SSE-NEXT:    movdqa %xmm3, %xmm1
 ; SSE-NEXT:    psrld $16, %xmm1
 ; SSE-NEXT:    cvtdq2pd %xmm1, %xmm4

 ; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [65535,65535,65535,65535]
 ; SSE-NEXT:    movdqa %xmm0, %xmm3
 ; SSE-NEXT:    pand %xmm2, %xmm3
 ; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [1258291200,1258291200,1258291200,1258291200]
 ; SSE-NEXT:    por %xmm4, %xmm3
 ; SSE-NEXT:    psrld $16, %xmm0
 ; SSE-NEXT:    movdqa {{.*#+}} xmm5 = [1392508928,1392508928,1392508928,1392508928]
 ; SSE-NEXT:    por %xmm5, %xmm0
-; SSE-NEXT:    movaps {{.*#+}} xmm6 = [-5.497642e+11,-5.497642e+11,-5.497642e+11,-5.497642e+11]
+; SSE-NEXT:    movaps {{.*#+}} xmm6 = [-5.49764202E+11,-5.49764202E+11,-5.49764202E+11,-5.49764202E+11]
 ; SSE-NEXT:    addps %xmm6, %xmm0
 ; SSE-NEXT:    addps %xmm3, %xmm0
 ; SSE-NEXT:    pand %xmm1, %xmm2
 ; SSE-NEXT:    por %xmm4, %xmm2
 ; SSE-NEXT:    psrld $16, %xmm1
 ; SSE-NEXT:    por %xmm5, %xmm1
 ; SSE-NEXT:    addps %xmm6, %xmm1
 ; SSE-NEXT:    addps %xmm2, %xmm1

 
 define <2 x double> @uitofp_load_2i64_to_2f64(<2 x i64> *%a) {
 ; SSE-LABEL: uitofp_load_2i64_to_2f64:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movdqa (%rdi), %xmm1
 ; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [1127219200,1160773632,0,0]
 ; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[2,3,0,1]
 ; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSE-NEXT:    movapd {{.*#+}} xmm4 = [4.503600e+15,1.934281e+25]
+; SSE-NEXT:    movapd {{.*#+}} xmm4 = [4503599627370496,1.9342813113834067E+25]
 ; SSE-NEXT:    subpd %xmm4, %xmm1
 ; SSE-NEXT:    movapd %xmm1, %xmm0
 ; SSE-NEXT:    unpckhpd {{.*#+}} xmm0 = xmm0[1],xmm1[1]
 ; SSE-NEXT:    addpd %xmm1, %xmm0
 ; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
 ; SSE-NEXT:    subpd %xmm4, %xmm3
 ; SSE-NEXT:    movapd %xmm3, %xmm1
 ; SSE-NEXT:    unpckhpd {{.*#+}} xmm1 = xmm1[1],xmm3[1]
 ; SSE-NEXT:    addpd %xmm3, %xmm1
 ; SSE-NEXT:    unpcklpd {{.*#+}} xmm0 = xmm0[0],xmm1[0]
 ; SSE-NEXT:    retq
 ;
 ; VEX-LABEL: uitofp_load_2i64_to_2f64:
 ; VEX:       # %bb.0:
 ; VEX-NEXT:    vmovapd (%rdi), %xmm0
 ; VEX-NEXT:    vmovapd {{.*#+}} xmm1 = [1127219200,1160773632,0,0]
 ; VEX-NEXT:    vunpcklps {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; VEX-NEXT:    vmovapd {{.*#+}} xmm3 = [4.503600e+15,1.934281e+25]
+; VEX-NEXT:    vmovapd {{.*#+}} xmm3 = [4503599627370496,1.9342813113834067E+25]
 ; VEX-NEXT:    vsubpd %xmm3, %xmm2, %xmm2
 ; VEX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[2,3,0,1]
 ; VEX-NEXT:    vunpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
 ; VEX-NEXT:    vsubpd %xmm3, %xmm0, %xmm0
 ; VEX-NEXT:    vhaddpd %xmm0, %xmm2, %xmm0
 ; VEX-NEXT:    retq
 ;
 ; AVX512F-LABEL: uitofp_load_2i64_to_2f64:

 define <4 x double> @uitofp_load_4i64_to_4f64(<4 x i64> *%a) {
 ; SSE-LABEL: uitofp_load_4i64_to_4f64:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movdqa (%rdi), %xmm1
 ; SSE-NEXT:    movdqa 16(%rdi), %xmm2
 ; SSE-NEXT:    movdqa {{.*#+}} xmm3 = [1127219200,1160773632,0,0]
 ; SSE-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[2,3,0,1]
 ; SSE-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
-; SSE-NEXT:    movapd {{.*#+}} xmm5 = [4.503600e+15,1.934281e+25]
+; SSE-NEXT:    movapd {{.*#+}} xmm5 = [4503599627370496,1.9342813113834067E+25]
 ; SSE-NEXT:    subpd %xmm5, %xmm1
 ; SSE-NEXT:    movapd %xmm1, %xmm0
 ; SSE-NEXT:    unpckhpd {{.*#+}} xmm0 = xmm0[1],xmm1[1]
 ; SSE-NEXT:    addpd %xmm1, %xmm0
 ; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
 ; SSE-NEXT:    subpd %xmm5, %xmm4
 ; SSE-NEXT:    movapd %xmm4, %xmm1
 ; SSE-NEXT:    unpckhpd {{.*#+}} xmm1 = xmm1[1],xmm4[1]

 ; SSE-NEXT:    retq
 ;
 ; VEX-LABEL: uitofp_load_4i64_to_4f64:
 ; VEX:       # %bb.0:
 ; VEX-NEXT:    vmovapd (%rdi), %ymm0
 ; VEX-NEXT:    vextractf128 $1, %ymm0, %xmm1
 ; VEX-NEXT:    vmovapd {{.*#+}} xmm2 = [1127219200,1160773632,0,0]
 ; VEX-NEXT:    vunpcklps {{.*#+}} xmm3 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; VEX-NEXT:    vmovapd {{.*#+}} xmm4 = [4.503600e+15,1.934281e+25]
+; VEX-NEXT:    vmovapd {{.*#+}} xmm4 = [4503599627370496,1.9342813113834067E+25]
 ; VEX-NEXT:    vsubpd %xmm4, %xmm3, %xmm3
 ; VEX-NEXT:    vpermilps {{.*#+}} xmm1 = xmm1[2,3,0,1]
 ; VEX-NEXT:    vunpcklps {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
 ; VEX-NEXT:    vsubpd %xmm4, %xmm1, %xmm1
 ; VEX-NEXT:    vhaddpd %xmm1, %xmm3, %xmm1
 ; VEX-NEXT:    vunpcklps {{.*#+}} xmm3 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
 ; VEX-NEXT:    vsubpd %xmm4, %xmm3, %xmm3
 ; VEX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[2,3,0,1]

 
 define <4 x double> @uitofp_load_4i32_to_4f64(<4 x i32> *%a) {
 ; SSE-LABEL: uitofp_load_4i32_to_4f64:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movdqa (%rdi), %xmm0
 ; SSE-NEXT:    movdqa %xmm0, %xmm1
 ; SSE-NEXT:    psrld $16, %xmm1
 ; SSE-NEXT:    cvtdq2pd %xmm1, %xmm1
-; SSE-NEXT:    movapd {{.*#+}} xmm2 = [6.553600e+04,6.553600e+04]
+; SSE-NEXT:    movapd {{.*#+}} xmm2 = [65536,65536]
 ; SSE-NEXT:    mulpd %xmm2, %xmm1
 ; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
 ; SSE-NEXT:    pand {{.*}}(%rip), %xmm0
 ; SSE-NEXT:    cvtdq2pd %xmm0, %xmm0
 ; SSE-NEXT:    addpd %xmm1, %xmm0
 ; SSE-NEXT:    movdqa %xmm3, %xmm1
 ; SSE-NEXT:    psrld $16, %xmm1
 ; SSE-NEXT:    cvtdq2pd %xmm1, %xmm4

 ; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [65535,65535,65535,65535]
 ; SSE-NEXT:    movdqa %xmm0, %xmm3
 ; SSE-NEXT:    pand %xmm2, %xmm3
 ; SSE-NEXT:    movdqa {{.*#+}} xmm4 = [1258291200,1258291200,1258291200,1258291200]
 ; SSE-NEXT:    por %xmm4, %xmm3
 ; SSE-NEXT:    psrld $16, %xmm0
 ; SSE-NEXT:    movdqa {{.*#+}} xmm5 = [1392508928,1392508928,1392508928,1392508928]
 ; SSE-NEXT:    por %xmm5, %xmm0
-; SSE-NEXT:    movaps {{.*#+}} xmm6 = [-5.497642e+11,-5.497642e+11,-5.497642e+11,-5.497642e+11]
+; SSE-NEXT:    movaps {{.*#+}} xmm6 = [-5.49764202E+11,-5.49764202E+11,-5.49764202E+11,-5.49764202E+11]
 ; SSE-NEXT:    addps %xmm6, %xmm0
 ; SSE-NEXT:    addps %xmm3, %xmm0
 ; SSE-NEXT:    pand %xmm1, %xmm2
 ; SSE-NEXT:    por %xmm4, %xmm2
 ; SSE-NEXT:    psrld $16, %xmm1
 ; SSE-NEXT:    por %xmm5, %xmm1
 ; SSE-NEXT:    addps %xmm6, %xmm1
 ; SSE-NEXT:    addps %xmm2, %xmm1

llvm/trunk/test/CodeGen/X86/vec_ss_load_fold.ll

; X64_AVX-NEXT:    retq
   %X = call <4 x float> @llvm.x86.sse41.round.ss(<4 x float> %q, <4 x float> %B, i32 4)
   ret <4 x float> %X
 }
 
 ; PR13576
 define  <2 x double> @test5() nounwind uwtable readnone noinline {
 ; X32-LABEL: test5:
 ; X32:       ## %bb.0: ## %entry
-; X32-NEXT:    movaps {{.*#+}} xmm0 = [1.280000e+02,1.233210e+02]
+; X32-NEXT:    movaps {{.*#+}} xmm0 = [128,123.321]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: test5:
 ; X64:       ## %bb.0: ## %entry
-; X64-NEXT:    movaps {{.*#+}} xmm0 = [1.280000e+02,1.233210e+02]
+; X64-NEXT:    movaps {{.*#+}} xmm0 = [128,123.321]
 ; X64-NEXT:    retq
 ;
 ; X32_AVX-LABEL: test5:
 ; X32_AVX:       ## %bb.0: ## %entry
-; X32_AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [1.280000e+02,1.233210e+02]
+; X32_AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [128,123.321]
 ; X32_AVX-NEXT:    retl
 ;
 ; X64_AVX-LABEL: test5:
 ; X64_AVX:       ## %bb.0: ## %entry
-; X64_AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [1.280000e+02,1.233210e+02]
+; X64_AVX-NEXT:    vmovaps {{.*#+}} xmm0 = [128,123.321]
 ; X64_AVX-NEXT:    retq
 entry:
   %0 = tail call <2 x double> @llvm.x86.sse2.cvtsi2sd(<2 x double> <double 4.569870e+02, double 1.233210e+02>, i32 128) nounwind readnone
   ret <2 x double> %0
 }
 
 declare <2 x double> @llvm.x86.sse2.cvtsi2sd(<2 x double>, i32) nounwind readnone
 

llvm/trunk/test/CodeGen/X86/vec_uint_to_fp-fastmath.ll

 ; AVX2-NEXT: .long 1199570944 # float 65536
 
 define <8 x float> @test_uitofp_v8i32_to_v8f32(<8 x i32> %arg) {
 ; SSE2-LABEL: test_uitofp_v8i32_to_v8f32:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    movdqa %xmm0, %xmm2
 ; SSE2-NEXT:    psrld $16, %xmm2
 ; SSE2-NEXT:    cvtdq2ps %xmm2, %xmm2
-; SSE2-NEXT:    movaps {{.*#+}} xmm3 = [6.553600e+04,6.553600e+04,6.553600e+04,6.553600e+04]
+; SSE2-NEXT:    movaps {{.*#+}} xmm3 = [65536,65536,65536,65536]
 ; SSE2-NEXT:    mulps %xmm3, %xmm2
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [65535,65535,65535,65535]
 ; SSE2-NEXT:    pand %xmm4, %xmm0
 ; SSE2-NEXT:    cvtdq2ps %xmm0, %xmm0
 ; SSE2-NEXT:    addps %xmm2, %xmm0
 ; SSE2-NEXT:    movdqa %xmm1, %xmm2
 ; SSE2-NEXT:    psrld $16, %xmm2
 ; SSE2-NEXT:    cvtdq2ps %xmm2, %xmm2
 ; SSE2-NEXT:    mulps %xmm3, %xmm2
 ; SSE2-NEXT:    pand %xmm4, %xmm1
 ; SSE2-NEXT:    cvtdq2ps %xmm1, %xmm1
 ; SSE2-NEXT:    addps %xmm2, %xmm1
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: test_uitofp_v8i32_to_v8f32:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    movdqa %xmm0, %xmm2
 ; SSE41-NEXT:    psrld $16, %xmm2
 ; SSE41-NEXT:    cvtdq2ps %xmm2, %xmm2
-; SSE41-NEXT:    movaps {{.*#+}} xmm3 = [6.553600e+04,6.553600e+04,6.553600e+04,6.553600e+04]
+; SSE41-NEXT:    movaps {{.*#+}} xmm3 = [65536,65536,65536,65536]
 ; SSE41-NEXT:    mulps %xmm3, %xmm2
 ; SSE41-NEXT:    pxor %xmm4, %xmm4
 ; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0],xmm4[1],xmm0[2],xmm4[3],xmm0[4],xmm4[5],xmm0[6],xmm4[7]
 ; SSE41-NEXT:    cvtdq2ps %xmm0, %xmm0
 ; SSE41-NEXT:    addps %xmm2, %xmm0
 ; SSE41-NEXT:    movdqa %xmm1, %xmm2
 ; SSE41-NEXT:    psrld $16, %xmm2
 ; SSE41-NEXT:    cvtdq2ps %xmm2, %xmm2

