[DAGCombiner][x86] add transform/hook to decompose integer multiply into shift/add

[spatel]

Summary

This is an alternative to D37896. I don't see a way to decompose multiplies generically without a target hook to tell us when it's profitable.

As a first step, I'm just trying to get the vector cases requested in PR34474:
https://bugs.llvm.org/show_bug.cgi?id=34474

The shakiest test diff here may be SSE4.1 code that uses 'pmulld' with a constant pool load. That can become 4 instructions like:

movdqa %xmm0, %xmm1
pslld $4, %xmm1
paddd %xmm0, %xmm1
movdqa %xmm1, %xmm0

...but I think despite the code-size increase, this is still better performing code. A scan of Agner's timing tables says pmulld is always at least 4 cycle latency, but possibly as much as 11 cycles. So replacing that with fast ops (and removing the constant load) should be a win even in the minimal case.

[RKSimon]

The shakiest test diff here may be SSE4.1 code that uses 'pmulld' with a constant pool load. That can become 4 instructions like:

pmullw could be worse - that's often just 2/3cy latency

[spatel]

Patch updated:
A more conservative first step for x86 - don't do the transform if the vector multiply is legal (pmullw/pmulld). The remaining cases should always be clear improvements in speed and size.

[RKSimon]

LGTM

llvm/trunk/include/llvm/CodeGen/TargetLowering.h

public:
 
   /// Return true if a select of constants (select Cond, C1, C2) should be
   /// transformed into simple math ops with the condition value. For example:
   /// select Cond, C1, C1-1 --> add (zext Cond), C1-1
   virtual bool convertSelectOfConstantsToMath(EVT VT) const {
     return false;
   }
 
+  /// Return true if it is profitable to transform an integer
+  /// multiplication-by-constant into simpler operations like shifts and adds.
+  /// This may be true if the target does not directly support the
+  /// multiplication operation for the specified type or the sequence of simpler
+  /// ops is faster than the multiply.
+  virtual bool decomposeMulByConstant(EVT VT, SDValue C) const {
+    return false;
+  }
+
   //===--------------------------------------------------------------------===//
   // TargetLowering Configuration Methods - These methods should be invoked by
   // the derived class constructor to configure this object for the target.
   //
 protected:
   /// Specify how the target extends the result of integer and floating point
   /// boolean values from i1 to a wider type.  See getBooleanContents.
   void setBooleanContents(BooleanContent Ty) {

llvm/trunk/lib/CodeGen/SelectionDAG/DAGCombiner.cpp

if (N1IsConst && !N1IsOpaqueConst && (-ConstValue1).isPowerOf2()) {
     // single-use add), we should put the negate there.
     return DAG.getNode(ISD::SUB, DL, VT,
                        DAG.getConstant(0, DL, VT),
                        DAG.getNode(ISD::SHL, DL, VT, N0,
                             DAG.getConstant(Log2Val, DL,
                                       getShiftAmountTy(N0.getValueType()))));
   }
 
+  // Try to transform multiply-by-(power-of-2 +/- 1) into shift and add/sub.
+  // Examples: x * 33 --> (x << 5) + x
+  //           x * 15 --> (x << 4) - x
+  if (N1IsConst && TLI.decomposeMulByConstant(VT, N1)) {
+    // TODO: Negative constants can be handled by negating the result.
+    // TODO: We could handle more general decomposition of any constant by
+    //       having the target set a limit on number of ops and making a
+    //       callback to determine that sequence (similar to sqrt expansion).
+    unsigned MathOp = ISD::DELETED_NODE;
+    if ((ConstValue1 - 1).isPowerOf2())
+      MathOp = ISD::ADD;
+    else if ((ConstValue1 + 1).isPowerOf2())
+      MathOp = ISD::SUB;
+
+    if (MathOp != ISD::DELETED_NODE) {
+      unsigned ShAmt = MathOp == ISD::ADD ? (ConstValue1 - 1).logBase2()
+                                          : (ConstValue1 + 1).logBase2();
+      assert(ShAmt > 0 && ShAmt < VT.getScalarSizeInBits() &&
+             "Not expecting multiply-by-constant that could have simplified");
+      SDLoc DL(N);
+      SDValue Shl = DAG.getNode(ISD::SHL, DL, VT, N0,
+                                DAG.getConstant(ShAmt, DL, VT));
+      return DAG.getNode(MathOp, DL, VT, Shl, N0);
+    }
+  }
+
   // (mul (shl X, c1), c2) -> (mul X, c2 << c1)
   if (N0.getOpcode() == ISD::SHL &&
       isConstantOrConstantVector(N1, /* NoOpaques */ true) &&
       isConstantOrConstantVector(N0.getOperand(1), /* NoOpaques */ true)) {
     SDValue C3 = DAG.getNode(ISD::SHL, SDLoc(N), VT, N1, N0.getOperand(1));
     if (isConstantOrConstantVector(C3))
       return DAG.getNode(ISD::MUL, SDLoc(N), VT, N0.getOperand(0), C3);
   }

llvm/trunk/lib/Target/X86/X86ISelLowering.h

public:
 
     /// Returns true if it is beneficial to convert a load of a constant
     /// to just the constant itself.
     bool shouldConvertConstantLoadToIntImm(const APInt &Imm,
                                            Type *Ty) const override;
 
     bool convertSelectOfConstantsToMath(EVT VT) const override;
 
+    bool decomposeMulByConstant(EVT VT, SDValue C) const override;
+
     /// Return true if EXTRACT_SUBVECTOR is cheap for this result type
     /// with this index.
     bool isExtractSubvectorCheap(EVT ResVT, EVT SrcVT,
                                  unsigned Index) const override;
 
     bool storeOfVectorConstantIsCheap(EVT MemVT, unsigned NumElem,
                                       unsigned AddrSpace) const override {
       // If we can replace more than 2 scalar stores, there will be a reduction

llvm/trunk/lib/Target/X86/X86ISelLowering.cpp

bool X86TargetLowering::convertSelectOfConstantsToMath(EVT VT) const {
   // TODO: It might be a win to ease or lift this restriction, but the generic
   // folds in DAGCombiner conflict with vector folds for an AVX512 target.
   if (VT.isVector() && Subtarget.hasAVX512())
     return false;
 
   return true;
 }
 
+bool X86TargetLowering::decomposeMulByConstant(EVT VT, SDValue C) const {
+  // TODO: We handle scalars using custom code, but generic combining could make
+  // that unnecessary.
+  APInt MulC;
+  if (!ISD::isConstantSplatVector(C.getNode(), MulC))
+    return false;
+
+  // If vector multiply is legal, assume that's faster than shl + add/sub.
+  // TODO: Multiply is a complex op with higher latency and lower througput in
+  //       most implementations, so this check could be loosened based on type
+  //       and/or a CPU attribute.
+  if (isOperationLegal(ISD::MUL, VT))
+    return false;
+
+  return (MulC + 1).isPowerOf2() || (MulC - 1).isPowerOf2();
+}
+
 bool X86TargetLowering::isExtractSubvectorCheap(EVT ResVT, EVT SrcVT,
                                                 unsigned Index) const {
   if (!isOperationLegalOrCustom(ISD::EXTRACT_SUBVECTOR, ResVT))
     return false;
 
   // Mask vectors support all subregister combinations and operations that
   // extract half of vector.
   if (ResVT.getVectorElementType() == MVT::i1)

llvm/trunk/test/CodeGen/X86/urem-seteq-vec-nonsplat.ll

 ; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm2
 ; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,3,2,3]
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
 ; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm3
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[1,3,2,3]
 ; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
 ; CHECK-SSE2-NEXT:    psrld $2, %xmm2
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [5,5,5,5]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm3
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[0,2,2,3]
-; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
-; CHECK-SSE2-NEXT:    psubd %xmm2, %xmm0
+; CHECK-SSE2-NEXT:    movdqa %xmm2, %xmm1
+; CHECK-SSE2-NEXT:    pslld $2, %xmm1
+; CHECK-SSE2-NEXT:    paddd %xmm2, %xmm1
+; CHECK-SSE2-NEXT:    psubd %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    pcmpeqd {{.*}}(%rip), %xmm0
 ; CHECK-SSE2-NEXT:    psrld $31, %xmm0
 ; CHECK-SSE2-NEXT:    retq
 ;
 ; CHECK-SSE41-LABEL: test_urem_comp_nonsplat:
 ; CHECK-SSE41:       # %bb.0:
 ; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,3,3]
 ; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [3435973837,3435973837,3435973837,3435973837]

 ; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm2
 ; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,3,2,3]
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
 ; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm3
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[1,3,2,3]
 ; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
 ; CHECK-SSE2-NEXT:    psrld $2, %xmm2
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [5,5,5,5]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm3
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[0,2,2,3]
-; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
-; CHECK-SSE2-NEXT:    psubd %xmm2, %xmm0
+; CHECK-SSE2-NEXT:    movdqa %xmm2, %xmm1
+; CHECK-SSE2-NEXT:    pslld $2, %xmm1
+; CHECK-SSE2-NEXT:    paddd %xmm2, %xmm1
+; CHECK-SSE2-NEXT:    psubd %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    pxor %xmm1, %xmm1
 ; CHECK-SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    psrld $31, %xmm0
 ; CHECK-SSE2-NEXT:    retq
 ;
 ; CHECK-SSE41-LABEL: test_urem_comp_undef:
 ; CHECK-SSE41:       # %bb.0:
 ; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,3,3]

llvm/trunk/test/CodeGen/X86/urem-seteq-vec-splat.ll

 ; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm2
 ; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,3,2,3]
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
 ; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm3
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[1,3,2,3]
 ; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
 ; CHECK-SSE2-NEXT:    psrld $2, %xmm2
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [5,5,5,5]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm3
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[0,2,2,3]
-; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
-; CHECK-SSE2-NEXT:    psubd %xmm2, %xmm0
+; CHECK-SSE2-NEXT:    movdqa %xmm2, %xmm1
+; CHECK-SSE2-NEXT:    pslld $2, %xmm1
+; CHECK-SSE2-NEXT:    paddd %xmm2, %xmm1
+; CHECK-SSE2-NEXT:    psubd %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    pxor %xmm1, %xmm1
 ; CHECK-SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    psrld $31, %xmm0
 ; CHECK-SSE2-NEXT:    retq
 ;
 ; CHECK-SSE41-LABEL: test_urem_odd_vec_i32:
 ; CHECK-SSE41:       # %bb.0:
 ; CHECK-SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,3,3]

 ; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm2
 ; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,3,2,3]
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
 ; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm3
 ; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[1,3,2,3]
 ; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
 ; CHECK-SSE2-NEXT:    psrld $2, %xmm2
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [5,5,5,5]
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm2
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
-; CHECK-SSE2-NEXT:    pmuludq %xmm1, %xmm3
-; CHECK-SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[0,2,2,3]
-; CHECK-SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
-; CHECK-SSE2-NEXT:    psubd %xmm2, %xmm0
+; CHECK-SSE2-NEXT:    movdqa %xmm2, %xmm1
+; CHECK-SSE2-NEXT:    pslld $2, %xmm1
+; CHECK-SSE2-NEXT:    paddd %xmm2, %xmm1
+; CHECK-SSE2-NEXT:    psubd %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    pxor %xmm1, %xmm1
 ; CHECK-SSE2-NEXT:    pcmpeqd %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    psrld $31, %xmm0
 ; CHECK-SSE2-NEXT:    retq
 ;
 ; CHECK-SSE41-LABEL: test_urem_odd_vec_i16:
 ; CHECK-SSE41:       # %bb.0:
 ; CHECK-SSE41-NEXT:    pxor %xmm1, %xmm1

llvm/trunk/test/CodeGen/X86/vector-idiv-sdiv-128.ll

 ; SSE2-NEXT:    pand %xmm1, %xmm3
 ; SSE2-NEXT:    paddd %xmm0, %xmm3
 ; SSE2-NEXT:    psubd %xmm3, %xmm2
 ; SSE2-NEXT:    paddd %xmm0, %xmm2
 ; SSE2-NEXT:    movdqa %xmm2, %xmm1
 ; SSE2-NEXT:    psrld $31, %xmm1
 ; SSE2-NEXT:    psrad $2, %xmm2
 ; SSE2-NEXT:    paddd %xmm1, %xmm2
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [7,7,7,7]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm1, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
-; SSE2-NEXT:    pmuludq %xmm1, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
-; SSE2-NEXT:    psubd %xmm2, %xmm0
+; SSE2-NEXT:    movdqa %xmm2, %xmm1
+; SSE2-NEXT:    pslld $3, %xmm1
+; SSE2-NEXT:    psubd %xmm1, %xmm2
+; SSE2-NEXT:    paddd %xmm2, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: test_rem7_4i32:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,3,3]
 ; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [2454267027,2454267027,2454267027,2454267027]
 ; SSE41-NEXT:    pmuldq %xmm2, %xmm1
 ; SSE41-NEXT:    pmuldq %xmm0, %xmm2

