[llvm] 52cb0bb - [X86] LowerRotate - use vXi8 custom lowering for non-uniform constant amounts

Simon Pilgrim via llvm-commits llvm-commits at lists.llvm.org
Wed Dec 15 06:52:17 PST 2021


Author: Simon Pilgrim
Date: 2021-12-15T14:51:15Z
New Revision: 52cb0bbec34fc9320ff8ec1cdd617ea90e56a5e1

URL: https://github.com/llvm/llvm-project/commit/52cb0bbec34fc9320ff8ec1cdd617ea90e56a5e1
DIFF: https://github.com/llvm/llvm-project/commit/52cb0bbec34fc9320ff8ec1cdd617ea90e56a5e1.diff

LOG: [X86] LowerRotate - use vXi8 custom lowering for non-uniform constant amounts

Instead of bailing and using the default expansion, we can more efficiently use the shl(unpack(x,x),unpack(amt,zero)) pattern for vXi8 rotl, as we'll then use vXi16 fast PMULLW (or PSLLVW).

This required some minor changes to improve constant folding during unpack shuffle creation and convertShiftLeftToScale to support constants that have already been lowered to constant pools.

Added: 
    

Modified: 
    llvm/lib/Target/X86/X86ISelLowering.cpp
    llvm/test/CodeGen/X86/min-legal-vector-width.ll
    llvm/test/CodeGen/X86/srem-seteq-vec-nonsplat.ll
    llvm/test/CodeGen/X86/vector-fshl-rot-128.ll
    llvm/test/CodeGen/X86/vector-fshl-rot-256.ll
    llvm/test/CodeGen/X86/vector-fshl-rot-512.ll
    llvm/test/CodeGen/X86/vector-fshr-rot-128.ll
    llvm/test/CodeGen/X86/vector-fshr-rot-256.ll
    llvm/test/CodeGen/X86/vector-fshr-rot-512.ll
    llvm/test/CodeGen/X86/vector-rotate-128.ll
    llvm/test/CodeGen/X86/vector-rotate-256.ll
    llvm/test/CodeGen/X86/vector-rotate-512.ll

Removed: 
    


################################################################################
diff  --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index ef8783423c23c..8308129eac8c1 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -6782,12 +6782,33 @@ void llvm::createSplat2ShuffleMask(MVT VT, SmallVectorImpl<int> &Mask,
   }
 }
 
+// Attempt to constant fold, else just create a VECTOR_SHUFFLE.
+static SDValue getVectorShuffle(SelectionDAG &DAG, EVT VT, const SDLoc &dl,
+                                SDValue V1, SDValue V2, ArrayRef<int> Mask) {
+  if ((ISD::isBuildVectorOfConstantSDNodes(V1.getNode()) || V1.isUndef()) &&
+      (ISD::isBuildVectorOfConstantSDNodes(V2.getNode()) || V2.isUndef())) {
+    SmallVector<SDValue> Ops(Mask.size(), DAG.getUNDEF(VT.getScalarType()));
+    for (int I = 0, NumElts = Mask.size(); I != NumElts; ++I) {
+      int M = Mask[I];
+      if (M < 0)
+        continue;
+      SDValue V = (M < NumElts) ? V1 : V2;
+      if (V.isUndef())
+        continue;
+      Ops[I] = V.getOperand(M % NumElts);
+    }
+    return DAG.getBuildVector(VT, dl, Ops);
+  }
+
+  return DAG.getVectorShuffle(VT, dl, V1, V2, Mask);
+}
+
 /// Returns a vector_shuffle node for an unpackl operation.
 static SDValue getUnpackl(SelectionDAG &DAG, const SDLoc &dl, EVT VT,
                           SDValue V1, SDValue V2) {
   SmallVector<int, 8> Mask;
   createUnpackShuffleMask(VT, Mask, /* Lo = */ true, /* Unary = */ false);
-  return DAG.getVectorShuffle(VT, dl, V1, V2, Mask);
+  return getVectorShuffle(DAG, VT, dl, V1, V2, Mask);
 }
 