llvm/trunk/test/CodeGen/X86/vec_uint_to_fp.ll

 ; SSE: movdqa {{.*#+}} [[MASK:xmm[0-9]+]] = [65535,65535,65535,65535]
 ; SSE-NEXT: movdqa %xmm0, [[VECLOW:%xmm[0-9]+]]
 ; SSE-NEXT: pand %[[MASK]], [[VECLOW]]
 ; SSE-NEXT: movdqa {{.*#+}} [[LOWCST:xmm[0-9]+]] = [1258291200,1258291200,1258291200,1258291200]
 ; SSE-NEXT: por %[[LOWCST]], [[VECLOW]]
 ; SSE-NEXT: psrld $16, %xmm0
 ; SSE-NEXT: movdqa {{.*#+}} [[HIGHCST:xmm[0-9]+]] = [1392508928,1392508928,1392508928,1392508928]
 ; SSE-NEXT: por %[[HIGHCST]], %xmm0
-; SSE-NEXT: movaps {{.*#+}} [[MAGICCST:xmm[0-9]+]] = [-5.497642e+11,-5.497642e+11,-5.497642e+11,-5.497642e+11]
+; SSE-NEXT: movaps {{.*#+}} [[MAGICCST:xmm[0-9]+]] = [-5.49764202E+11,-5.49764202E+11,-5.49764202E+11,-5.49764202E+11]
 ; SSE-NEXT: addps %[[MAGICCST]], %xmm0
 ; SSE-NEXT: addps [[VECLOW]], %xmm0
 ; MASK is the low vector of the second part after this point.
 ; SSE-NEXT: pand %xmm1, %[[MASK]]
 ; SSE-NEXT: por %[[LOWCST]], %[[MASK]]
 ; SSE-NEXT: psrld $16, %xmm1
 ; SSE-NEXT: por %[[HIGHCST]], %xmm1
 ; SSE-NEXT: addps %[[MAGICCST]], %xmm1
 ; SSE-NEXT: addps %[[MASK]], %xmm1
 ; SSE-NEXT: retq
 ;
 ; SSE41: movdqa {{.*#+}} [[LOWCST:xmm[0-9]+]] = [1258291200,1258291200,1258291200,1258291200]
 ; SSE41-NEXT: movdqa %xmm0, [[VECLOW:%xmm[0-9]+]]
 ; SSE41-NEXT: pblendw $170, %[[LOWCST]], [[VECLOW]]
 ; SSE41-NEXT: psrld $16, %xmm0
 ; SSE41-NEXT: movdqa {{.*#+}} [[HIGHCST:xmm[0-9]+]] = [1392508928,1392508928,1392508928,1392508928]
 ; SSE41-NEXT: pblendw $170, %[[HIGHCST]], %xmm0
-; SSE41-NEXT: movaps {{.*#+}} [[MAGICCST:xmm[0-9]+]] = [-5.497642e+11,-5.497642e+11,-5.497642e+11,-5.497642e+11]
+; SSE41-NEXT: movaps {{.*#+}} [[MAGICCST:xmm[0-9]+]] = [-5.49764202E+11,-5.49764202E+11,-5.49764202E+11,-5.49764202E+11]
 ; SSE41-NEXT: addps %[[MAGICCST]], %xmm0
 ; SSE41-NEXT: addps [[VECLOW]], %xmm0
 ; LOWCST is the low vector of the second part after this point.
 ; The operands of the blend are inverted because we reuse xmm1
 ; in the next shift.
 ; SSE41-NEXT: pblendw $85, %xmm1, %[[LOWCST]]
 ; SSE41-NEXT: psrld $16, %xmm1
 ; SSE41-NEXT: pblendw $170, %[[HIGHCST]], %xmm1

llvm/trunk/test/CodeGen/X86/vector-constrained-fp-intrinsics.ll

%div = call <1 x float> @llvm.experimental.constrained.fdiv.v1f32(
            metadata !"round.dynamic",
            metadata !"fpexcept.strict")
   ret <1 x float> %div
 }
 
 define <2 x double> @constrained_vector_fdiv_v2f64() {
 ; NO-FMA-LABEL: constrained_vector_fdiv_v2f64:
 ; NO-FMA:       # %bb.0: # %entry
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1.000000e+00,2.000000e+00]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1,2]
 ; NO-FMA-NEXT:    divpd {{.*}}(%rip), %xmm0
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fdiv_v2f64:
 ; HAS-FMA:       # %bb.0: # %entry
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm0 = [1.000000e+00,2.000000e+00]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm0 = [1,2]
 ; HAS-FMA-NEXT:    vdivpd {{.*}}(%rip), %xmm0, %xmm0
 ; HAS-FMA-NEXT:    retq
 entry:
   %div = call <2 x double> @llvm.experimental.constrained.fdiv.v2f64(
            <2 x double> <double 1.000000e+00, double 2.000000e+00>,
            <2 x double> <double 1.000000e+01, double 1.000000e+01>,
            metadata !"round.dynamic",
            metadata !"fpexcept.strict")
%div = call <3 x float> @llvm.experimental.constrained.fdiv.v3f32(
            metadata !"round.dynamic",
            metadata !"fpexcept.strict")
   ret <3 x float> %div
 }
 
 define <3 x double> @constrained_vector_fdiv_v3f64() {
 ; NO-FMA-LABEL: constrained_vector_fdiv_v3f64:
 ; NO-FMA:       # %bb.0: # %entry
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1.000000e+00,2.000000e+00]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1,2]
 ; NO-FMA-NEXT:    divpd {{.*}}(%rip), %xmm0
 ; NO-FMA-NEXT:    movsd {{.*#+}} xmm1 = mem[0],zero
 ; NO-FMA-NEXT:    divsd {{.*}}(%rip), %xmm1
 ; NO-FMA-NEXT:    movsd %xmm1, -{{[0-9]+}}(%rsp)
 ; NO-FMA-NEXT:    movapd %xmm0, %xmm1
 ; NO-FMA-NEXT:    unpckhpd {{.*#+}} xmm1 = xmm1[1],xmm0[1]
 ; NO-FMA-NEXT:    fldl -{{[0-9]+}}(%rsp)
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fdiv_v3f64:
 ; HAS-FMA:       # %bb.0: # %entry
 ; HAS-FMA-NEXT:    vmovsd {{.*#+}} xmm0 = mem[0],zero
 ; HAS-FMA-NEXT:    vdivsd {{.*}}(%rip), %xmm0, %xmm0
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm1 = [1.000000e+00,2.000000e+00]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm1 = [1,2]
 ; HAS-FMA-NEXT:    vdivpd {{.*}}(%rip), %xmm1, %xmm1
 ; HAS-FMA-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; HAS-FMA-NEXT:    retq
 entry:
   %div = call <3 x double> @llvm.experimental.constrained.fdiv.v3f64(
            <3 x double> <double 1.000000e+00, double 2.000000e+00, double 3.000000e+00>,
            <3 x double> <double 1.000000e+01, double 1.000000e+01, double 1.000000e+01>,
            metadata !"round.dynamic",
            metadata !"fpexcept.strict")
   ret <3 x double> %div
 }
 
 define <4 x double> @constrained_vector_fdiv_v4f64() {
 ; NO-FMA-LABEL: constrained_vector_fdiv_v4f64:
 ; NO-FMA:       # %bb.0: # %entry
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm2 = [1.000000e+01,1.000000e+01]
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1.000000e+00,2.000000e+00]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm2 = [10,10]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1,2]
 ; NO-FMA-NEXT:    divpd %xmm2, %xmm0
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm1 = [3.000000e+00,4.000000e+00]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm1 = [3,4]
 ; NO-FMA-NEXT:    divpd %xmm2, %xmm1
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fdiv_v4f64:
 ; HAS-FMA:       # %bb.0: # %entry
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} ymm0 = [1.000000e+00,2.000000e+00,3.000000e+00,4.000000e+00]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} ymm0 = [1,2,3,4]
 ; HAS-FMA-NEXT:    vdivpd {{.*}}(%rip), %ymm0, %ymm0
 ; HAS-FMA-NEXT:    retq
 entry:
   %div = call <4 x double> @llvm.experimental.constrained.fdiv.v4f64(
            <4 x double> <double 1.000000e+00, double 2.000000e+00,
                          double 3.000000e+00, double 4.000000e+00>,
            <4 x double> <double 1.000000e+01, double 1.000000e+01,
                          double 1.000000e+01, double 1.000000e+01>,
%mul = call <1 x float> @llvm.experimental.constrained.fmul.v1f32(
            metadata !"round.dynamic",
            metadata !"fpexcept.strict")
   ret <1 x float> %mul
 }
 
 define <2 x double> @constrained_vector_fmul_v2f64() {
 ; NO-FMA-LABEL: constrained_vector_fmul_v2f64:
 ; NO-FMA:       # %bb.0: # %entry
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1.797693e+308,1.797693e+308]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1.7976931348623157E+308,1.7976931348623157E+308]
 ; NO-FMA-NEXT:    mulpd {{.*}}(%rip), %xmm0
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fmul_v2f64:
 ; HAS-FMA:       # %bb.0: # %entry
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm0 = [1.797693e+308,1.797693e+308]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm0 = [1.7976931348623157E+308,1.7976931348623157E+308]
 ; HAS-FMA-NEXT:    vmulpd {{.*}}(%rip), %xmm0, %xmm0
 ; HAS-FMA-NEXT:    retq
 entry:
   %mul = call <2 x double> @llvm.experimental.constrained.fmul.v2f64(
            <2 x double> <double 0x7FEFFFFFFFFFFFFF, double 0x7FEFFFFFFFFFFFFF>,
            <2 x double> <double 2.000000e+00, double 3.000000e+00>,
            metadata !"round.dynamic",
            metadata !"fpexcept.strict")
%mul = call <3 x float> @llvm.experimental.constrained.fmul.v3f32(
            metadata !"round.dynamic",
            metadata !"fpexcept.strict")
   ret <3 x float> %mul
 }
 