 ; SSE2-NEXT:    pand {{.*}}(%rip), %xmm2
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32]
 ; SSE2-NEXT:    pxor %xmm3, %xmm2
 ; SSE2-NEXT:    psubb %xmm3, %xmm2
 ; SSE2-NEXT:    psrlw $7, %xmm1
 ; SSE2-NEXT:    pand {{.*}}(%rip), %xmm1
 ; SSE2-NEXT:    paddb %xmm2, %xmm1
 ; SSE2-NEXT:    movdqa %xmm1, %xmm2
-; SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm0[8],xmm2[9],xmm0[9],xmm2[10],xmm0[10],xmm2[11],xmm0[11],xmm2[12],xmm0[12],xmm2[13],xmm0[13],xmm2[14],xmm0[14],xmm2[15],xmm0[15]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
-; SSE2-NEXT:    pmullw %xmm3, %xmm2
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
-; SSE2-NEXT:    pand %xmm4, %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE2-NEXT:    pmullw %xmm3, %xmm1
-; SSE2-NEXT:    pand %xmm4, %xmm1
-; SSE2-NEXT:    packuswb %xmm2, %xmm1
-; SSE2-NEXT:    psubb %xmm1, %xmm0
+; SSE2-NEXT:    psllw $3, %xmm2
+; SSE2-NEXT:    pand {{.*}}(%rip), %xmm2
+; SSE2-NEXT:    psubb %xmm2, %xmm1
+; SSE2-NEXT:    paddb %xmm0, %xmm1
+; SSE2-NEXT:    movdqa %xmm1, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: test_rem7_16i8:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pmovsxbw %xmm0, %xmm1
 ; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [65427,65427,65427,65427,65427,65427,65427,65427]
 ; SSE41-NEXT:    pmullw %xmm2, %xmm1
 ; SSE41-NEXT:    psrlw $8, %xmm1
 ; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
 ; SSE41-NEXT:    pmovsxbw %xmm3, %xmm3
 ; SSE41-NEXT:    pmullw %xmm2, %xmm3
 ; SSE41-NEXT:    psrlw $8, %xmm3
 ; SSE41-NEXT:    packuswb %xmm3, %xmm1
 ; SSE41-NEXT:    paddb %xmm0, %xmm1
 ; SSE41-NEXT:    movdqa %xmm1, %xmm2
 ; SSE41-NEXT:    psrlw $2, %xmm2
 ; SSE41-NEXT:    pand {{.*}}(%rip), %xmm2
 ; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32]
 ; SSE41-NEXT:    pxor %xmm3, %xmm2
 ; SSE41-NEXT:    psubb %xmm3, %xmm2
 ; SSE41-NEXT:    psrlw $7, %xmm1
 ; SSE41-NEXT:    pand {{.*}}(%rip), %xmm1
 ; SSE41-NEXT:    paddb %xmm2, %xmm1
-; SSE41-NEXT:    pmovzxbw {{.*#+}} xmm2 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
-; SSE41-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
-; SSE41-NEXT:    pmullw %xmm3, %xmm1
-; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
-; SSE41-NEXT:    pand %xmm4, %xmm1
-; SSE41-NEXT:    pmullw %xmm3, %xmm2
-; SSE41-NEXT:    pand %xmm4, %xmm2
-; SSE41-NEXT:    packuswb %xmm1, %xmm2
-; SSE41-NEXT:    psubb %xmm2, %xmm0
+; SSE41-NEXT:    movdqa %xmm1, %xmm2
+; SSE41-NEXT:    psllw $3, %xmm2
+; SSE41-NEXT:    pand {{.*}}(%rip), %xmm2
+; SSE41-NEXT:    psubb %xmm2, %xmm1
+; SSE41-NEXT:    paddb %xmm1, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: test_rem7_16i8:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vpmovsxbw %xmm0, %xmm1
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [65427,65427,65427,65427,65427,65427,65427,65427]
 ; AVX1-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
 ; AVX1-NEXT:    vpsrlw $8, %xmm1, %xmm1
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
 ; AVX1-NEXT:    vpmovsxbw %xmm3, %xmm3
 ; AVX1-NEXT:    vpmullw %xmm2, %xmm3, %xmm2
 ; AVX1-NEXT:    vpsrlw $8, %xmm2, %xmm2
 ; AVX1-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
 ; AVX1-NEXT:    vpaddb %xmm0, %xmm1, %xmm1
 ; AVX1-NEXT:    vpsrlw $2, %xmm1, %xmm2
 ; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32]
 ; AVX1-NEXT:    vpxor %xmm3, %xmm2, %xmm2
 ; AVX1-NEXT:    vpsubb %xmm3, %xmm2, %xmm2
 ; AVX1-NEXT:    vpsrlw $7, %xmm1, %xmm1
 ; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
 ; AVX1-NEXT:    vpaddb %xmm1, %xmm2, %xmm1
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
-; AVX1-NEXT:    vpmullw %xmm3, %xmm2, %xmm2
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpand %xmm4, %xmm2, %xmm2
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
-; AVX1-NEXT:    vpmullw %xmm3, %xmm1, %xmm1
-; AVX1-NEXT:    vpand %xmm4, %xmm1, %xmm1
-; AVX1-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpsubb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpsllw $3, %xmm1, %xmm2
+; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
+; AVX1-NEXT:    vpsubb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2NOBW-LABEL: test_rem7_16i8:
 ; AVX2NOBW:       # %bb.0:
 ; AVX2NOBW-NEXT:    vpmovsxbw %xmm0, %ymm1
 ; AVX2NOBW-NEXT:    vpmullw {{.*}}(%rip), %ymm1, %ymm1
 ; AVX2NOBW-NEXT:    vpsrlw $8, %ymm1, %ymm1
 ; AVX2NOBW-NEXT:    vextracti128 $1, %ymm1, %xmm2
 ; AVX2NOBW-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
 ; AVX2NOBW-NEXT:    vpaddb %xmm0, %xmm1, %xmm1
 ; AVX2NOBW-NEXT:    vpsrlw $2, %xmm1, %xmm2
 ; AVX2NOBW-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
 ; AVX2NOBW-NEXT:    vmovdqa {{.*#+}} xmm3 = [32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32]
 ; AVX2NOBW-NEXT:    vpxor %xmm3, %xmm2, %xmm2
 ; AVX2NOBW-NEXT:    vpsubb %xmm3, %xmm2, %xmm2
 ; AVX2NOBW-NEXT:    vpsrlw $7, %xmm1, %xmm1
 ; AVX2NOBW-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
 ; AVX2NOBW-NEXT:    vpaddb %xmm1, %xmm2, %xmm1
-; AVX2NOBW-NEXT:    vpmovsxbw %xmm1, %ymm1
-; AVX2NOBW-NEXT:    vpmullw {{.*}}(%rip), %ymm1, %ymm1
-; AVX2NOBW-NEXT:    vextracti128 $1, %ymm1, %xmm2
-; AVX2NOBW-NEXT:    vmovdqa {{.*#+}} xmm3 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2NOBW-NEXT:    vpshufb %xmm3, %xmm2, %xmm2
-; AVX2NOBW-NEXT:    vpshufb %xmm3, %xmm1, %xmm1
-; AVX2NOBW-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
-; AVX2NOBW-NEXT:    vpsubb %xmm1, %xmm0, %xmm0
+; AVX2NOBW-NEXT:    vpsllw $3, %xmm1, %xmm2
+; AVX2NOBW-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
+; AVX2NOBW-NEXT:    vpsubb %xmm2, %xmm1, %xmm1
+; AVX2NOBW-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX2NOBW-NEXT:    vzeroupper
 ; AVX2NOBW-NEXT:    retq
 ;
 ; AVX512BW-LABEL: test_rem7_16i8:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vpmovsxbw %xmm0, %ymm1
 ; AVX512BW-NEXT:    vpmullw {{.*}}(%rip), %ymm1, %ymm1
 ; AVX512BW-NEXT:    vpsrlw $8, %ymm1, %ymm1
 ; AVX512BW-NEXT:    vpmovwb %zmm1, %ymm1
 ; AVX512BW-NEXT:    vpaddb %xmm0, %xmm1, %xmm1
 ; AVX512BW-NEXT:    vpsrlw $2, %xmm1, %xmm2
 ; AVX512BW-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
 ; AVX512BW-NEXT:    vmovdqa {{.*#+}} xmm3 = [32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32]
 ; AVX512BW-NEXT:    vpxor %xmm3, %xmm2, %xmm2
 ; AVX512BW-NEXT:    vpsubb %xmm3, %xmm2, %xmm2
 ; AVX512BW-NEXT:    vpsrlw $7, %xmm1, %xmm1
 ; AVX512BW-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
 ; AVX512BW-NEXT:    vpaddb %xmm1, %xmm2, %xmm1
-; AVX512BW-NEXT:    vpmovsxbw %xmm1, %ymm1
-; AVX512BW-NEXT:    vpmullw {{.*}}(%rip), %ymm1, %ymm1
-; AVX512BW-NEXT:    vpmovwb %zmm1, %ymm1
-; AVX512BW-NEXT:    vpsubb %xmm1, %xmm0, %xmm0
+; AVX512BW-NEXT:    vpsllw $3, %xmm1, %xmm2
+; AVX512BW-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
+; AVX512BW-NEXT:    vpsubb %xmm2, %xmm1, %xmm1
+; AVX512BW-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vzeroupper
 ; AVX512BW-NEXT:    retq
   %res = srem <16 x i8> %a, <i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7>
   ret <16 x i8> %res
 }

llvm/trunk/test/CodeGen/X86/vector-idiv-sdiv-256.ll

 ; AVX1-NEXT:    vpmuldq %xmm3, %xmm2, %xmm2
 ; AVX1-NEXT:    vpmuldq %xmm3, %xmm1, %xmm4
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm4[1,1,3,3]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm4[0,1],xmm2[2,3],xmm4[4,5],xmm2[6,7]
 ; AVX1-NEXT:    vpaddd %xmm1, %xmm2, %xmm2
 ; AVX1-NEXT:    vpsrld $31, %xmm2, %xmm4
 ; AVX1-NEXT:    vpsrad $2, %xmm2, %xmm2
 ; AVX1-NEXT:    vpaddd %xmm4, %xmm2, %xmm2
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [7,7,7,7]
-; AVX1-NEXT:    vpmulld %xmm4, %xmm2, %xmm2
-; AVX1-NEXT:    vpsubd %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpslld $3, %xmm2, %xmm4
+; AVX1-NEXT:    vpsubd %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddd %xmm2, %xmm1, %xmm1
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
 ; AVX1-NEXT:    vpmuldq %xmm3, %xmm2, %xmm2
 ; AVX1-NEXT:    vpmuldq %xmm3, %xmm0, %xmm3
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm3[0,1],xmm2[2,3],xmm3[4,5],xmm2[6,7]
 ; AVX1-NEXT:    vpaddd %xmm0, %xmm2, %xmm2
 ; AVX1-NEXT:    vpsrld $31, %xmm2, %xmm3
 ; AVX1-NEXT:    vpsrad $2, %xmm2, %xmm2
 ; AVX1-NEXT:    vpaddd %xmm3, %xmm2, %xmm2
-; AVX1-NEXT:    vpmulld %xmm4, %xmm2, %xmm2
-; AVX1-NEXT:    vpsubd %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpslld $3, %xmm2, %xmm3
+; AVX1-NEXT:    vpsubd %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddd %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: test_rem7_8i32:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm0[1,1,3,3,5,5,7,7]
 ; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm2 = [2454267027,2454267027,2454267027,2454267027,2454267027,2454267027,2454267027,2454267027]
 ; AVX2-NEXT:    vpmuldq %ymm2, %ymm1, %ymm1