 /// Returns a vector_shuffle node for an unpackh operation.
@@ -6795,7 +6816,7 @@ static SDValue getUnpackh(SelectionDAG &DAG, const SDLoc &dl, EVT VT,
                           SDValue V1, SDValue V2) {
   SmallVector<int, 8> Mask;
   createUnpackShuffleMask(VT, Mask, /* Lo = */ false, /* Unary = */ false);
-  return DAG.getVectorShuffle(VT, dl, V1, V2, Mask);
+  return getVectorShuffle(DAG, VT, dl, V1, V2, Mask);
 }
 
 /// Returns a node that packs the LHS + RHS nodes together at half width.
@@ -29164,28 +29185,20 @@ static SDValue convertShiftLeftToScale(SDValue Amt, const SDLoc &dl,
         (Subtarget.hasBWI() && VT == MVT::v64i8)))
     return SDValue();
 
-  if (ISD::isBuildVectorOfConstantSDNodes(Amt.getNode())) {
-    SmallVector<SDValue, 8> Elts;
-    MVT SVT = VT.getVectorElementType();
-    unsigned SVTBits = SVT.getSizeInBits();
-    APInt One(SVTBits, 1);
-    unsigned NumElems = VT.getVectorNumElements();
-
-    for (unsigned i = 0; i != NumElems; ++i) {
-      SDValue Op = Amt->getOperand(i);
-      if (Op->isUndef()) {
-        Elts.push_back(Op);
-        continue;
-      }
+  MVT SVT = VT.getVectorElementType();
+  unsigned SVTBits = SVT.getSizeInBits();
+  unsigned NumElems = VT.getVectorNumElements();
 
-      ConstantSDNode *ND = cast<ConstantSDNode>(Op);
-      APInt C(SVTBits, ND->getZExtValue());
-      uint64_t ShAmt = C.getZExtValue();
-      if (ShAmt >= SVTBits) {
-        Elts.push_back(DAG.getUNDEF(SVT));
+  APInt UndefElts;
+  SmallVector<APInt> EltBits;
+  if (getTargetConstantBitsFromNode(Amt, SVTBits, UndefElts, EltBits)) {
+    APInt One(SVTBits, 1);
+    SmallVector<SDValue> Elts(NumElems, DAG.getUNDEF(SVT));
+    for (unsigned I = 0; I != NumElems; ++I) {
+      if (UndefElts[I] || EltBits[I].uge(SVTBits))
         continue;
-      }
-      Elts.push_back(DAG.getConstant(One.shl(ShAmt), dl, SVT));
+      uint64_t ShAmt = EltBits[I].getZExtValue();
+      Elts[I] = DAG.getConstant(One.shl(ShAmt), dl, SVT);
     }
     return DAG.getBuildVector(VT, dl, Elts);
   }
@@ -29861,9 +29874,7 @@ static SDValue LowerRotate(SDValue Op, const X86Subtarget &Subtarget,
   // the amount bit.
   // TODO: We're doing nothing here that we couldn't do for funnel shifts.
   if (EltSizeInBits == 8) {
-    if (ISD::isBuildVectorOfConstantSDNodes(Amt.getNode()))
-      return SDValue();
-
+    bool IsConstAmt = ISD::isBuildVectorOfConstantSDNodes(Amt.getNode());
     // Check for a hidden ISD::ROTR, vXi8 lowering can handle both, but we
     // currently hit infinite loops in legalization if we allow ISD::ROTR.
     // FIXME: Infinite ROTL<->ROTR legalization in TargetLowering::expandROT.
@@ -29894,6 +29905,9 @@ static SDValue LowerRotate(SDValue Op, const X86Subtarget &Subtarget,
       return getPack(DAG, Subtarget, DL, VT, Lo, Hi, !HiddenROTRAmt);
     } else if (supportedVectorVarShift(WideVT, Subtarget, ShiftOpc) &&
                supportedVectorShiftWithImm(WideVT, Subtarget, ShiftOpc)) {
+      // If we're rotating by constant, just use default promotion.
+      if (IsConstAmt)
+        return SDValue();
       // See if we can perform this by widening to vXi16 or vXi32.
       R = DAG.getNode(ISD::ZERO_EXTEND, DL, WideVT, R);
       R = DAG.getNode(
@@ -29904,9 +29918,10 @@ static SDValue LowerRotate(SDValue Op, const X86Subtarget &Subtarget,
       if (!HiddenROTRAmt)
         R = getTargetVShiftByConstNode(X86ISD::VSRLI, DL, WideVT, R, 8, DAG);
       return DAG.getNode(ISD::TRUNCATE, DL, VT, R);
-    } else if (supportedVectorVarShift(ExtVT, Subtarget, ShiftOpc)) {
+    } else if (IsConstAmt ||
+               supportedVectorVarShift(ExtVT, Subtarget, ShiftOpc)) {
       // See if we can perform this by unpacking to lo/hi vXi16.
-      SDValue Z = getZeroVector(VT, Subtarget, DAG, DL);
+      SDValue Z = DAG.getConstant(0, DL, VT);
       SDValue RLo = DAG.getBitcast(ExtVT, getUnpackl(DAG, DL, VT, R, R));
       SDValue RHi = DAG.getBitcast(ExtVT, getUnpackh(DAG, DL, VT, R, R));
       SDValue ALo = DAG.getBitcast(ExtVT, getUnpackl(DAG, DL, VT, AmtMod, Z));