 define <3 x double> @constrained_vector_fmul_v3f64() {
 ; NO-FMA-LABEL: constrained_vector_fmul_v3f64:
 ; NO-FMA:       # %bb.0: # %entry
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1.797693e+308,1.797693e+308]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1.7976931348623157E+308,1.7976931348623157E+308]
 ; NO-FMA-NEXT:    mulpd {{.*}}(%rip), %xmm0
 ; NO-FMA-NEXT:    movsd {{.*#+}} xmm1 = mem[0],zero
 ; NO-FMA-NEXT:    mulsd {{.*}}(%rip), %xmm1
 ; NO-FMA-NEXT:    movsd %xmm1, -{{[0-9]+}}(%rsp)
 ; NO-FMA-NEXT:    movapd %xmm0, %xmm1
 ; NO-FMA-NEXT:    unpckhpd {{.*#+}} xmm1 = xmm1[1],xmm0[1]
 ; NO-FMA-NEXT:    fldl -{{[0-9]+}}(%rsp)
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fmul_v3f64:
 ; HAS-FMA:       # %bb.0: # %entry
 ; HAS-FMA-NEXT:    vmovsd {{.*#+}} xmm0 = mem[0],zero
 ; HAS-FMA-NEXT:    vmulsd {{.*}}(%rip), %xmm0, %xmm0
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm1 = [1.797693e+308,1.797693e+308]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm1 = [1.7976931348623157E+308,1.7976931348623157E+308]
 ; HAS-FMA-NEXT:    vmulpd {{.*}}(%rip), %xmm1, %xmm1
 ; HAS-FMA-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; HAS-FMA-NEXT:    retq
 entry:
   %mul = call <3 x double> @llvm.experimental.constrained.fmul.v3f64(
            <3 x double> <double 0x7FEFFFFFFFFFFFFF, double 0x7FEFFFFFFFFFFFFF,
                          double 0x7FEFFFFFFFFFFFFF>,
            <3 x double> <double 1.000000e+00, double 1.000000e+01, double 1.000000e+02>,
            metadata !"round.dynamic",
            metadata !"fpexcept.strict")
   ret <3 x double> %mul
 }
 
 define <4 x double> @constrained_vector_fmul_v4f64() {
 ; NO-FMA-LABEL: constrained_vector_fmul_v4f64:
 ; NO-FMA:       # %bb.0: # %entry
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm1 = [1.797693e+308,1.797693e+308]
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [2.000000e+00,3.000000e+00]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm1 = [1.7976931348623157E+308,1.7976931348623157E+308]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [2,3]
 ; NO-FMA-NEXT:    mulpd %xmm1, %xmm0
 ; NO-FMA-NEXT:    mulpd {{.*}}(%rip), %xmm1
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fmul_v4f64:
 ; HAS-FMA:       # %bb.0: # %entry
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} ymm0 = [1.797693e+308,1.797693e+308,1.797693e+308,1.797693e+308]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} ymm0 = [1.7976931348623157E+308,1.7976931348623157E+308,1.7976931348623157E+308,1.7976931348623157E+308]
 ; HAS-FMA-NEXT:    vmulpd {{.*}}(%rip), %ymm0, %ymm0
 ; HAS-FMA-NEXT:    retq
 entry:
   %mul = call <4 x double> @llvm.experimental.constrained.fmul.v4f64(
            <4 x double> <double 0x7FEFFFFFFFFFFFFF, double 0x7FEFFFFFFFFFFFFF,
                          double 0x7FEFFFFFFFFFFFFF, double 0x7FEFFFFFFFFFFFFF>,
            <4 x double> <double 2.000000e+00, double 3.000000e+00,
                          double 4.000000e+00, double 5.000000e+00>,
%add = call <1 x float> @llvm.experimental.constrained.fadd.v1f32(
            metadata !"round.dynamic",
            metadata !"fpexcept.strict")
   ret <1 x float> %add
 }
 
 define <2 x double> @constrained_vector_fadd_v2f64() {
 ; NO-FMA-LABEL: constrained_vector_fadd_v2f64:
 ; NO-FMA:       # %bb.0: # %entry
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1.797693e+308,1.797693e+308]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1.7976931348623157E+308,1.7976931348623157E+308]
 ; NO-FMA-NEXT:    addpd {{.*}}(%rip), %xmm0
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fadd_v2f64:
 ; HAS-FMA:       # %bb.0: # %entry
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm0 = [1.797693e+308,1.797693e+308]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm0 = [1.7976931348623157E+308,1.7976931348623157E+308]
 ; HAS-FMA-NEXT:    vaddpd {{.*}}(%rip), %xmm0, %xmm0
 ; HAS-FMA-NEXT:    retq
 entry:
   %add = call <2 x double> @llvm.experimental.constrained.fadd.v2f64(
            <2 x double> <double 0x7FEFFFFFFFFFFFFF, double 0x7FEFFFFFFFFFFFFF>,
            <2 x double> <double 1.000000e+00, double 1.000000e-01>,
            metadata !"round.dynamic",
            metadata !"fpexcept.strict")
%add = call <3 x float> @llvm.experimental.constrained.fadd.v3f32(
            metadata !"round.dynamic",
            metadata !"fpexcept.strict")
   ret <3 x float> %add
 }
 
 define <3 x double> @constrained_vector_fadd_v3f64() {
 ; NO-FMA-LABEL: constrained_vector_fadd_v3f64:
 ; NO-FMA:       # %bb.0: # %entry
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1.797693e+308,1.797693e+308]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1.7976931348623157E+308,1.7976931348623157E+308]
 ; NO-FMA-NEXT:    addpd {{.*}}(%rip), %xmm0
 ; NO-FMA-NEXT:    xorpd %xmm1, %xmm1
 ; NO-FMA-NEXT:    addsd {{.*}}(%rip), %xmm1
 ; NO-FMA-NEXT:    movsd %xmm1, -{{[0-9]+}}(%rsp)
 ; NO-FMA-NEXT:    movapd %xmm0, %xmm1
 ; NO-FMA-NEXT:    unpckhpd {{.*#+}} xmm1 = xmm1[1],xmm0[1]
 ; NO-FMA-NEXT:    fldl -{{[0-9]+}}(%rsp)
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fadd_v3f64:
 ; HAS-FMA:       # %bb.0: # %entry
 ; HAS-FMA-NEXT:    vxorpd %xmm0, %xmm0, %xmm0
 ; HAS-FMA-NEXT:    vaddsd {{.*}}(%rip), %xmm0, %xmm0
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm1 = [1.797693e+308,1.797693e+308]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm1 = [1.7976931348623157E+308,1.7976931348623157E+308]
 ; HAS-FMA-NEXT:    vaddpd {{.*}}(%rip), %xmm1, %xmm1
 ; HAS-FMA-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; HAS-FMA-NEXT:    retq
 entry:
   %add = call <3 x double> @llvm.experimental.constrained.fadd.v3f64(
            <3 x double> <double 0x7FEFFFFFFFFFFFFF, double 0x7FEFFFFFFFFFFFFF,
                          double 0x7FEFFFFFFFFFFFFF>,
            <3 x double> <double 2.0, double 1.0, double 0.0>,
            metadata !"round.dynamic",
            metadata !"fpexcept.strict")
   ret <3 x double> %add
 }
 