 ; AVX1-LABEL: test_rem7_16i16:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [18725,18725,18725,18725,18725,18725,18725,18725]
 ; AVX1-NEXT:    vpmulhw %xmm2, %xmm1, %xmm3
 ; AVX1-NEXT:    vpsrlw $15, %xmm3, %xmm4
 ; AVX1-NEXT:    vpsraw $1, %xmm3, %xmm3
 ; AVX1-NEXT:    vpaddw %xmm4, %xmm3, %xmm3
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [7,7,7,7,7,7,7,7]
-; AVX1-NEXT:    vpmullw %xmm4, %xmm3, %xmm3
-; AVX1-NEXT:    vpsubw %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpsllw $3, %xmm3, %xmm4
+; AVX1-NEXT:    vpsubw %xmm4, %xmm3, %xmm3
+; AVX1-NEXT:    vpaddw %xmm3, %xmm1, %xmm1
 ; AVX1-NEXT:    vpmulhw %xmm2, %xmm0, %xmm2
 ; AVX1-NEXT:    vpsrlw $15, %xmm2, %xmm3
 ; AVX1-NEXT:    vpsraw $1, %xmm2, %xmm2
 ; AVX1-NEXT:    vpaddw %xmm3, %xmm2, %xmm2
-; AVX1-NEXT:    vpmullw %xmm4, %xmm2, %xmm2
-; AVX1-NEXT:    vpsubw %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpsllw $3, %xmm2, %xmm3
+; AVX1-NEXT:    vpsubw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddw %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: test_rem7_16i16:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpmulhw {{.*}}(%rip), %ymm0, %ymm1
 ; AVX2-NEXT:    vpsrlw $15, %ymm1, %ymm2
 ; AVX2-NEXT:    vpsraw $1, %ymm1, %ymm1
 ; AVX2-NEXT:    vpaddw %ymm2, %ymm1, %ymm1
 ; AVX2-NEXT:    vpmullw {{.*}}(%rip), %ymm1, %ymm1
 ; AVX2-NEXT:    vpsubw %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
   %res = srem <16 x i16> %a, <i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7>
   ret <16 x i16> %res
 }
 
 define <32 x i8> @test_rem7_32i8(<32 x i8> %a) nounwind {
 ; AVX1-LABEL: test_rem7_32i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
-; AVX1-NEXT:    vpmovsxbw %xmm2, %xmm3
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [65427,65427,65427,65427,65427,65427,65427,65427]
-; AVX1-NEXT:    vpmullw %xmm1, %xmm3, %xmm3
-; AVX1-NEXT:    vpsrlw $8, %xmm3, %xmm3
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm2[2,3,0,1]
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpmovsxbw %xmm1, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [65427,65427,65427,65427,65427,65427,65427,65427]
+; AVX1-NEXT:    vpmullw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrlw $8, %xmm2, %xmm2
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm1[2,3,0,1]
 ; AVX1-NEXT:    vpmovsxbw %xmm4, %xmm4
-; AVX1-NEXT:    vpmullw %xmm1, %xmm4, %xmm4
+; AVX1-NEXT:    vpmullw %xmm3, %xmm4, %xmm4
 ; AVX1-NEXT:    vpsrlw $8, %xmm4, %xmm4
-; AVX1-NEXT:    vpackuswb %xmm4, %xmm3, %xmm3
-; AVX1-NEXT:    vpaddb %xmm2, %xmm3, %xmm3
-; AVX1-NEXT:    vpsrlw $7, %xmm3, %xmm4
+; AVX1-NEXT:    vpackuswb %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddb %xmm1, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrlw $7, %xmm2, %xmm4
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm8 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
 ; AVX1-NEXT:    vpand %xmm8, %xmm4, %xmm4
-; AVX1-NEXT:    vpsrlw $2, %xmm3, %xmm3
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm9 = [63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63]
-; AVX1-NEXT:    vpand %xmm9, %xmm3, %xmm3
+; AVX1-NEXT:    vpsrlw $2, %xmm2, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm6 = [63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63]
+; AVX1-NEXT:    vpand %xmm6, %xmm2, %xmm2
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm7 = [32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32]
-; AVX1-NEXT:    vpxor %xmm7, %xmm3, %xmm3
-; AVX1-NEXT:    vpsubb %xmm7, %xmm3, %xmm3
-; AVX1-NEXT:    vpaddb %xmm4, %xmm3, %xmm3
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm4 = xmm3[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm5 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
-; AVX1-NEXT:    vpmullw %xmm5, %xmm4, %xmm4
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm6 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpand %xmm6, %xmm4, %xmm4
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero,xmm3[4],zero,xmm3[5],zero,xmm3[6],zero,xmm3[7],zero
-; AVX1-NEXT:    vpmullw %xmm5, %xmm3, %xmm3
-; AVX1-NEXT:    vpand %xmm6, %xmm3, %xmm3
-; AVX1-NEXT:    vpackuswb %xmm4, %xmm3, %xmm3
-; AVX1-NEXT:    vpsubb %xmm3, %xmm2, %xmm2
-; AVX1-NEXT:    vpmovsxbw %xmm0, %xmm3
-; AVX1-NEXT:    vpmullw %xmm1, %xmm3, %xmm3
-; AVX1-NEXT:    vpsrlw $8, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpsubb %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddb %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpsllw $3, %xmm2, %xmm4
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm5 = [248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248]
+; AVX1-NEXT:    vpand %xmm5, %xmm4, %xmm4
+; AVX1-NEXT:    vpsubb %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpmovsxbw %xmm0, %xmm2
+; AVX1-NEXT:    vpmullw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrlw $8, %xmm2, %xmm2
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm0[2,3,0,1]
 ; AVX1-NEXT:    vpmovsxbw %xmm4, %xmm4
-; AVX1-NEXT:    vpmullw %xmm1, %xmm4, %xmm1
-; AVX1-NEXT:    vpsrlw $8, %xmm1, %xmm1
-; AVX1-NEXT:    vpackuswb %xmm1, %xmm3, %xmm1
-; AVX1-NEXT:    vpaddb %xmm0, %xmm1, %xmm1
-; AVX1-NEXT:    vpsrlw $7, %xmm1, %xmm3
+; AVX1-NEXT:    vpmullw %xmm3, %xmm4, %xmm3
+; AVX1-NEXT:    vpsrlw $8, %xmm3, %xmm3
+; AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddb %xmm0, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrlw $7, %xmm2, %xmm3
 ; AVX1-NEXT:    vpand %xmm8, %xmm3, %xmm3
-; AVX1-NEXT:    vpsrlw $2, %xmm1, %xmm1
-; AVX1-NEXT:    vpand %xmm9, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm7, %xmm1, %xmm1
-; AVX1-NEXT:    vpsubb %xmm7, %xmm1, %xmm1
-; AVX1-NEXT:    vpaddb %xmm3, %xmm1, %xmm1
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vpmullw %xmm5, %xmm3, %xmm3
-; AVX1-NEXT:    vpand %xmm6, %xmm3, %xmm3
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
-; AVX1-NEXT:    vpmullw %xmm5, %xmm1, %xmm1
-; AVX1-NEXT:    vpand %xmm6, %xmm1, %xmm1
-; AVX1-NEXT:    vpackuswb %xmm3, %xmm1, %xmm1
-; AVX1-NEXT:    vpsubb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vpsrlw $2, %xmm2, %xmm2
+; AVX1-NEXT:    vpand %xmm6, %xmm2, %xmm2
+; AVX1-NEXT:    vpxor %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpsubb %xmm7, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpsllw $3, %xmm2, %xmm3
+; AVX1-NEXT:    vpand %xmm5, %xmm3, %xmm3
+; AVX1-NEXT:    vpsubb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2NOBW-LABEL: test_rem7_32i8:
 ; AVX2NOBW:       # %bb.0:
 ; AVX2NOBW-NEXT:    vextracti128 $1, %ymm0, %xmm1
 ; AVX2NOBW-NEXT:    vpmovsxbw %xmm1, %ymm1
 ; AVX2NOBW-NEXT:    vmovdqa {{.*#+}} ymm2 = [65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427]
 ; AVX2NOBW-NEXT:    vpmullw %ymm2, %ymm1, %ymm1
 ; AVX2NOBW-NEXT:    vpsrlw $8, %ymm1, %ymm1
 ; AVX2NOBW-NEXT:    vpmovsxbw %xmm0, %ymm3
 ; AVX2NOBW-NEXT:    vpmullw %ymm2, %ymm3, %ymm2
 ; AVX2NOBW-NEXT:    vpsrlw $8, %ymm2, %ymm2
 ; AVX2NOBW-NEXT:    vpackuswb %ymm1, %ymm2, %ymm1
 ; AVX2NOBW-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,1,3]
 ; AVX2NOBW-NEXT:    vpaddb %ymm0, %ymm1, %ymm1
 ; AVX2NOBW-NEXT:    vpsrlw $2, %ymm1, %ymm2
 ; AVX2NOBW-NEXT:    vpand {{.*}}(%rip), %ymm2, %ymm2
 ; AVX2NOBW-NEXT:    vmovdqa {{.*#+}} ymm3 = [32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32]
 ; AVX2NOBW-NEXT:    vpxor %ymm3, %ymm2, %ymm2
 ; AVX2NOBW-NEXT:    vpsubb %ymm3, %ymm2, %ymm2
 ; AVX2NOBW-NEXT:    vpsrlw $7, %ymm1, %ymm1
 ; AVX2NOBW-NEXT:    vpand {{.*}}(%rip), %ymm1, %ymm1
 ; AVX2NOBW-NEXT:    vpaddb %ymm1, %ymm2, %ymm1
-; AVX2NOBW-NEXT:    vextracti128 $1, %ymm1, %xmm2
-; AVX2NOBW-NEXT:    vpmovsxbw %xmm2, %ymm2
-; AVX2NOBW-NEXT:    vmovdqa {{.*#+}} ymm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
-; AVX2NOBW-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
-; AVX2NOBW-NEXT:    vextracti128 $1, %ymm2, %xmm4
-; AVX2NOBW-NEXT:    vmovdqa {{.*#+}} xmm5 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2NOBW-NEXT:    vpshufb %xmm5, %xmm4, %xmm4
-; AVX2NOBW-NEXT:    vpshufb %xmm5, %xmm2, %xmm2
-; AVX2NOBW-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm4[0]
-; AVX2NOBW-NEXT:    vpmovsxbw %xmm1, %ymm1
-; AVX2NOBW-NEXT:    vpmullw %ymm3, %ymm1, %ymm1
-; AVX2NOBW-NEXT:    vextracti128 $1, %ymm1, %xmm3
-; AVX2NOBW-NEXT:    vpshufb %xmm5, %xmm3, %xmm3
-; AVX2NOBW-NEXT:    vpshufb %xmm5, %xmm1, %xmm1
-; AVX2NOBW-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm3[0]
-; AVX2NOBW-NEXT:    vinserti128 $1, %xmm2, %ymm1, %ymm1
-; AVX2NOBW-NEXT:    vpsubb %ymm1, %ymm0, %ymm0
+; AVX2NOBW-NEXT:    vpsllw $3, %ymm1, %ymm2
+; AVX2NOBW-NEXT:    vpand {{.*}}(%rip), %ymm2, %ymm2
+; AVX2NOBW-NEXT:    vpsubb %ymm2, %ymm1, %ymm1
+; AVX2NOBW-NEXT:    vpaddb %ymm1, %ymm0, %ymm0
 ; AVX2NOBW-NEXT:    retq
 ;
 ; AVX512BW-LABEL: test_rem7_32i8:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vpmovsxbw %ymm0, %zmm1
 ; AVX512BW-NEXT:    vpmullw {{.*}}(%rip), %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpsrlw $8, %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpmovwb %zmm1, %ymm1
 ; AVX512BW-NEXT:    vpaddb %ymm0, %ymm1, %ymm1
 ; AVX512BW-NEXT:    vpsrlw $2, %ymm1, %ymm2
 ; AVX512BW-NEXT:    vpand {{.*}}(%rip), %ymm2, %ymm2
 ; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm3 = [32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32]
 ; AVX512BW-NEXT:    vpxor %ymm3, %ymm2, %ymm2
 ; AVX512BW-NEXT:    vpsubb %ymm3, %ymm2, %ymm2
 ; AVX512BW-NEXT:    vpsrlw $7, %ymm1, %ymm1
 ; AVX512BW-NEXT:    vpand {{.*}}(%rip), %ymm1, %ymm1
 ; AVX512BW-NEXT:    vpaddb %ymm1, %ymm2, %ymm1
-; AVX512BW-NEXT:    vpmovsxbw %ymm1, %zmm1
-; AVX512BW-NEXT:    vpmullw {{.*}}(%rip), %zmm1, %zmm1
-; AVX512BW-NEXT:    vpmovwb %zmm1, %ymm1
-; AVX512BW-NEXT:    vpsubb %ymm1, %ymm0, %ymm0
+; AVX512BW-NEXT:    vpsllw $3, %ymm1, %ymm2
+; AVX512BW-NEXT:    vpand {{.*}}(%rip), %ymm2, %ymm2
+; AVX512BW-NEXT:    vpsubb %ymm2, %ymm1, %ymm1
+; AVX512BW-NEXT:    vpaddb %ymm1, %ymm0, %ymm0
 ; AVX512BW-NEXT:    retq
   %res = srem <32 x i8> %a, <i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7>
   ret <32 x i8> %res
 }