 define <4 x double> @constrained_vector_fadd_v4f64() {
 ; NO-FMA-LABEL: constrained_vector_fadd_v4f64:
 ; NO-FMA:       # %bb.0: # %entry
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm1 = [1.797693e+308,1.797693e+308]
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1.000000e+00,1.000000e-01]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm1 = [1.7976931348623157E+308,1.7976931348623157E+308]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [1,0.10000000000000001]
 ; NO-FMA-NEXT:    addpd %xmm1, %xmm0
 ; NO-FMA-NEXT:    addpd {{.*}}(%rip), %xmm1
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fadd_v4f64:
 ; HAS-FMA:       # %bb.0: # %entry
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} ymm0 = [1.797693e+308,1.797693e+308,1.797693e+308,1.797693e+308]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} ymm0 = [1.7976931348623157E+308,1.7976931348623157E+308,1.7976931348623157E+308,1.7976931348623157E+308]
 ; HAS-FMA-NEXT:    vaddpd {{.*}}(%rip), %ymm0, %ymm0
 ; HAS-FMA-NEXT:    retq
 entry:
   %add = call <4 x double> @llvm.experimental.constrained.fadd.v4f64(
            <4 x double> <double 0x7FEFFFFFFFFFFFFF, double 0x7FEFFFFFFFFFFFFF,
                          double 0x7FEFFFFFFFFFFFFF, double 0x7FEFFFFFFFFFFFFF>,
            <4 x double> <double 1.000000e+00, double 1.000000e-01,
                          double 2.000000e+00, double 2.000000e-01>,
%sub = call <1 x float> @llvm.experimental.constrained.fsub.v1f32(
            metadata !"round.dynamic",
            metadata !"fpexcept.strict")
   ret <1 x float> %sub
 }
 
 define <2 x double> @constrained_vector_fsub_v2f64() {
 ; NO-FMA-LABEL: constrained_vector_fsub_v2f64:
 ; NO-FMA:       # %bb.0: # %entry
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [-1.797693e+308,-1.797693e+308]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [-1.7976931348623157E+308,-1.7976931348623157E+308]
 ; NO-FMA-NEXT:    subpd {{.*}}(%rip), %xmm0
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fsub_v2f64:
 ; HAS-FMA:       # %bb.0: # %entry
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm0 = [-1.797693e+308,-1.797693e+308]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm0 = [-1.7976931348623157E+308,-1.7976931348623157E+308]
 ; HAS-FMA-NEXT:    vsubpd {{.*}}(%rip), %xmm0, %xmm0
 ; HAS-FMA-NEXT:    retq
 entry:
   %sub = call <2 x double> @llvm.experimental.constrained.fsub.v2f64(
            <2 x double> <double 0xFFEFFFFFFFFFFFFF, double 0xFFEFFFFFFFFFFFFF>,
            <2 x double> <double 1.000000e+00, double 1.000000e-01>,
            metadata !"round.dynamic",
            metadata !"fpexcept.strict")

 }
 
 define <3 x double> @constrained_vector_fsub_v3f64() {
 ; NO-FMA-LABEL: constrained_vector_fsub_v3f64:
 ; NO-FMA:       # %bb.0: # %entry
 ; NO-FMA-NEXT:    xorpd %xmm0, %xmm0
 ; NO-FMA-NEXT:    movsd {{.*#+}} xmm1 = mem[0],zero
 ; NO-FMA-NEXT:    subsd %xmm0, %xmm1
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [-1.797693e+308,-1.797693e+308]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm0 = [-1.7976931348623157E+308,-1.7976931348623157E+308]
 ; NO-FMA-NEXT:    subpd {{.*}}(%rip), %xmm0
 ; NO-FMA-NEXT:    movsd %xmm1, -{{[0-9]+}}(%rsp)
 ; NO-FMA-NEXT:    movapd %xmm0, %xmm1
 ; NO-FMA-NEXT:    unpckhpd {{.*#+}} xmm1 = xmm1[1],xmm0[1]
 ; NO-FMA-NEXT:    fldl -{{[0-9]+}}(%rsp)
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fsub_v3f64:
 ; HAS-FMA:       # %bb.0: # %entry
 ; HAS-FMA-NEXT:    vxorpd %xmm0, %xmm0, %xmm0
 ; HAS-FMA-NEXT:    vmovsd {{.*#+}} xmm1 = mem[0],zero
 ; HAS-FMA-NEXT:    vsubsd %xmm0, %xmm1, %xmm0
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm1 = [-1.797693e+308,-1.797693e+308]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm1 = [-1.7976931348623157E+308,-1.7976931348623157E+308]
 ; HAS-FMA-NEXT:    vsubpd {{.*}}(%rip), %xmm1, %xmm1
 ; HAS-FMA-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; HAS-FMA-NEXT:    retq
 entry:
   %sub = call <3 x double> @llvm.experimental.constrained.fsub.v3f64(
            <3 x double> <double 0xFFEFFFFFFFFFFFFF, double 0xFFEFFFFFFFFFFFFF,
                          double 0xFFEFFFFFFFFFFFFF>,
            <3 x double> <double 2.0, double 1.0, double 0.0>,
            metadata !"round.dynamic",
            metadata !"fpexcept.strict")
   ret <3 x double> %sub
 }
 
 define <4 x double> @constrained_vector_fsub_v4f64() {
 ; NO-FMA-LABEL: constrained_vector_fsub_v4f64:
 ; NO-FMA:       # %bb.0: # %entry
-; NO-FMA-NEXT:    movapd {{.*#+}} xmm1 = [-1.797693e+308,-1.797693e+308]
+; NO-FMA-NEXT:    movapd {{.*#+}} xmm1 = [-1.7976931348623157E+308,-1.7976931348623157E+308]
 ; NO-FMA-NEXT:    movapd %xmm1, %xmm0
 ; NO-FMA-NEXT:    subpd {{.*}}(%rip), %xmm0
 ; NO-FMA-NEXT:    subpd {{.*}}(%rip), %xmm1
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fsub_v4f64:
 ; HAS-FMA:       # %bb.0: # %entry
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} ymm0 = [-1.797693e+308,-1.797693e+308,-1.797693e+308,-1.797693e+308]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} ymm0 = [-1.7976931348623157E+308,-1.7976931348623157E+308,-1.7976931348623157E+308,-1.7976931348623157E+308]
 ; HAS-FMA-NEXT:    vsubpd {{.*}}(%rip), %ymm0, %ymm0
 ; HAS-FMA-NEXT:    retq
 entry:
   %sub = call <4 x double> @llvm.experimental.constrained.fsub.v4f64(
            <4 x double> <double 0xFFEFFFFFFFFFFFFF, double 0xFFEFFFFFFFFFFFFF,
                          double 0xFFEFFFFFFFFFFFFF, double 0xFFEFFFFFFFFFFFFF>,
            <4 x double> <double 1.000000e+00, double 1.000000e-01,
                          double 2.000000e+00, double 2.000000e-01>,

 ; NO-FMA-NEXT:    unpcklpd (%rsp), %xmm0 # 16-byte Folded Reload
 ; NO-FMA-NEXT:    # xmm0 = xmm0[0],mem[0]
 ; NO-FMA-NEXT:    addq $24, %rsp
 ; NO-FMA-NEXT:    .cfi_def_cfa_offset 8
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fma_v2f64:
 ; HAS-FMA:       # %bb.0: # %entry
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm1 = [1.500000e+00,5.000000e-01]
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm0 = [3.500000e+00,2.500000e+00]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm1 = [1.5,0.5]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm0 = [3.5,2.5]
 ; HAS-FMA-NEXT:    vfmadd213pd {{.*#+}} xmm0 = (xmm1 * xmm0) + mem
 ; HAS-FMA-NEXT:    retq
 entry:
   %fma = call <2 x double> @llvm.experimental.constrained.fma.v2f64(
            <2 x double> <double 1.5, double 0.5>,
            <2 x double> <double 3.5, double 2.5>,
            <2 x double> <double 5.5, double 4.5>,
            metadata !"round.dynamic",

 ; NO-FMA-NEXT:    .cfi_def_cfa_offset 8
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fma_v3f64:
 ; HAS-FMA:       # %bb.0: # %entry
 ; HAS-FMA-NEXT:    vmovsd {{.*#+}} xmm0 = mem[0],zero
 ; HAS-FMA-NEXT:    vmovsd {{.*#+}} xmm1 = mem[0],zero
 ; HAS-FMA-NEXT:    vfmadd213sd {{.*#+}} xmm1 = (xmm0 * xmm1) + mem
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm0 = [2.500000e+00,1.500000e+00]
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm2 = [5.500000e+00,4.500000e+00]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm0 = [2.5,1.5]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} xmm2 = [5.5,4.5]
 ; HAS-FMA-NEXT:    vfmadd213pd {{.*#+}} xmm2 = (xmm0 * xmm2) + mem
 ; HAS-FMA-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm0
 ; HAS-FMA-NEXT:    retq
 entry:
   %fma = call <3 x double> @llvm.experimental.constrained.fma.v3f64(
            <3 x double> <double 2.5, double 1.5, double 0.5>,
            <3 x double> <double 5.5, double 4.5, double 3.5>,
            <3 x double> <double 8.5, double 7.5, double 6.5>,