llvm/trunk/test/CodeGen/X86/vector-idiv-sdiv-512.ll

; AVX512BW-NEXT:    retq
   %res = srem <32 x i16> %a, <i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7>
   ret <32 x i16> %res
 }
 
 define <64 x i8> @test_rem7_64i8(<64 x i8> %a) nounwind {
 ; AVX512F-LABEL: test_rem7_64i8:
 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm2
-; AVX512F-NEXT:    vpmovsxbw %xmm2, %ymm3
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm2 = [65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427]
-; AVX512F-NEXT:    vpmullw %ymm2, %ymm3, %ymm3
-; AVX512F-NEXT:    vpsrlw $8, %ymm3, %ymm3
+; AVX512F-NEXT:    vpmovsxbw %xmm2, %ymm2
+; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm3 = [65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427]
+; AVX512F-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
+; AVX512F-NEXT:    vpsrlw $8, %ymm2, %ymm2
 ; AVX512F-NEXT:    vpmovsxbw %xmm0, %ymm4
-; AVX512F-NEXT:    vpmullw %ymm2, %ymm4, %ymm4
+; AVX512F-NEXT:    vpmullw %ymm3, %ymm4, %ymm4
 ; AVX512F-NEXT:    vpsrlw $8, %ymm4, %ymm4
-; AVX512F-NEXT:    vpackuswb %ymm3, %ymm4, %ymm3
-; AVX512F-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,1,3]
-; AVX512F-NEXT:    vpaddb %ymm0, %ymm3, %ymm3
-; AVX512F-NEXT:    vpsrlw $7, %ymm3, %ymm5
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm4 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
-; AVX512F-NEXT:    vpand %ymm4, %ymm5, %ymm7
-; AVX512F-NEXT:    vpsrlw $2, %ymm3, %ymm3
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm5 = [63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63]
-; AVX512F-NEXT:    vpand %ymm5, %ymm3, %ymm3
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm6 = [32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32]
-; AVX512F-NEXT:    vpxor %ymm6, %ymm3, %ymm3
-; AVX512F-NEXT:    vpsubb %ymm6, %ymm3, %ymm3
-; AVX512F-NEXT:    vpaddb %ymm7, %ymm3, %ymm7
-; AVX512F-NEXT:    vpmovsxbw %xmm7, %ymm8
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
-; AVX512F-NEXT:    vpmullw %ymm3, %ymm8, %ymm8
-; AVX512F-NEXT:    vpmovsxwd %ymm8, %zmm8
-; AVX512F-NEXT:    vpmovdb %zmm8, %xmm8
-; AVX512F-NEXT:    vextracti128 $1, %ymm7, %xmm7
-; AVX512F-NEXT:    vpmovsxbw %xmm7, %ymm7
-; AVX512F-NEXT:    vpmullw %ymm3, %ymm7, %ymm7
-; AVX512F-NEXT:    vpmovsxwd %ymm7, %zmm7
-; AVX512F-NEXT:    vpmovdb %zmm7, %xmm7
-; AVX512F-NEXT:    vinserti128 $1, %xmm7, %ymm8, %ymm7
-; AVX512F-NEXT:    vpsubb %ymm7, %ymm0, %ymm0
-; AVX512F-NEXT:    vextracti128 $1, %ymm1, %xmm7
-; AVX512F-NEXT:    vpmovsxbw %xmm7, %ymm7
-; AVX512F-NEXT:    vpmullw %ymm2, %ymm7, %ymm7
-; AVX512F-NEXT:    vpsrlw $8, %ymm7, %ymm7
-; AVX512F-NEXT:    vpmovsxbw %xmm1, %ymm8
-; AVX512F-NEXT:    vpmullw %ymm2, %ymm8, %ymm2
-; AVX512F-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX512F-NEXT:    vpackuswb %ymm7, %ymm2, %ymm2
+; AVX512F-NEXT:    vpackuswb %ymm2, %ymm4, %ymm2
 ; AVX512F-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,1,3]
-; AVX512F-NEXT:    vpaddb %ymm1, %ymm2, %ymm2
-; AVX512F-NEXT:    vpsrlw $7, %ymm2, %ymm7
-; AVX512F-NEXT:    vpand %ymm4, %ymm7, %ymm4
+; AVX512F-NEXT:    vpaddb %ymm0, %ymm2, %ymm2
+; AVX512F-NEXT:    vpsrlw $7, %ymm2, %ymm4
+; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm5 = [1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1,1]
+; AVX512F-NEXT:    vpand %ymm5, %ymm4, %ymm4
 ; AVX512F-NEXT:    vpsrlw $2, %ymm2, %ymm2
-; AVX512F-NEXT:    vpand %ymm5, %ymm2, %ymm2
-; AVX512F-NEXT:    vpxor %ymm6, %ymm2, %ymm2
-; AVX512F-NEXT:    vpsubb %ymm6, %ymm2, %ymm2
+; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm6 = [63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63]
+; AVX512F-NEXT:    vpand %ymm6, %ymm2, %ymm2
+; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm7 = [32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32]
+; AVX512F-NEXT:    vpxor %ymm7, %ymm2, %ymm2
+; AVX512F-NEXT:    vpsubb %ymm7, %ymm2, %ymm2
 ; AVX512F-NEXT:    vpaddb %ymm4, %ymm2, %ymm2
-; AVX512F-NEXT:    vpmovsxbw %xmm2, %ymm4
-; AVX512F-NEXT:    vpmullw %ymm3, %ymm4, %ymm4
-; AVX512F-NEXT:    vpmovsxwd %ymm4, %zmm4
-; AVX512F-NEXT:    vpmovdb %zmm4, %xmm4
-; AVX512F-NEXT:    vextracti128 $1, %ymm2, %xmm2
+; AVX512F-NEXT:    vpsllw $3, %ymm2, %ymm4
+; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm8 = [248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248]
+; AVX512F-NEXT:    vpand %ymm8, %ymm4, %ymm4
+; AVX512F-NEXT:    vpsubb %ymm4, %ymm2, %ymm2
+; AVX512F-NEXT:    vpaddb %ymm2, %ymm0, %ymm0
+; AVX512F-NEXT:    vextracti128 $1, %ymm1, %xmm2
 ; AVX512F-NEXT:    vpmovsxbw %xmm2, %ymm2
 ; AVX512F-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
-; AVX512F-NEXT:    vpmovsxwd %ymm2, %zmm2
-; AVX512F-NEXT:    vpmovdb %zmm2, %xmm2
-; AVX512F-NEXT:    vinserti128 $1, %xmm2, %ymm4, %ymm2
-; AVX512F-NEXT:    vpsubb %ymm2, %ymm1, %ymm1
+; AVX512F-NEXT:    vpsrlw $8, %ymm2, %ymm2
+; AVX512F-NEXT:    vpmovsxbw %xmm1, %ymm4
+; AVX512F-NEXT:    vpmullw %ymm3, %ymm4, %ymm3
+; AVX512F-NEXT:    vpsrlw $8, %ymm3, %ymm3
+; AVX512F-NEXT:    vpackuswb %ymm2, %ymm3, %ymm2
+; AVX512F-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,1,3]
+; AVX512F-NEXT:    vpaddb %ymm1, %ymm2, %ymm2
+; AVX512F-NEXT:    vpsrlw $7, %ymm2, %ymm3
+; AVX512F-NEXT:    vpand %ymm5, %ymm3, %ymm3
+; AVX512F-NEXT:    vpsrlw $2, %ymm2, %ymm2
+; AVX512F-NEXT:    vpand %ymm6, %ymm2, %ymm2
+; AVX512F-NEXT:    vpxor %ymm7, %ymm2, %ymm2
+; AVX512F-NEXT:    vpsubb %ymm7, %ymm2, %ymm2
+; AVX512F-NEXT:    vpaddb %ymm3, %ymm2, %ymm2
+; AVX512F-NEXT:    vpsllw $3, %ymm2, %ymm3
+; AVX512F-NEXT:    vpand %ymm8, %ymm3, %ymm3
+; AVX512F-NEXT:    vpsubb %ymm3, %ymm2, %ymm2
+; AVX512F-NEXT:    vpaddb %ymm2, %ymm1, %ymm1
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: test_rem7_64i8:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vpmovsxbw %ymm0, %zmm1
 ; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm2 = [65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427,65427]
 ; AVX512BW-NEXT:    vpmullw %zmm2, %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpsrlw $8, %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpmovwb %zmm1, %ymm1
 ; AVX512BW-NEXT:    vextracti64x4 $1, %zmm0, %ymm3
 ; AVX512BW-NEXT:    vpmovsxbw %ymm3, %zmm3
 ; AVX512BW-NEXT:    vpmullw %zmm2, %zmm3, %zmm2
 ; AVX512BW-NEXT:    vpsrlw $8, %zmm2, %zmm2
 ; AVX512BW-NEXT:    vpmovwb %zmm2, %ymm2
 ; AVX512BW-NEXT:    vinserti64x4 $1, %ymm2, %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpaddb %zmm0, %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpsrlw $2, %zmm1, %zmm2
 ; AVX512BW-NEXT:    vpandq {{.*}}(%rip), %zmm2, %zmm2
 ; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32,32]
 ; AVX512BW-NEXT:    vpxorq %zmm3, %zmm2, %zmm2
 ; AVX512BW-NEXT:    vpsubb %zmm3, %zmm2, %zmm2
 ; AVX512BW-NEXT:    vpsrlw $7, %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpandq {{.*}}(%rip), %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpaddb %zmm1, %zmm2, %zmm1
-; AVX512BW-NEXT:    vpmovsxbw %ymm1, %zmm2
-; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
-; AVX512BW-NEXT:    vpmullw %zmm3, %zmm2, %zmm2
-; AVX512BW-NEXT:    vpmovwb %zmm2, %ymm2
-; AVX512BW-NEXT:    vextracti64x4 $1, %zmm1, %ymm1
-; AVX512BW-NEXT:    vpmovsxbw %ymm1, %zmm1
-; AVX512BW-NEXT:    vpmullw %zmm3, %zmm1, %zmm1
-; AVX512BW-NEXT:    vpmovwb %zmm1, %ymm1
-; AVX512BW-NEXT:    vinserti64x4 $1, %ymm1, %zmm2, %zmm1
-; AVX512BW-NEXT:    vpsubb %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpsllw $3, %zmm1, %zmm2
+; AVX512BW-NEXT:    vpandq {{.*}}(%rip), %zmm2, %zmm2
+; AVX512BW-NEXT:    vpsubb %zmm2, %zmm1, %zmm1
+; AVX512BW-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    retq
   %res = srem <64 x i8> %a, <i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7>
   ret <64 x i8> %res
 }

llvm/trunk/test/CodeGen/X86/vector-idiv-udiv-128.ll

 ; SSE2-NEXT:    pmuludq %xmm1, %xmm3
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[1,3,2,3]
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
 ; SSE2-NEXT:    movdqa %xmm0, %xmm1
 ; SSE2-NEXT:    psubd %xmm2, %xmm1
 ; SSE2-NEXT:    psrld $1, %xmm1
 ; SSE2-NEXT:    paddd %xmm2, %xmm1
 ; SSE2-NEXT:    psrld $2, %xmm1
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [7,7,7,7]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[1,1,3,3]
-; SSE2-NEXT:    pmuludq %xmm2, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
-; SSE2-NEXT:    pmuludq %xmm2, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[0,2,2,3]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1]
-; SSE2-NEXT:    psubd %xmm1, %xmm0
+; SSE2-NEXT:    movdqa %xmm1, %xmm2
+; SSE2-NEXT:    pslld $3, %xmm2
+; SSE2-NEXT:    psubd %xmm2, %xmm1
+; SSE2-NEXT:    paddd %xmm0, %xmm1
+; SSE2-NEXT:    movdqa %xmm1, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: test_rem7_4i32:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,3,3]
 ; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [613566757,613566757,613566757,613566757]
 ; SSE41-NEXT:    pmuludq %xmm2, %xmm1
 ; SSE41-NEXT:    pmuludq %xmm0, %xmm2

 ; SSE2-NEXT:    movdqa %xmm0, %xmm1
 ; SSE2-NEXT:    psubb %xmm4, %xmm1
 ; SSE2-NEXT:    psrlw $1, %xmm1
 ; SSE2-NEXT:    pand {{.*}}(%rip), %xmm1
 ; SSE2-NEXT:    paddb %xmm4, %xmm1
 ; SSE2-NEXT:    psrlw $2, %xmm1
 ; SSE2-NEXT:    pand {{.*}}(%rip), %xmm1
 ; SSE2-NEXT:    movdqa %xmm1, %xmm2
-; SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm0[8],xmm2[9],xmm0[9],xmm2[10],xmm0[10],xmm2[11],xmm0[11],xmm2[12],xmm0[12],xmm2[13],xmm0[13],xmm2[14],xmm0[14],xmm2[15],xmm0[15]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
-; SSE2-NEXT:    pmullw %xmm3, %xmm2
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
-; SSE2-NEXT:    pand %xmm4, %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
-; SSE2-NEXT:    pmullw %xmm3, %xmm1
-; SSE2-NEXT:    pand %xmm4, %xmm1
-; SSE2-NEXT:    packuswb %xmm2, %xmm1
-; SSE2-NEXT:    psubb %xmm1, %xmm0
+; SSE2-NEXT:    psllw $3, %xmm2
+; SSE2-NEXT:    pand {{.*}}(%rip), %xmm2
+; SSE2-NEXT:    psubb %xmm2, %xmm1
+; SSE2-NEXT:    paddb %xmm0, %xmm1
+; SSE2-NEXT:    movdqa %xmm1, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSE41-LABEL: test_rem7_16i8:
 ; SSE41:       # %bb.0:
-; SSE41-NEXT:    pmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [37,37,37,37,37,37,37,37]
-; SSE41-NEXT:    pmullw %xmm2, %xmm1
-; SSE41-NEXT:    psrlw $8, %xmm1
+; SSE41-NEXT:    pmovzxbw {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
+; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [37,37,37,37,37,37,37,37]
+; SSE41-NEXT:    pmullw %xmm1, %xmm2
+; SSE41-NEXT:    psrlw $8, %xmm2
 ; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
 ; SSE41-NEXT:    pmovzxbw {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero,xmm3[4],zero,xmm3[5],zero,xmm3[6],zero,xmm3[7],zero
-; SSE41-NEXT:    pmullw %xmm2, %xmm3
+; SSE41-NEXT:    pmullw %xmm1, %xmm3
 ; SSE41-NEXT:    psrlw $8, %xmm3
-; SSE41-NEXT:    packuswb %xmm3, %xmm1
-; SSE41-NEXT:    movdqa %xmm0, %xmm2
-; SSE41-NEXT:    psubb %xmm1, %xmm2
-; SSE41-NEXT:    psrlw $1, %xmm2
-; SSE41-NEXT:    pand {{.*}}(%rip), %xmm2
-; SSE41-NEXT:    paddb %xmm1, %xmm2
-; SSE41-NEXT:    psrlw $2, %xmm2
+; SSE41-NEXT:    packuswb %xmm3, %xmm2
+; SSE41-NEXT:    movdqa %xmm0, %xmm1
+; SSE41-NEXT:    psubb %xmm2, %xmm1
+; SSE41-NEXT:    psrlw $1, %xmm1
+; SSE41-NEXT:    pand {{.*}}(%rip), %xmm1
+; SSE41-NEXT:    paddb %xmm2, %xmm1
+; SSE41-NEXT:    psrlw $2, %xmm1
+; SSE41-NEXT:    pand {{.*}}(%rip), %xmm1
+; SSE41-NEXT:    movdqa %xmm1, %xmm2
+; SSE41-NEXT:    psllw $3, %xmm2
 ; SSE41-NEXT:    pand {{.*}}(%rip), %xmm2
-; SSE41-NEXT:    pmovzxbw {{.*#+}} xmm1 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
-; SSE41-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
-; SSE41-NEXT:    pmullw %xmm3, %xmm2
-; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
-; SSE41-NEXT:    pand %xmm4, %xmm2
-; SSE41-NEXT:    pmullw %xmm3, %xmm1
-; SSE41-NEXT:    pand %xmm4, %xmm1
-; SSE41-NEXT:    packuswb %xmm2, %xmm1
-; SSE41-NEXT:    psubb %xmm1, %xmm0
+; SSE41-NEXT:    psubb %xmm2, %xmm1
+; SSE41-NEXT:    paddb %xmm0, %xmm1
+; SSE41-NEXT:    movdqa %xmm1, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: test_rem7_16i8:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [37,37,37,37,37,37,37,37]
 ; AVX1-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
 ; AVX1-NEXT:    vpsrlw $8, %xmm1, %xmm1
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero,xmm3[4],zero,xmm3[5],zero,xmm3[6],zero,xmm3[7],zero
 ; AVX1-NEXT:    vpmullw %xmm2, %xmm3, %xmm2
 ; AVX1-NEXT:    vpsrlw $8, %xmm2, %xmm2
 ; AVX1-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
 ; AVX1-NEXT:    vpsubb %xmm1, %xmm0, %xmm2
 ; AVX1-NEXT:    vpsrlw $1, %xmm2, %xmm2
 ; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
 ; AVX1-NEXT:    vpaddb %xmm1, %xmm2, %xmm1
 ; AVX1-NEXT:    vpsrlw $2, %xmm1, %xmm1
 ; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
-; AVX1-NEXT:    vpmullw %xmm3, %xmm2, %xmm2
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpand %xmm4, %xmm2, %xmm2
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
-; AVX1-NEXT:    vpmullw %xmm3, %xmm1, %xmm1
-; AVX1-NEXT:    vpand %xmm4, %xmm1, %xmm1
-; AVX1-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpsubb %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpsllw $3, %xmm1, %xmm2
+; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
+; AVX1-NEXT:    vpsubb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2NOBW-LABEL: test_rem7_16i8:
 ; AVX2NOBW:       # %bb.0:
 ; AVX2NOBW-NEXT:    vpmovzxbw {{.*#+}} ymm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
 ; AVX2NOBW-NEXT:    vpmullw {{.*}}(%rip), %ymm1, %ymm1
 ; AVX2NOBW-NEXT:    vpsrlw $8, %ymm1, %ymm1
 ; AVX2NOBW-NEXT:    vextracti128 $1, %ymm1, %xmm2
 ; AVX2NOBW-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
 ; AVX2NOBW-NEXT:    vpsubb %xmm1, %xmm0, %xmm2
 ; AVX2NOBW-NEXT:    vpsrlw $1, %xmm2, %xmm2
 ; AVX2NOBW-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
 ; AVX2NOBW-NEXT:    vpaddb %xmm1, %xmm2, %xmm1
 ; AVX2NOBW-NEXT:    vpsrlw $2, %xmm1, %xmm1
 ; AVX2NOBW-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
-; AVX2NOBW-NEXT:    vpmovsxbw %xmm1, %ymm1
-; AVX2NOBW-NEXT:    vpmullw {{.*}}(%rip), %ymm1, %ymm1
-; AVX2NOBW-NEXT:    vextracti128 $1, %ymm1, %xmm2
-; AVX2NOBW-NEXT:    vmovdqa {{.*#+}} xmm3 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2NOBW-NEXT:    vpshufb %xmm3, %xmm2, %xmm2
-; AVX2NOBW-NEXT:    vpshufb %xmm3, %xmm1, %xmm1
-; AVX2NOBW-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
-; AVX2NOBW-NEXT:    vpsubb %xmm1, %xmm0, %xmm0
+; AVX2NOBW-NEXT:    vpsllw $3, %xmm1, %xmm2
+; AVX2NOBW-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
+; AVX2NOBW-NEXT:    vpsubb %xmm2, %xmm1, %xmm1
+; AVX2NOBW-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX2NOBW-NEXT:    vzeroupper
 ; AVX2NOBW-NEXT:    retq
 ;
 ; AVX512BW-LABEL: test_rem7_16i8:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} ymm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
 ; AVX512BW-NEXT:    vpmullw {{.*}}(%rip), %ymm1, %ymm1
 ; AVX512BW-NEXT:    vpsrlw $8, %ymm1, %ymm1
 ; AVX512BW-NEXT:    vpmovwb %zmm1, %ymm1
 ; AVX512BW-NEXT:    vpsubb %xmm1, %xmm0, %xmm2
 ; AVX512BW-NEXT:    vpsrlw $1, %xmm2, %xmm2
 ; AVX512BW-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
 ; AVX512BW-NEXT:    vpaddb %xmm1, %xmm2, %xmm1
 ; AVX512BW-NEXT:    vpsrlw $2, %xmm1, %xmm1
 ; AVX512BW-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
-; AVX512BW-NEXT:    vpmovsxbw %xmm1, %ymm1
-; AVX512BW-NEXT:    vpmullw {{.*}}(%rip), %ymm1, %ymm1
-; AVX512BW-NEXT:    vpmovwb %zmm1, %ymm1
-; AVX512BW-NEXT:    vpsubb %xmm1, %xmm0, %xmm0
+; AVX512BW-NEXT:    vpsllw $3, %xmm1, %xmm2
+; AVX512BW-NEXT:    vpand {{.*}}(%rip), %xmm2, %xmm2
+; AVX512BW-NEXT:    vpsubb %xmm2, %xmm1, %xmm1
+; AVX512BW-NEXT:    vpaddb %xmm1, %xmm0, %xmm0
 ; AVX512BW-NEXT:    vzeroupper
 ; AVX512BW-NEXT:    retq
   %res = urem <16 x i8> %a, <i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7>
   ret <16 x i8> %res
 }

llvm/trunk/test/CodeGen/X86/vector-idiv-udiv-256.ll

 ; AVX1-NEXT:    vpmuludq %xmm3, %xmm2, %xmm2
 ; AVX1-NEXT:    vpmuludq %xmm3, %xmm1, %xmm4
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm4[1,1,3,3]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm4[0,1],xmm2[2,3],xmm4[4,5],xmm2[6,7]
 ; AVX1-NEXT:    vpsubd %xmm2, %xmm1, %xmm4
 ; AVX1-NEXT:    vpsrld $1, %xmm4, %xmm4
 ; AVX1-NEXT:    vpaddd %xmm2, %xmm4, %xmm2
 ; AVX1-NEXT:    vpsrld $2, %xmm2, %xmm2
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [7,7,7,7]
-; AVX1-NEXT:    vpmulld %xmm4, %xmm2, %xmm2
-; AVX1-NEXT:    vpsubd %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpslld $3, %xmm2, %xmm4
+; AVX1-NEXT:    vpsubd %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddd %xmm2, %xmm1, %xmm1
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
 ; AVX1-NEXT:    vpmuludq %xmm3, %xmm2, %xmm2
 ; AVX1-NEXT:    vpmuludq %xmm3, %xmm0, %xmm3
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm3[0,1],xmm2[2,3],xmm3[4,5],xmm2[6,7]
 ; AVX1-NEXT:    vpsubd %xmm2, %xmm0, %xmm3
 ; AVX1-NEXT:    vpsrld $1, %xmm3, %xmm3
 ; AVX1-NEXT:    vpaddd %xmm2, %xmm3, %xmm2
 ; AVX1-NEXT:    vpsrld $2, %xmm2, %xmm2
-; AVX1-NEXT:    vpmulld %xmm4, %xmm2, %xmm2
-; AVX1-NEXT:    vpsubd %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpslld $3, %xmm2, %xmm3
+; AVX1-NEXT:    vpsubd %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddd %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: test_rem7_8i32:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpshufd {{.*#+}} ymm1 = ymm0[1,1,3,3,5,5,7,7]
 ; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm2 = [613566757,613566757,613566757,613566757,613566757,613566757,613566757,613566757]
 ; AVX2-NEXT:    vpmuludq %ymm2, %ymm1, %ymm1

 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [9363,9363,9363,9363,9363,9363,9363,9363]
 ; AVX1-NEXT:    vpmulhuw %xmm2, %xmm1, %xmm3
 ; AVX1-NEXT:    vpsubw %xmm3, %xmm1, %xmm4
 ; AVX1-NEXT:    vpsrlw $1, %xmm4, %xmm4
 ; AVX1-NEXT:    vpaddw %xmm3, %xmm4, %xmm3
 ; AVX1-NEXT:    vpsrlw $2, %xmm3, %xmm3
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [7,7,7,7,7,7,7,7]
-; AVX1-NEXT:    vpmullw %xmm4, %xmm3, %xmm3
-; AVX1-NEXT:    vpsubw %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpsllw $3, %xmm3, %xmm4
+; AVX1-NEXT:    vpsubw %xmm4, %xmm3, %xmm3
+; AVX1-NEXT:    vpaddw %xmm3, %xmm1, %xmm1
 ; AVX1-NEXT:    vpmulhuw %xmm2, %xmm0, %xmm2
 ; AVX1-NEXT:    vpsubw %xmm2, %xmm0, %xmm3
 ; AVX1-NEXT:    vpsrlw $1, %xmm3, %xmm3
 ; AVX1-NEXT:    vpaddw %xmm2, %xmm3, %xmm2
 ; AVX1-NEXT:    vpsrlw $2, %xmm2, %xmm2
-; AVX1-NEXT:    vpmullw %xmm4, %xmm2, %xmm2
-; AVX1-NEXT:    vpsubw %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpsllw $3, %xmm2, %xmm3
+; AVX1-NEXT:    vpsubw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddw %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: test_rem7_16i16:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpmulhuw {{.*}}(%rip), %ymm0, %ymm1
 ; AVX2-NEXT:    vpsubw %ymm1, %ymm0, %ymm2
 ; AVX2-NEXT:    vpsrlw $1, %ymm2, %ymm2