 ; NO-FMA-NEXT:    # xmm1 = xmm1[0],mem[0]
 ; NO-FMA-NEXT:    movaps (%rsp), %xmm0 # 16-byte Reload
 ; NO-FMA-NEXT:    addq $40, %rsp
 ; NO-FMA-NEXT:    .cfi_def_cfa_offset 8
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fma_v4f64:
 ; HAS-FMA:       # %bb.0: # %entry
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} ymm1 = [3.500000e+00,2.500000e+00,1.500000e+00,5.000000e-01]
-; HAS-FMA-NEXT:    vmovapd {{.*#+}} ymm0 = [7.500000e+00,6.500000e+00,5.500000e+00,4.500000e+00]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} ymm1 = [3.5,2.5,1.5,0.5]
+; HAS-FMA-NEXT:    vmovapd {{.*#+}} ymm0 = [7.5,6.5,5.5,4.5]
 ; HAS-FMA-NEXT:    vfmadd213pd {{.*#+}} ymm0 = (ymm1 * ymm0) + mem
 ; HAS-FMA-NEXT:    retq
 entry:
   %fma = call <4 x double> @llvm.experimental.constrained.fma.v4f64(
            <4 x double> <double 3.5, double 2.5, double 1.5, double 0.5>,
            <4 x double> <double 7.5, double 6.5, double 5.5, double 4.5>,
            <4 x double> <double 11.5, double 10.5, double 9.5, double 8.5>,
            metadata !"round.dynamic",

 ; NO-FMA-NEXT:    unpcklpd (%rsp), %xmm0 # 16-byte Folded Reload
 ; NO-FMA-NEXT:    # xmm0 = xmm0[0],mem[0]
 ; NO-FMA-NEXT:    addq $40, %rsp
 ; NO-FMA-NEXT:    .cfi_def_cfa_offset 8
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fma_v4f32:
 ; HAS-FMA:       # %bb.0: # %entry
-; HAS-FMA-NEXT:    vmovaps {{.*#+}} xmm1 = [3.500000e+00,2.500000e+00,1.500000e+00,5.000000e-01]
-; HAS-FMA-NEXT:    vmovaps {{.*#+}} xmm0 = [7.500000e+00,6.500000e+00,5.500000e+00,4.500000e+00]
+; HAS-FMA-NEXT:    vmovaps {{.*#+}} xmm1 = [3.5,2.5,1.5,0.5]
+; HAS-FMA-NEXT:    vmovaps {{.*#+}} xmm0 = [7.5,6.5,5.5,4.5]
 ; HAS-FMA-NEXT:    vfmadd213ps {{.*#+}} xmm0 = (xmm1 * xmm0) + mem
 ; HAS-FMA-NEXT:    retq
 entry:
   %fma = call <4 x float> @llvm.experimental.constrained.fma.v4f32(
            <4 x float> <float 3.5, float 2.5, float 1.5, float 0.5>,
            <4 x float> <float 7.5, float 6.5, float 5.5, float 4.5>,
            <4 x float> <float 11.5, float 10.5, float 9.5, float 8.5>,
            metadata !"round.dynamic",

 ; NO-FMA-NEXT:    # xmm1 = xmm1[0],mem[0]
 ; NO-FMA-NEXT:    movaps {{[-0-9]+}}(%r{{[sb]}}p), %xmm0 # 16-byte Reload
 ; NO-FMA-NEXT:    addq $56, %rsp
 ; NO-FMA-NEXT:    .cfi_def_cfa_offset 8
 ; NO-FMA-NEXT:    retq
 ;
 ; HAS-FMA-LABEL: constrained_vector_fma_v8f32:
 ; HAS-FMA:       # %bb.0: # %entry
-; HAS-FMA-NEXT:    vmovaps {{.*#+}} ymm1 = [3.500000e+00,2.500000e+00,1.500000e+00,5.000000e-01,7.500000e+00,6.500000e+00,5.500000e+00,4.500000e+00]
-; HAS-FMA-NEXT:    vmovaps {{.*#+}} ymm0 = [7.500000e+00,6.500000e+00,5.500000e+00,4.500000e+00,1.150000e+01,1.050000e+01,9.500000e+00,8.500000e+00]
+; HAS-FMA-NEXT:    vmovaps {{.*#+}} ymm1 = [3.5,2.5,1.5,0.5,7.5,6.5,5.5,4.5]
+; HAS-FMA-NEXT:    vmovaps {{.*#+}} ymm0 = [7.5,6.5,5.5,4.5,11.5,10.5,9.5,8.5]
 ; HAS-FMA-NEXT:    vfmadd213ps {{.*#+}} ymm0 = (ymm1 * ymm0) + mem
 ; HAS-FMA-NEXT:    retq
 entry:
   %fma = call <8 x float> @llvm.experimental.constrained.fma.v8f32(
            <8 x float> <float 3.5, float 2.5, float 1.5, float 0.5,
                         float 7.5, float 6.5, float 5.5, float 4.5>,
            <8 x float> <float 7.5, float 6.5, float 5.5, float 4.5,
                         float 11.5, float 10.5, float 9.5, float 8.5>,

llvm/trunk/test/CodeGen/X86/vector-shuffle-avx512.ll

; KNL32-NEXT:    retl
    %res = shufflevector <4 x float> %addV, <4 x float> %a, <8 x i32> <i32 3, i32 3, i32 4, i32 0, i32 5, i32 0, i32 0, i32 0>
    ret <8 x float> %res
 }
 
 ;Negative test.
 define <8 x float> @expand15(<4 x float> %a) {
 ; SKX64-LABEL: expand15:
 ; SKX64:       # %bb.0:
-; SKX64-NEXT:    vmovaps {{.*#+}} xmm1 = [0.000000e+00,2.000000e+00,0.000000e+00,0.000000e+00]
+; SKX64-NEXT:    vmovaps {{.*#+}} xmm1 = [0,2,0,0]
 ; SKX64-NEXT:    vpermilps {{.*#+}} xmm2 = xmm0[0,1,1,3]
 ; SKX64-NEXT:    vmovaps {{.*#+}} ymm0 = [0,1,8,3,10,3,2,3]
 ; SKX64-NEXT:    vpermi2ps %ymm2, %ymm1, %ymm0
 ; SKX64-NEXT:    retq
 ;
 ; KNL64-LABEL: expand15:
 ; KNL64:       # %bb.0:
-; KNL64-NEXT:    vmovaps {{.*#+}} xmm1 = [0.000000e+00,2.000000e+00,0.000000e+00,0.000000e+00]
+; KNL64-NEXT:    vmovaps {{.*#+}} xmm1 = [0,2,0,0]
 ; KNL64-NEXT:    vpermpd {{.*#+}} ymm1 = ymm1[0,1,1,1]
 ; KNL64-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,1,1,3]
 ; KNL64-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,0,1,3]
 ; KNL64-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2],ymm1[3],ymm0[4],ymm1[5,6,7]
 ; KNL64-NEXT:    retq
 ;
 ; SKX32-LABEL: expand15:
 ; SKX32:       # %bb.0:
-; SKX32-NEXT:    vmovaps {{.*#+}} xmm1 = [0.000000e+00,2.000000e+00,0.000000e+00,0.000000e+00]
+; SKX32-NEXT:    vmovaps {{.*#+}} xmm1 = [0,2,0,0]
 ; SKX32-NEXT:    vpermilps {{.*#+}} xmm2 = xmm0[0,1,1,3]
 ; SKX32-NEXT:    vmovaps {{.*#+}} ymm0 = [0,1,8,3,10,3,2,3]
 ; SKX32-NEXT:    vpermi2ps %ymm2, %ymm1, %ymm0
 ; SKX32-NEXT:    retl
 ;
 ; KNL32-LABEL: expand15:
 ; KNL32:       # %bb.0:
-; KNL32-NEXT:    vmovaps {{.*#+}} xmm1 = [0.000000e+00,2.000000e+00,0.000000e+00,0.000000e+00]
+; KNL32-NEXT:    vmovaps {{.*#+}} xmm1 = [0,2,0,0]
 ; KNL32-NEXT:    vpermpd {{.*#+}} ymm1 = ymm1[0,1,1,1]
 ; KNL32-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,1,1,3]
 ; KNL32-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,0,1,3]
 ; KNL32-NEXT:    vblendps {{.*#+}} ymm0 = ymm1[0,1],ymm0[2],ymm1[3],ymm0[4],ymm1[5,6,7]
 ; KNL32-NEXT:    retl
    %addV = fadd <4 x float> <float 0.0,float 1.0,float 2.0,float 0.0> , <float 0.0,float 1.0,float 2.0,float 0.0>
    %res = shufflevector <4 x float> %addV, <4 x float> %a, <8 x i32> <i32 0, i32 1, i32 4, i32 0, i32 5, i32 0, i32 0, i32 0>
    ret <8 x float> %res

llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-avx.ll

; X64-NEXT:    retq
   %1 = call <4 x float> @llvm.x86.avx.vpermilvar.ps(<4 x float> %a0, <4 x i32> <i32 3, i32 2, i32 1, i32 0>)
   %2 = shufflevector <4 x float> %1, <4 x float> zeroinitializer, <4 x i32> <i32 2, i32 4, i32 1, i32 4>
   ret <4 x float> %2
 }
 
 define <2 x double> @constant_fold_vpermilvar_pd() {
 ; X32-LABEL: constant_fold_vpermilvar_pd:
 ; X32:       # %bb.0:
-; X32-NEXT:    vmovaps {{.*#+}} xmm0 = [2.000000e+00,1.000000e+00]
+; X32-NEXT:    vmovaps {{.*#+}} xmm0 = [2,1]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: constant_fold_vpermilvar_pd:
 ; X64:       # %bb.0:
-; X64-NEXT:    vmovaps {{.*#+}} xmm0 = [2.000000e+00,1.000000e+00]
+; X64-NEXT:    vmovaps {{.*#+}} xmm0 = [2,1]
 ; X64-NEXT:    retq
   %1 = call <2 x double> @llvm.x86.avx.vpermilvar.pd(<2 x double> <double 1.0, double 2.0>, <2 x i64> <i64 2, i64 0>)
   ret <2 x double> %1
 }
 
 define <4 x double> @constant_fold_vpermilvar_pd_256() {
 ; X32-LABEL: constant_fold_vpermilvar_pd_256:
 ; X32:       # %bb.0:
-; X32-NEXT:    vmovaps {{.*#+}} ymm0 = [2.000000e+00,1.000000e+00,3.000000e+00,4.000000e+00]
+; X32-NEXT:    vmovaps {{.*#+}} ymm0 = [2,1,3,4]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: constant_fold_vpermilvar_pd_256:
 ; X64:       # %bb.0:
-; X64-NEXT:    vmovaps {{.*#+}} ymm0 = [2.000000e+00,1.000000e+00,3.000000e+00,4.000000e+00]
+; X64-NEXT:    vmovaps {{.*#+}} ymm0 = [2,1,3,4]
 ; X64-NEXT:    retq
   %1 = call <4 x double> @llvm.x86.avx.vpermilvar.pd.256(<4 x double> <double 1.0, double 2.0, double 3.0, double 4.0>, <4 x i64> <i64 2, i64 0, i64 0, i64 2>)
   ret <4 x double> %1
 }
 
 define <4 x float> @constant_fold_vpermilvar_ps() {
 ; X32-LABEL: constant_fold_vpermilvar_ps:
 ; X32:       # %bb.0:
-; X32-NEXT:    vmovaps {{.*#+}} xmm0 = [4.000000e+00,1.000000e+00,3.000000e+00,2.000000e+00]
+; X32-NEXT:    vmovaps {{.*#+}} xmm0 = [4,1,3,2]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: constant_fold_vpermilvar_ps:
 ; X64:       # %bb.0:
-; X64-NEXT:    vmovaps {{.*#+}} xmm0 = [4.000000e+00,1.000000e+00,3.000000e+00,2.000000e+00]
+; X64-NEXT:    vmovaps {{.*#+}} xmm0 = [4,1,3,2]
 ; X64-NEXT:    retq
   %1 = call <4 x float> @llvm.x86.avx.vpermilvar.ps(<4 x float> <float 1.0, float 2.0, float 3.0, float 4.0>, <4 x i32> <i32 3, i32 0, i32 2, i32 1>)
   ret <4 x float> %1
 }
 
 define <8 x float> @constant_fold_vpermilvar_ps_256() {
 ; X32-LABEL: constant_fold_vpermilvar_ps_256:
 ; X32:       # %bb.0:
-; X32-NEXT:    vmovaps {{.*#+}} ymm0 = [1.000000e+00,1.000000e+00,3.000000e+00,2.000000e+00,5.000000e+00,6.000000e+00,6.000000e+00,6.000000e+00]
+; X32-NEXT:    vmovaps {{.*#+}} ymm0 = [1,1,3,2,5,6,6,6]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: constant_fold_vpermilvar_ps_256:
 ; X64:       # %bb.0:
-; X64-NEXT:    vmovaps {{.*#+}} ymm0 = [1.000000e+00,1.000000e+00,3.000000e+00,2.000000e+00,5.000000e+00,6.000000e+00,6.000000e+00,6.000000e+00]
+; X64-NEXT:    vmovaps {{.*#+}} ymm0 = [1,1,3,2,5,6,6,6]
 ; X64-NEXT:    retq
   %1 = call <8 x float> @llvm.x86.avx.vpermilvar.ps.256(<8 x float> <float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0, float 8.0>, <8 x i32> <i32 4, i32 0, i32 2, i32 1, i32 0, i32 1, i32 1, i32 1>)
   ret <8 x float> %1
 }

llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-avx2.ll

 ; CHECK-NEXT:    ret{{[l|q]}}
   %1 = call <8 x i32> @llvm.x86.avx2.permd(<8 x i32> <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>, <8 x i32> <i32 4, i32 6, i32 2, i32 1, i32 7, i32 1, i32 5, i32 0>)
   ret <8 x i32> %1
 }
 
 define <8 x float> @constant_fold_permps() {
 ; CHECK-LABEL: constant_fold_permps:
 ; CHECK:       # %bb.0:
-; CHECK-NEXT:    vmovaps {{.*#+}} ymm0 = [5.000000e+00,7.000000e+00,3.000000e+00,2.000000e+00,8.000000e+00,2.000000e+00,6.000000e+00,1.000000e+00]
+; CHECK-NEXT:    vmovaps {{.*#+}} ymm0 = [5,7,3,2,8,2,6,1]
 ; CHECK-NEXT:    ret{{[l|q]}}
   %1 = call <8 x float> @llvm.x86.avx2.permps(<8 x float> <float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0, float 8.0>, <8 x i32> <i32 4, i32 6, i32 2, i32 1, i32 7, i32 1, i32 5, i32 0>)
   ret <8 x float> %1
 }
 
 define <32 x i8> @constant_fold_pshufb_256() {
 ; CHECK-LABEL: constant_fold_pshufb_256:
 ; CHECK:       # %bb.0:

llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-xop.ll

; X64-NEXT:    retq
   %v3 = insertelement <4 x float> %v2,   float %b2, i32 3
   store <4 x float> %v3, <4 x float>* %ptr
   ret void
 }
 