 ; AVX1-NEXT:    vpsrlw $8, %xmm2, %xmm2
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm4 = xmm1[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm4 = xmm4[0],zero,xmm4[1],zero,xmm4[2],zero,xmm4[3],zero,xmm4[4],zero,xmm4[5],zero,xmm4[6],zero,xmm4[7],zero
 ; AVX1-NEXT:    vpmullw %xmm3, %xmm4, %xmm4
 ; AVX1-NEXT:    vpsrlw $8, %xmm4, %xmm4
 ; AVX1-NEXT:    vpackuswb %xmm4, %xmm2, %xmm2
 ; AVX1-NEXT:    vpsubb %xmm2, %xmm1, %xmm4
 ; AVX1-NEXT:    vpsrlw $1, %xmm4, %xmm4
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm8 = [127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127]
-; AVX1-NEXT:    vpand %xmm8, %xmm4, %xmm4
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm5 = [127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127]
+; AVX1-NEXT:    vpand %xmm5, %xmm4, %xmm4
 ; AVX1-NEXT:    vpaddb %xmm2, %xmm4, %xmm2
 ; AVX1-NEXT:    vpsrlw $2, %xmm2, %xmm2
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63]
 ; AVX1-NEXT:    vpand %xmm4, %xmm2, %xmm2
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm6 = xmm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm7 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
-; AVX1-NEXT:    vpmullw %xmm7, %xmm6, %xmm6
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm5 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpand %xmm5, %xmm6, %xmm6
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
-; AVX1-NEXT:    vpmullw %xmm7, %xmm2, %xmm2
-; AVX1-NEXT:    vpand %xmm5, %xmm2, %xmm2
-; AVX1-NEXT:    vpackuswb %xmm6, %xmm2, %xmm2
-; AVX1-NEXT:    vpsubb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpsllw $3, %xmm2, %xmm6
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm7 = [248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248]
+; AVX1-NEXT:    vpand %xmm7, %xmm6, %xmm6
+; AVX1-NEXT:    vpsubb %xmm6, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddb %xmm2, %xmm1, %xmm1
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX1-NEXT:    vpmullw %xmm3, %xmm2, %xmm2
 ; AVX1-NEXT:    vpsrlw $8, %xmm2, %xmm2
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm6 = xmm0[2,3,0,1]
 ; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm6 = xmm6[0],zero,xmm6[1],zero,xmm6[2],zero,xmm6[3],zero,xmm6[4],zero,xmm6[5],zero,xmm6[6],zero,xmm6[7],zero
 ; AVX1-NEXT:    vpmullw %xmm3, %xmm6, %xmm3
 ; AVX1-NEXT:    vpsrlw $8, %xmm3, %xmm3
 ; AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
 ; AVX1-NEXT:    vpsubb %xmm2, %xmm0, %xmm3
 ; AVX1-NEXT:    vpsrlw $1, %xmm3, %xmm3
-; AVX1-NEXT:    vpand %xmm8, %xmm3, %xmm3
+; AVX1-NEXT:    vpand %xmm5, %xmm3, %xmm3
 ; AVX1-NEXT:    vpaddb %xmm2, %xmm3, %xmm2
 ; AVX1-NEXT:    vpsrlw $2, %xmm2, %xmm2
 ; AVX1-NEXT:    vpand %xmm4, %xmm2, %xmm2
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vpmullw %xmm7, %xmm3, %xmm3
-; AVX1-NEXT:    vpand %xmm5, %xmm3, %xmm3
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero
-; AVX1-NEXT:    vpmullw %xmm7, %xmm2, %xmm2
-; AVX1-NEXT:    vpand %xmm5, %xmm2, %xmm2
-; AVX1-NEXT:    vpackuswb %xmm3, %xmm2, %xmm2
-; AVX1-NEXT:    vpsubb %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpsllw $3, %xmm2, %xmm3
+; AVX1-NEXT:    vpand %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpsubb %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpaddb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2NOBW-LABEL: test_rem7_32i8:
 ; AVX2NOBW:       # %bb.0:
 ; AVX2NOBW-NEXT:    vextracti128 $1, %ymm0, %xmm1
 ; AVX2NOBW-NEXT:    vpmovzxbw {{.*#+}} ymm1 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero
 ; AVX2NOBW-NEXT:    vmovdqa {{.*#+}} ymm2 = [37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37]
 ; AVX2NOBW-NEXT:    vpmullw %ymm2, %ymm1, %ymm1
 ; AVX2NOBW-NEXT:    vpsrlw $8, %ymm1, %ymm1
 ; AVX2NOBW-NEXT:    vpmovzxbw {{.*#+}} ymm3 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
 ; AVX2NOBW-NEXT:    vpmullw %ymm2, %ymm3, %ymm2
 ; AVX2NOBW-NEXT:    vpsrlw $8, %ymm2, %ymm2
 ; AVX2NOBW-NEXT:    vpackuswb %ymm1, %ymm2, %ymm1
 ; AVX2NOBW-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,1,3]
 ; AVX2NOBW-NEXT:    vpsubb %ymm1, %ymm0, %ymm2
 ; AVX2NOBW-NEXT:    vpsrlw $1, %ymm2, %ymm2
 ; AVX2NOBW-NEXT:    vpand {{.*}}(%rip), %ymm2, %ymm2
 ; AVX2NOBW-NEXT:    vpaddb %ymm1, %ymm2, %ymm1
 ; AVX2NOBW-NEXT:    vpsrlw $2, %ymm1, %ymm1
 ; AVX2NOBW-NEXT:    vpand {{.*}}(%rip), %ymm1, %ymm1
-; AVX2NOBW-NEXT:    vextracti128 $1, %ymm1, %xmm2
-; AVX2NOBW-NEXT:    vpmovsxbw %xmm2, %ymm2
-; AVX2NOBW-NEXT:    vmovdqa {{.*#+}} ymm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
-; AVX2NOBW-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
-; AVX2NOBW-NEXT:    vextracti128 $1, %ymm2, %xmm4
-; AVX2NOBW-NEXT:    vmovdqa {{.*#+}} xmm5 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX2NOBW-NEXT:    vpshufb %xmm5, %xmm4, %xmm4
-; AVX2NOBW-NEXT:    vpshufb %xmm5, %xmm2, %xmm2
-; AVX2NOBW-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm4[0]
-; AVX2NOBW-NEXT:    vpmovsxbw %xmm1, %ymm1
-; AVX2NOBW-NEXT:    vpmullw %ymm3, %ymm1, %ymm1
-; AVX2NOBW-NEXT:    vextracti128 $1, %ymm1, %xmm3
-; AVX2NOBW-NEXT:    vpshufb %xmm5, %xmm3, %xmm3
-; AVX2NOBW-NEXT:    vpshufb %xmm5, %xmm1, %xmm1
-; AVX2NOBW-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm3[0]
-; AVX2NOBW-NEXT:    vinserti128 $1, %xmm2, %ymm1, %ymm1
-; AVX2NOBW-NEXT:    vpsubb %ymm1, %ymm0, %ymm0
+; AVX2NOBW-NEXT:    vpsllw $3, %ymm1, %ymm2
+; AVX2NOBW-NEXT:    vpand {{.*}}(%rip), %ymm2, %ymm2
+; AVX2NOBW-NEXT:    vpsubb %ymm2, %ymm1, %ymm1
+; AVX2NOBW-NEXT:    vpaddb %ymm1, %ymm0, %ymm0
 ; AVX2NOBW-NEXT:    retq
 ;
 ; AVX512BW-LABEL: test_rem7_32i8:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} zmm1 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero,ymm0[16],zero,ymm0[17],zero,ymm0[18],zero,ymm0[19],zero,ymm0[20],zero,ymm0[21],zero,ymm0[22],zero,ymm0[23],zero,ymm0[24],zero,ymm0[25],zero,ymm0[26],zero,ymm0[27],zero,ymm0[28],zero,ymm0[29],zero,ymm0[30],zero,ymm0[31],zero
 ; AVX512BW-NEXT:    vpmullw {{.*}}(%rip), %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpsrlw $8, %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpmovwb %zmm1, %ymm1
 ; AVX512BW-NEXT:    vpsubb %ymm1, %ymm0, %ymm2
 ; AVX512BW-NEXT:    vpsrlw $1, %ymm2, %ymm2
 ; AVX512BW-NEXT:    vpand {{.*}}(%rip), %ymm2, %ymm2
 ; AVX512BW-NEXT:    vpaddb %ymm1, %ymm2, %ymm1
 ; AVX512BW-NEXT:    vpsrlw $2, %ymm1, %ymm1
 ; AVX512BW-NEXT:    vpand {{.*}}(%rip), %ymm1, %ymm1
-; AVX512BW-NEXT:    vpmovsxbw %ymm1, %zmm1
-; AVX512BW-NEXT:    vpmullw {{.*}}(%rip), %zmm1, %zmm1
-; AVX512BW-NEXT:    vpmovwb %zmm1, %ymm1
-; AVX512BW-NEXT:    vpsubb %ymm1, %ymm0, %ymm0
+; AVX512BW-NEXT:    vpsllw $3, %ymm1, %ymm2
+; AVX512BW-NEXT:    vpand {{.*}}(%rip), %ymm2, %ymm2
+; AVX512BW-NEXT:    vpsubb %ymm2, %ymm1, %ymm1
+; AVX512BW-NEXT:    vpaddb %ymm1, %ymm0, %ymm0
 ; AVX512BW-NEXT:    retq
   %res = urem <32 x i8> %a, <i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7>
   ret <32 x i8> %res
 }

llvm/trunk/test/CodeGen/X86/vector-idiv-udiv-512.ll

; AVX512BW-NEXT:    retq
   %res = urem <32 x i16> %a, <i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7>
   ret <32 x i16> %res
 }
 