 define <2 x double> @constant_fold_vpermil2pd() {
 ; X32-LABEL: constant_fold_vpermil2pd:
 ; X32:       # %bb.0:
-; X32-NEXT:    vmovaps {{.*#+}} xmm0 = [-2.000000e+00,2.000000e+00]
+; X32-NEXT:    vmovaps {{.*#+}} xmm0 = [-2,2]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: constant_fold_vpermil2pd:
 ; X64:       # %bb.0:
-; X64-NEXT:    vmovaps {{.*#+}} xmm0 = [-2.000000e+00,2.000000e+00]
+; X64-NEXT:    vmovaps {{.*#+}} xmm0 = [-2,2]
 ; X64-NEXT:    retq
   %1 = call <2 x double> @llvm.x86.xop.vpermil2pd(<2 x double> <double 1.0, double 2.0>, <2 x double> <double -2.0, double -1.0>, <2 x i64> <i64 4, i64 2>, i8 2)
   ret <2 x double> %1
 }
 
 define <4 x double> @constant_fold_vpermil2pd_256() {
 ; X32-LABEL: constant_fold_vpermil2pd_256:
 ; X32:       # %bb.0:
-; X32-NEXT:    vmovaps {{.*#+}} ymm0 = [-4.000000e+00,0.000000e+00,4.000000e+00,3.000000e+00]
+; X32-NEXT:    vmovaps {{.*#+}} ymm0 = [-4,0,4,3]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: constant_fold_vpermil2pd_256:
 ; X64:       # %bb.0:
-; X64-NEXT:    vmovaps {{.*#+}} ymm0 = [-4.000000e+00,0.000000e+00,4.000000e+00,3.000000e+00]
+; X64-NEXT:    vmovaps {{.*#+}} ymm0 = [-4,0,4,3]
 ; X64-NEXT:    retq
   %1 = call <4 x double> @llvm.x86.xop.vpermil2pd.256(<4 x double> <double 1.0, double 2.0, double 3.0, double 4.0>, <4 x double> <double -4.0, double -3.0, double -2.0, double -1.0>, <4 x i64> <i64 4, i64 8, i64 2, i64 0>, i8 2)
   ret <4 x double> %1
 }
 
 define <4 x float> @constant_fold_vpermil2ps() {
 ; X32-LABEL: constant_fold_vpermil2ps:
 ; X32:       # %bb.0:
-; X32-NEXT:    vmovaps {{.*#+}} xmm0 = [-4.000000e+00,1.000000e+00,3.000000e+00,0.000000e+00]
+; X32-NEXT:    vmovaps {{.*#+}} xmm0 = [-4,1,3,0]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: constant_fold_vpermil2ps:
 ; X64:       # %bb.0:
-; X64-NEXT:    vmovaps {{.*#+}} xmm0 = [-4.000000e+00,1.000000e+00,3.000000e+00,0.000000e+00]
+; X64-NEXT:    vmovaps {{.*#+}} xmm0 = [-4,1,3,0]
 ; X64-NEXT:    retq
   %1 = call <4 x float> @llvm.x86.xop.vpermil2ps(<4 x float> <float 1.0, float 2.0, float 3.0, float 4.0>, <4 x float> <float -4.0, float -3.0, float -2.0, float -1.0>, <4 x i32> <i32 4, i32 0, i32 2, i32 8>, i8 2)
   ret <4 x float> %1
 }
 
 define <8 x float> @constant_fold_vpermil2ps_256() {
 ; X32-LABEL: constant_fold_vpermil2ps_256:
 ; X32:       # %bb.0:
-; X32-NEXT:    vmovaps {{.*#+}} ymm0 = [-8.000000e+00,1.000000e+00,3.000000e+00,0.000000e+00,5.000000e+00,0.000000e+00,5.000000e+00,7.000000e+00]
+; X32-NEXT:    vmovaps {{.*#+}} ymm0 = [-8,1,3,0,5,0,5,7]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: constant_fold_vpermil2ps_256:
 ; X64:       # %bb.0:
-; X64-NEXT:    vmovaps {{.*#+}} ymm0 = [-8.000000e+00,1.000000e+00,3.000000e+00,0.000000e+00,5.000000e+00,0.000000e+00,5.000000e+00,7.000000e+00]
+; X64-NEXT:    vmovaps {{.*#+}} ymm0 = [-8,1,3,0,5,0,5,7]
 ; X64-NEXT:    retq
   %1 = call <8 x float> @llvm.x86.xop.vpermil2ps.256(<8 x float> <float 1.0, float 2.0, float 3.0, float 4.0, float 5.0, float 6.0, float 7.0, float 8.0>, <8 x float> <float -8.0, float -7.0, float -6.0, float -5.0, float -4.0, float -3.0, float -2.0, float -1.0>, <8 x i32> <i32 4, i32 0, i32 2, i32 8, i32 0, i32 8, i32 0, i32 2>, i8 2)
   ret <8 x float> %1
 }
 
 define <16 x i8> @constant_fold_vpperm() {
 ; X32-LABEL: constant_fold_vpperm:
 ; X32:       # %bb.0:

llvm/trunk/test/CodeGen/X86/vselect-avx.ll

 ; AVX1:       ## %bb.0: ## %bb
 ; AVX1-NEXT:    vpslld $31, %xmm0, %xmm0
 ; AVX1-NEXT:    vpsrad $31, %xmm0, %xmm0
 ; AVX1-NEXT:    vpmovsxdq %xmm0, %xmm1
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
 ; AVX1-NEXT:    vpmovsxdq %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    movq (%rdi,%rsi,8), %rax
-; AVX1-NEXT:    vmovapd {{.*#+}} ymm1 = [5.000000e-01,5.000000e-01,5.000000e-01,5.000000e-01]
+; AVX1-NEXT:    vmovapd {{.*#+}} ymm1 = [0.5,0.5,0.5,0.5]
 ; AVX1-NEXT:    vblendvpd %ymm0, {{.*}}(%rip), %ymm1, %ymm0
 ; AVX1-NEXT:    vmovupd %ymm0, (%rax)
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: test2:
 ; AVX2:       ## %bb.0: ## %bb
 ; AVX2-NEXT:    vpslld $31, %xmm0, %xmm0

llvm/trunk/test/CodeGen/X86/x86-setcc-int-to-fp-combine.ll

 define void @foo2(<4 x float>* noalias %result) nounwind {
 ; CHECK-LABEL: LCPI2_0:
 ; CHECK-NEXT: .long 1082130432              ## float 4
 ; CHECK-NEXT: .long 1084227584              ## float 5
 ; CHECK-NEXT: .long 1086324736              ## float 6
 ; CHECK-NEXT: .long 1088421888              ## float 7
 ; CHECK-LABEL: foo2:
 ; CHECK:       ## %bb.0:
-; CHECK-NEXT:    movaps {{.*#+}} xmm0 = [4.000000e+00,5.000000e+00,6.000000e+00,7.000000e+00]
+; CHECK-NEXT:    movaps {{.*#+}} xmm0 = [4,5,6,7]
 ; CHECK-NEXT:    movaps %xmm0, (%rdi)
 ; CHECK-NEXT:    retq
   %val = uitofp <4 x i32> <i32 4, i32 5, i32 6, i32 7> to <4 x float>
   store <4 x float> %val, <4 x float>* %result
   ret void
 }
 
 ; Fold explicit AND operations when the constant isn't a splat of a single

 define void @foo4(<4 x float>* noalias %result) nounwind {
 ; CHECK-LABEL: LCPI4_0:
 ; CHECK-NEXT: .long 1065353216              ## float 1
 ; CHECK-NEXT: .long 1123942400              ## float 127
 ; CHECK-NEXT: .long 1124073472              ## float 128
 ; CHECK-NEXT: .long 1132396544              ## float 255
 ; CHECK-LABEL: foo4:
 ; CHECK:       ## %bb.0:
-; CHECK-NEXT:    movaps {{.*#+}} xmm0 = [1.000000e+00,1.270000e+02,1.280000e+02,2.550000e+02]
+; CHECK-NEXT:    movaps {{.*#+}} xmm0 = [1,127,128,255]
 ; CHECK-NEXT:    movaps %xmm0, (%rdi)
 ; CHECK-NEXT:    retq
   %val = uitofp <4 x i8> <i8 1, i8 127, i8 -128, i8 -1> to <4 x float>
   store <4 x float> %val, <4 x float>* %result
   ret void
 }