 define <64 x i8> @test_rem7_64i8(<64 x i8> %a) nounwind {
 ; AVX512F-LABEL: test_rem7_64i8:
 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    vextracti128 $1, %ymm0, %xmm2
-; AVX512F-NEXT:    vpmovzxbw {{.*#+}} ymm3 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero,xmm2[8],zero,xmm2[9],zero,xmm2[10],zero,xmm2[11],zero,xmm2[12],zero,xmm2[13],zero,xmm2[14],zero,xmm2[15],zero
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm2 = [37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37]
-; AVX512F-NEXT:    vpmullw %ymm2, %ymm3, %ymm3
-; AVX512F-NEXT:    vpsrlw $8, %ymm3, %ymm3
+; AVX512F-NEXT:    vpmovzxbw {{.*#+}} ymm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero,xmm2[8],zero,xmm2[9],zero,xmm2[10],zero,xmm2[11],zero,xmm2[12],zero,xmm2[13],zero,xmm2[14],zero,xmm2[15],zero
+; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm3 = [37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37]
+; AVX512F-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
+; AVX512F-NEXT:    vpsrlw $8, %ymm2, %ymm2
 ; AVX512F-NEXT:    vpmovzxbw {{.*#+}} ymm4 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero,xmm0[8],zero,xmm0[9],zero,xmm0[10],zero,xmm0[11],zero,xmm0[12],zero,xmm0[13],zero,xmm0[14],zero,xmm0[15],zero
-; AVX512F-NEXT:    vpmullw %ymm2, %ymm4, %ymm4
+; AVX512F-NEXT:    vpmullw %ymm3, %ymm4, %ymm4
 ; AVX512F-NEXT:    vpsrlw $8, %ymm4, %ymm4
-; AVX512F-NEXT:    vpackuswb %ymm3, %ymm4, %ymm3
-; AVX512F-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,2,1,3]
-; AVX512F-NEXT:    vpsubb %ymm3, %ymm0, %ymm4
+; AVX512F-NEXT:    vpackuswb %ymm2, %ymm4, %ymm2
+; AVX512F-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,1,3]
+; AVX512F-NEXT:    vpsubb %ymm2, %ymm0, %ymm4
 ; AVX512F-NEXT:    vpsrlw $1, %ymm4, %ymm4
 ; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm5 = [127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127,127]
 ; AVX512F-NEXT:    vpand %ymm5, %ymm4, %ymm4
-; AVX512F-NEXT:    vpaddb %ymm3, %ymm4, %ymm3
-; AVX512F-NEXT:    vpsrlw $2, %ymm3, %ymm3
+; AVX512F-NEXT:    vpaddb %ymm2, %ymm4, %ymm2
+; AVX512F-NEXT:    vpsrlw $2, %ymm2, %ymm2
 ; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm4 = [63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63,63]
-; AVX512F-NEXT:    vpand %ymm4, %ymm3, %ymm6
-; AVX512F-NEXT:    vpmovsxbw %xmm6, %ymm7
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
-; AVX512F-NEXT:    vpmullw %ymm3, %ymm7, %ymm7
-; AVX512F-NEXT:    vpmovsxwd %ymm7, %zmm7
-; AVX512F-NEXT:    vpmovdb %zmm7, %xmm7
-; AVX512F-NEXT:    vextracti128 $1, %ymm6, %xmm6
-; AVX512F-NEXT:    vpmovsxbw %xmm6, %ymm6
-; AVX512F-NEXT:    vpmullw %ymm3, %ymm6, %ymm6
-; AVX512F-NEXT:    vpmovsxwd %ymm6, %zmm6
-; AVX512F-NEXT:    vpmovdb %zmm6, %xmm6
-; AVX512F-NEXT:    vinserti128 $1, %xmm6, %ymm7, %ymm6
-; AVX512F-NEXT:    vpsubb %ymm6, %ymm0, %ymm0
-; AVX512F-NEXT:    vextracti128 $1, %ymm1, %xmm6
-; AVX512F-NEXT:    vpmovzxbw {{.*#+}} ymm6 = xmm6[0],zero,xmm6[1],zero,xmm6[2],zero,xmm6[3],zero,xmm6[4],zero,xmm6[5],zero,xmm6[6],zero,xmm6[7],zero,xmm6[8],zero,xmm6[9],zero,xmm6[10],zero,xmm6[11],zero,xmm6[12],zero,xmm6[13],zero,xmm6[14],zero,xmm6[15],zero
-; AVX512F-NEXT:    vpmullw %ymm2, %ymm6, %ymm6
-; AVX512F-NEXT:    vpsrlw $8, %ymm6, %ymm6
-; AVX512F-NEXT:    vpmovzxbw {{.*#+}} ymm7 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero
-; AVX512F-NEXT:    vpmullw %ymm2, %ymm7, %ymm2
+; AVX512F-NEXT:    vpand %ymm4, %ymm2, %ymm2
+; AVX512F-NEXT:    vpsllw $3, %ymm2, %ymm6
+; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm7 = [248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248,248]
+; AVX512F-NEXT:    vpand %ymm7, %ymm6, %ymm6
+; AVX512F-NEXT:    vpsubb %ymm6, %ymm2, %ymm2
+; AVX512F-NEXT:    vpaddb %ymm2, %ymm0, %ymm0
+; AVX512F-NEXT:    vextracti128 $1, %ymm1, %xmm2
+; AVX512F-NEXT:    vpmovzxbw {{.*#+}} ymm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero,xmm2[4],zero,xmm2[5],zero,xmm2[6],zero,xmm2[7],zero,xmm2[8],zero,xmm2[9],zero,xmm2[10],zero,xmm2[11],zero,xmm2[12],zero,xmm2[13],zero,xmm2[14],zero,xmm2[15],zero
+; AVX512F-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
 ; AVX512F-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX512F-NEXT:    vpackuswb %ymm6, %ymm2, %ymm2
+; AVX512F-NEXT:    vpmovzxbw {{.*#+}} ymm6 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero,xmm1[8],zero,xmm1[9],zero,xmm1[10],zero,xmm1[11],zero,xmm1[12],zero,xmm1[13],zero,xmm1[14],zero,xmm1[15],zero
+; AVX512F-NEXT:    vpmullw %ymm3, %ymm6, %ymm3
+; AVX512F-NEXT:    vpsrlw $8, %ymm3, %ymm3
+; AVX512F-NEXT:    vpackuswb %ymm2, %ymm3, %ymm2
 ; AVX512F-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,1,3]
-; AVX512F-NEXT:    vpsubb %ymm2, %ymm1, %ymm6
-; AVX512F-NEXT:    vpsrlw $1, %ymm6, %ymm6
-; AVX512F-NEXT:    vpand %ymm5, %ymm6, %ymm5
-; AVX512F-NEXT:    vpaddb %ymm2, %ymm5, %ymm2
+; AVX512F-NEXT:    vpsubb %ymm2, %ymm1, %ymm3
+; AVX512F-NEXT:    vpsrlw $1, %ymm3, %ymm3
+; AVX512F-NEXT:    vpand %ymm5, %ymm3, %ymm3
+; AVX512F-NEXT:    vpaddb %ymm2, %ymm3, %ymm2
 ; AVX512F-NEXT:    vpsrlw $2, %ymm2, %ymm2
 ; AVX512F-NEXT:    vpand %ymm4, %ymm2, %ymm2
-; AVX512F-NEXT:    vpmovsxbw %xmm2, %ymm4
-; AVX512F-NEXT:    vpmullw %ymm3, %ymm4, %ymm4
-; AVX512F-NEXT:    vpmovsxwd %ymm4, %zmm4
-; AVX512F-NEXT:    vpmovdb %zmm4, %xmm4
-; AVX512F-NEXT:    vextracti128 $1, %ymm2, %xmm2
-; AVX512F-NEXT:    vpmovsxbw %xmm2, %ymm2
-; AVX512F-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
-; AVX512F-NEXT:    vpmovsxwd %ymm2, %zmm2
-; AVX512F-NEXT:    vpmovdb %zmm2, %xmm2
-; AVX512F-NEXT:    vinserti128 $1, %xmm2, %ymm4, %ymm2
-; AVX512F-NEXT:    vpsubb %ymm2, %ymm1, %ymm1
+; AVX512F-NEXT:    vpsllw $3, %ymm2, %ymm3
+; AVX512F-NEXT:    vpand %ymm7, %ymm3, %ymm3
+; AVX512F-NEXT:    vpsubb %ymm3, %ymm2, %ymm2
+; AVX512F-NEXT:    vpaddb %ymm2, %ymm1, %ymm1
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: test_rem7_64i8:
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} zmm1 = ymm0[0],zero,ymm0[1],zero,ymm0[2],zero,ymm0[3],zero,ymm0[4],zero,ymm0[5],zero,ymm0[6],zero,ymm0[7],zero,ymm0[8],zero,ymm0[9],zero,ymm0[10],zero,ymm0[11],zero,ymm0[12],zero,ymm0[13],zero,ymm0[14],zero,ymm0[15],zero,ymm0[16],zero,ymm0[17],zero,ymm0[18],zero,ymm0[19],zero,ymm0[20],zero,ymm0[21],zero,ymm0[22],zero,ymm0[23],zero,ymm0[24],zero,ymm0[25],zero,ymm0[26],zero,ymm0[27],zero,ymm0[28],zero,ymm0[29],zero,ymm0[30],zero,ymm0[31],zero
 ; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm2 = [37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37,37]
 ; AVX512BW-NEXT:    vpmullw %zmm2, %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpsrlw $8, %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpmovwb %zmm1, %ymm1
 ; AVX512BW-NEXT:    vextracti64x4 $1, %zmm0, %ymm3
 ; AVX512BW-NEXT:    vpmovzxbw {{.*#+}} zmm3 = ymm3[0],zero,ymm3[1],zero,ymm3[2],zero,ymm3[3],zero,ymm3[4],zero,ymm3[5],zero,ymm3[6],zero,ymm3[7],zero,ymm3[8],zero,ymm3[9],zero,ymm3[10],zero,ymm3[11],zero,ymm3[12],zero,ymm3[13],zero,ymm3[14],zero,ymm3[15],zero,ymm3[16],zero,ymm3[17],zero,ymm3[18],zero,ymm3[19],zero,ymm3[20],zero,ymm3[21],zero,ymm3[22],zero,ymm3[23],zero,ymm3[24],zero,ymm3[25],zero,ymm3[26],zero,ymm3[27],zero,ymm3[28],zero,ymm3[29],zero,ymm3[30],zero,ymm3[31],zero
 ; AVX512BW-NEXT:    vpmullw %zmm2, %zmm3, %zmm2
 ; AVX512BW-NEXT:    vpsrlw $8, %zmm2, %zmm2
 ; AVX512BW-NEXT:    vpmovwb %zmm2, %ymm2
 ; AVX512BW-NEXT:    vinserti64x4 $1, %ymm2, %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpsubb %zmm1, %zmm0, %zmm2
 ; AVX512BW-NEXT:    vpsrlw $1, %zmm2, %zmm2
 ; AVX512BW-NEXT:    vpandq {{.*}}(%rip), %zmm2, %zmm2
 ; AVX512BW-NEXT:    vpaddb %zmm1, %zmm2, %zmm1
 ; AVX512BW-NEXT:    vpsrlw $2, %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpandq {{.*}}(%rip), %zmm1, %zmm1
-; AVX512BW-NEXT:    vpmovsxbw %ymm1, %zmm2
-; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7,7]
-; AVX512BW-NEXT:    vpmullw %zmm3, %zmm2, %zmm2
-; AVX512BW-NEXT:    vpmovwb %zmm2, %ymm2
-; AVX512BW-NEXT:    vextracti64x4 $1, %zmm1, %ymm1
-; AVX512BW-NEXT:    vpmovsxbw %ymm1, %zmm1
-; AVX512BW-NEXT:    vpmullw %zmm3, %zmm1, %zmm1
-; AVX512BW-NEXT:    vpmovwb %zmm1, %ymm1
-; AVX512BW-NEXT:    vinserti64x4 $1, %ymm1, %zmm2, %zmm1
-; AVX512BW-NEXT:    vpsubb %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpsllw $3, %zmm1, %zmm2
+; AVX512BW-NEXT:    vpandq {{.*}}(%rip), %zmm2, %zmm2
+; AVX512BW-NEXT:    vpsubb %zmm2, %zmm1, %zmm1
+; AVX512BW-NEXT:    vpaddb %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    retq
   %res = urem <64 x i8> %a, <i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7, i8 7,i8 7, i8 7, i8 7, i8 7>
   ret <64 x i8> %res
 }

llvm/trunk/test/CodeGen/X86/vector-mul.ll

 
 ;
 ; PowOf2 + 1 (uniform)
 ;
 
 define <2 x i64> @mul_v2i64_17(<2 x i64> %a0) nounwind {
 ; X86-LABEL: mul_v2i64_17:
 ; X86:       # %bb.0:
-; X86-NEXT:    movdqa {{.*#+}} xmm1 = [17,0,17,0]
-; X86-NEXT:    movdqa %xmm0, %xmm2
-; X86-NEXT:    pmuludq %xmm1, %xmm2
-; X86-NEXT:    psrlq $32, %xmm0
-; X86-NEXT:    pmuludq %xmm1, %xmm0
-; X86-NEXT:    psllq $32, %xmm0
-; X86-NEXT:    paddq %xmm2, %xmm0
+; X86-NEXT:    movdqa %xmm0, %xmm1
+; X86-NEXT:    psllq $4, %xmm1
+; X86-NEXT:    paddq %xmm0, %xmm1
+; X86-NEXT:    movdqa %xmm1, %xmm0
 ; X86-NEXT:    retl
 ;
 ; X64-LABEL: mul_v2i64_17:
 ; X64:       # %bb.0:
-; X64-NEXT:    movdqa {{.*#+}} xmm1 = [17,17]
-; X64-NEXT:    movdqa %xmm0, %xmm2
-; X64-NEXT:    pmuludq %xmm1, %xmm2
-; X64-NEXT:    psrlq $32, %xmm0
-; X64-NEXT:    pmuludq %xmm1, %xmm0
-; X64-NEXT:    psllq $32, %xmm0
-; X64-NEXT:    paddq %xmm2, %xmm0
+; X64-NEXT:    movdqa %xmm0, %xmm1
+; X64-NEXT:    psllq $4, %xmm1
+; X64-NEXT:    paddq %xmm0, %xmm1
+; X64-NEXT:    movdqa %xmm1, %xmm0
 ; X64-NEXT:    retq
 ;
 ; X64-AVX-LABEL: mul_v2i64_17:
 ; X64-AVX:       # %bb.0:
-; X64-AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [17,17]
-; X64-AVX-NEXT:    vpmuludq %xmm1, %xmm0, %xmm2
-; X64-AVX-NEXT:    vpsrlq $32, %xmm0, %xmm0
-; X64-AVX-NEXT:    vpmuludq %xmm1, %xmm0, %xmm0
-; X64-AVX-NEXT:    vpsllq $32, %xmm0, %xmm0
-; X64-AVX-NEXT:    vpaddq %xmm0, %xmm2, %xmm0
+; X64-AVX-NEXT:    vpsllq $4, %xmm0, %xmm1
+; X64-AVX-NEXT:    vpaddq %xmm0, %xmm1, %xmm0
 ; X64-AVX-NEXT:    retq
   %1 = mul <2 x i64> %a0, <i64 17, i64 17>
   ret <2 x i64> %1
 }
 
 define <4 x i32> @mul_v4i32_17(<4 x i32> %a0) nounwind {
 ; X86-LABEL: mul_v4i32_17:
 ; X86:       # %bb.0:

 ; X64-AVX-NEXT:    retq
   %1 = mul <8 x i16> %a0, <i16 17, i16 17, i16 17, i16 17, i16 17, i16 17, i16 17, i16 17>
   ret <8 x i16> %1
 }
 
 define <16 x i8> @mul_v16i8_17(<16 x i8> %a0) nounwind {
 ; X86-LABEL: mul_v16i8_17:
 ; X86:       # %bb.0:
-; X86-NEXT:    pmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; X86-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; X86-NEXT:    movdqa {{.*#+}} xmm2 = [17,17,17,17,17,17,17,17,17,17,17,17,17,17,17,17]
-; X86-NEXT:    pmullw %xmm2, %xmm0
-; X86-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; X86-NEXT:    pand %xmm3, %xmm0
-; X86-NEXT:    pmullw %xmm2, %xmm1
-; X86-NEXT:    pand %xmm3, %xmm1
-; X86-NEXT:    packuswb %xmm0, %xmm1
+; X86-NEXT:    movdqa %xmm0, %xmm1
+; X86-NEXT:    psllw $4, %xmm1
+; X86-NEXT:    pand {{\.LCPI.*}}, %xmm1
+; X86-NEXT:    paddb %xmm0, %xmm1
 ; X86-NEXT:    movdqa %xmm1, %xmm0
 ; X86-NEXT:    retl
 ;
 ; X64-LABEL: mul_v16i8_17:
 ; X64:       # %bb.0:
-; X64-NEXT:    pmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; X64-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; X64-NEXT:    movdqa {{.*#+}} xmm2 = [17,17,17,17,17,17,17,17,17,17,17,17,17,17,17,17]
-; X64-NEXT:    pmullw %xmm2, %xmm0
-; X64-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; X64-NEXT:    pand %xmm3, %xmm0
-; X64-NEXT:    pmullw %xmm2, %xmm1
-; X64-NEXT:    pand %xmm3, %xmm1
-; X64-NEXT:    packuswb %xmm0, %xmm1
+; X64-NEXT:    movdqa %xmm0, %xmm1
+; X64-NEXT:    psllw $4, %xmm1
+; X64-NEXT:    pand {{.*}}(%rip), %xmm1
+; X64-NEXT:    paddb %xmm0, %xmm1
 ; X64-NEXT:    movdqa %xmm1, %xmm0
 ; X64-NEXT:    retq
 ;
 ; X64-XOP-LABEL: mul_v16i8_17:
 ; X64-XOP:       # %bb.0:
-; X64-XOP-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; X64-XOP-NEXT:    vmovdqa {{.*#+}} xmm2 = [17,17,17,17,17,17,17,17,17,17,17,17,17,17,17,17]
-; X64-XOP-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
-; X64-XOP-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; X64-XOP-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; X64-XOP-NEXT:    vpperm {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14],xmm1[0,2,4,6,8,10,12,14]
+; X64-XOP-NEXT:    vpshlb {{.*}}(%rip), %xmm0, %xmm1
+; X64-XOP-NEXT:    vpaddb %xmm0, %xmm1, %xmm0
 ; X64-XOP-NEXT:    retq
 ;
 ; X64-AVX2-LABEL: mul_v16i8_17:
 ; X64-AVX2:       # %bb.0:
-; X64-AVX2-NEXT:    vpmovsxbw %xmm0, %ymm0
-; X64-AVX2-NEXT:    vpmullw {{.*}}(%rip), %ymm0, %ymm0
-; X64-AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; X64-AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; X64-AVX2-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; X64-AVX2-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; X64-AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; X64-AVX2-NEXT:    vzeroupper
+; X64-AVX2-NEXT:    vpsllw $4, %xmm0, %xmm1
+; X64-AVX2-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
+; X64-AVX2-NEXT:    vpaddb %xmm0, %xmm1, %xmm0
 ; X64-AVX2-NEXT:    retq
   %1 = mul <16 x i8> %a0, <i8 17, i8 17, i8 17, i8 17, i8 17, i8 17, i8 17, i8 17, i8 17, i8 17, i8 17, i8 17, i8 17, i8 17, i8 17, i8 17>
   ret <16 x i8> %1
 }
 
 ;
 ; PowOf2 + 1 (non-uniform)
 ;

 
 ;
 ; PowOf2 - 1 (uniform)
 ;
 
 define <2 x i64> @mul_v2i64_7(<2 x i64> %a0) nounwind {
 ; X86-LABEL: mul_v2i64_7:
 ; X86:       # %bb.0:
-; X86-NEXT:    movdqa {{.*#+}} xmm1 = [7,0,7,0]
-; X86-NEXT:    movdqa %xmm0, %xmm2
-; X86-NEXT:    pmuludq %xmm1, %xmm2
-; X86-NEXT:    psrlq $32, %xmm0
-; X86-NEXT:    pmuludq %xmm1, %xmm0
-; X86-NEXT:    psllq $32, %xmm0
-; X86-NEXT:    paddq %xmm2, %xmm0
+; X86-NEXT:    movdqa %xmm0, %xmm1
+; X86-NEXT:    psllq $3, %xmm1
+; X86-NEXT:    psubq %xmm0, %xmm1
+; X86-NEXT:    movdqa %xmm1, %xmm0
 ; X86-NEXT:    retl
 ;
 ; X64-LABEL: mul_v2i64_7:
 ; X64:       # %bb.0:
-; X64-NEXT:    movdqa {{.*#+}} xmm1 = [7,7]
-; X64-NEXT:    movdqa %xmm0, %xmm2
-; X64-NEXT:    pmuludq %xmm1, %xmm2
-; X64-NEXT:    psrlq $32, %xmm0
-; X64-NEXT:    pmuludq %xmm1, %xmm0
-; X64-NEXT:    psllq $32, %xmm0
-; X64-NEXT:    paddq %xmm2, %xmm0
+; X64-NEXT:    movdqa %xmm0, %xmm1
+; X64-NEXT:    psllq $3, %xmm1
+; X64-NEXT:    psubq %xmm0, %xmm1
+; X64-NEXT:    movdqa %xmm1, %xmm0
 ; X64-NEXT:    retq
 ;
 ; X64-AVX-LABEL: mul_v2i64_7:
 ; X64-AVX:       # %bb.0:
-; X64-AVX-NEXT:    vmovdqa {{.*#+}} xmm1 = [7,7]
-; X64-AVX-NEXT:    vpmuludq %xmm1, %xmm0, %xmm2
-; X64-AVX-NEXT:    vpsrlq $32, %xmm0, %xmm0
-; X64-AVX-NEXT:    vpmuludq %xmm1, %xmm0, %xmm0
-; X64-AVX-NEXT:    vpsllq $32, %xmm0, %xmm0
-; X64-AVX-NEXT:    vpaddq %xmm0, %xmm2, %xmm0
+; X64-AVX-NEXT:    vpsllq $3, %xmm0, %xmm1
+; X64-AVX-NEXT:    vpsubq %xmm0, %xmm1, %xmm0
 ; X64-AVX-NEXT:    retq
   %1 = mul <2 x i64> %a0, <i64 7, i64 7>
   ret <2 x i64> %1
 }
 
 define <4 x i32> @mul_v4i32_7(<4 x i32> %a0) nounwind {
 ; X86-LABEL: mul_v4i32_7:
 ; X86:       # %bb.0:

 ; X64-AVX-NEXT:    retq
   %1 = mul <8 x i16> %a0, <i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7, i16 7>
   ret <8 x i16> %1
 }
 
 define <16 x i8> @mul_v16i8_31(<16 x i8> %a0) nounwind {
 ; X86-LABEL: mul_v16i8_31:
 ; X86:       # %bb.0:
-; X86-NEXT:    pmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; X86-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; X86-NEXT:    movdqa {{.*#+}} xmm2 = [31,31,31,31,31,31,31,31,31,31,31,31,31,31,31,31]
-; X86-NEXT:    pmullw %xmm2, %xmm0
-; X86-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; X86-NEXT:    pand %xmm3, %xmm0
-; X86-NEXT:    pmullw %xmm2, %xmm1
-; X86-NEXT:    pand %xmm3, %xmm1
-; X86-NEXT:    packuswb %xmm0, %xmm1
+; X86-NEXT:    movdqa %xmm0, %xmm1
+; X86-NEXT:    psllw $5, %xmm1
+; X86-NEXT:    pand {{\.LCPI.*}}, %xmm1
+; X86-NEXT:    psubb %xmm0, %xmm1
 ; X86-NEXT:    movdqa %xmm1, %xmm0
 ; X86-NEXT:    retl
 ;
 ; X64-LABEL: mul_v16i8_31:
 ; X64:       # %bb.0:
-; X64-NEXT:    pmovzxbw {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; X64-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; X64-NEXT:    movdqa {{.*#+}} xmm2 = [31,31,31,31,31,31,31,31,31,31,31,31,31,31,31,31]
-; X64-NEXT:    pmullw %xmm2, %xmm0
-; X64-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; X64-NEXT:    pand %xmm3, %xmm0
-; X64-NEXT:    pmullw %xmm2, %xmm1
-; X64-NEXT:    pand %xmm3, %xmm1
-; X64-NEXT:    packuswb %xmm0, %xmm1
+; X64-NEXT:    movdqa %xmm0, %xmm1
+; X64-NEXT:    psllw $5, %xmm1
+; X64-NEXT:    pand {{.*}}(%rip), %xmm1
+; X64-NEXT:    psubb %xmm0, %xmm1
 ; X64-NEXT:    movdqa %xmm1, %xmm0
 ; X64-NEXT:    retq
 ;
 ; X64-XOP-LABEL: mul_v16i8_31:
 ; X64-XOP:       # %bb.0:
-; X64-XOP-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; X64-XOP-NEXT:    vmovdqa {{.*#+}} xmm2 = [31,31,31,31,31,31,31,31,31,31,31,31,31,31,31,31]
-; X64-XOP-NEXT:    vpmullw %xmm2, %xmm1, %xmm1
-; X64-XOP-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; X64-XOP-NEXT:    vpmullw %xmm2, %xmm0, %xmm0
-; X64-XOP-NEXT:    vpperm {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14],xmm1[0,2,4,6,8,10,12,14]
+; X64-XOP-NEXT:    vpshlb {{.*}}(%rip), %xmm0, %xmm1
+; X64-XOP-NEXT:    vpsubb %xmm0, %xmm1, %xmm0
 ; X64-XOP-NEXT:    retq
 ;
 ; X64-AVX2-LABEL: mul_v16i8_31:
 ; X64-AVX2:       # %bb.0:
-; X64-AVX2-NEXT:    vpmovsxbw %xmm0, %ymm0
-; X64-AVX2-NEXT:    vpmullw {{.*}}(%rip), %ymm0, %ymm0
-; X64-AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; X64-AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; X64-AVX2-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; X64-AVX2-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; X64-AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; X64-AVX2-NEXT:    vzeroupper
+; X64-AVX2-NEXT:    vpsllw $5, %xmm0, %xmm1
+; X64-AVX2-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm1
+; X64-AVX2-NEXT:    vpsubb %xmm0, %xmm1, %xmm0
 ; X64-AVX2-NEXT:    retq
   %1 = mul <16 x i8> %a0, <i8 31, i8 31, i8 31, i8 31, i8 31, i8 31, i8 31, i8 31, i8 31, i8 31, i8 31, i8 31, i8 31, i8 31, i8 31, i8 31>
   ret <16 x i8> %1
 }
 
 ;
 ; PowOf2 - 1 (non-uniform)
 ;