diff  --git a/llvm/test/CodeGen/X86/min-legal-vector-width.ll b/llvm/test/CodeGen/X86/min-legal-vector-width.ll
index 3493a5433310d..a6ccd731f1867 100644
--- a/llvm/test/CodeGen/X86/min-legal-vector-width.ll
+++ b/llvm/test/CodeGen/X86/min-legal-vector-width.ll
@@ -1716,45 +1716,16 @@ define <32 x i8> @splatvar_rotate_v32i8(<32 x i8> %a, <32 x i8> %b) nounwind "mi
 }
 
 define <32 x i8> @constant_rotate_v32i8(<32 x i8> %a) nounwind "min-legal-vector-width"="256" {
-; CHECK-AVX512-LABEL: constant_rotate_v32i8:
-; CHECK-AVX512:       # %bb.0:
-; CHECK-AVX512-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; CHECK-AVX512-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31]
-; CHECK-AVX512-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; CHECK-AVX512-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; CHECK-AVX512-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23]
-; CHECK-AVX512-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm1, %ymm1
-; CHECK-AVX512-NEXT:    vpsrlw $8, %ymm1, %ymm1
-; CHECK-AVX512-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
-; CHECK-AVX512-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; CHECK-AVX512-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; CHECK-AVX512-NEXT:    vmovdqa {{.*#+}} ymm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; CHECK-AVX512-NEXT:    vpand %ymm3, %ymm2, %ymm2
-; CHECK-AVX512-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
-; CHECK-AVX512-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; CHECK-AVX512-NEXT:    vpand %ymm3, %ymm0, %ymm0
-; CHECK-AVX512-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
-; CHECK-AVX512-NEXT:    vpor %ymm1, %ymm0, %ymm0
-; CHECK-AVX512-NEXT:    retq
-;
-; CHECK-VBMI-LABEL: constant_rotate_v32i8:
-; CHECK-VBMI:       # %bb.0:
-; CHECK-VBMI-NEXT:    vpunpckhbw {{.*#+}} ymm1 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; CHECK-VBMI-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm1, %ymm1
-; CHECK-VBMI-NEXT:    vpunpcklbw {{.*#+}} ymm2 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
-; CHECK-VBMI-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; CHECK-VBMI-NEXT:    vmovdqa {{.*#+}} ymm3 = [0,2,4,6,8,10,12,14,32,34,36,38,40,42,44,46,16,18,20,22,24,26,28,30,48,50,52,54,56,58,60,62]
-; CHECK-VBMI-NEXT:    vpermi2b %ymm1, %ymm2, %ymm3
-; CHECK-VBMI-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; CHECK-VBMI-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31]
-; CHECK-VBMI-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; CHECK-VBMI-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; CHECK-VBMI-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23]
-; CHECK-VBMI-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; CHECK-VBMI-NEXT:    vpsrlw $8, %ymm0, %ymm0
-; CHECK-VBMI-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
-; CHECK-VBMI-NEXT:    vpor %ymm0, %ymm3, %ymm0
-; CHECK-VBMI-NEXT:    retq
+; CHECK-LABEL: constant_rotate_v32i8:
+; CHECK:       # %bb.0:
+; CHECK-NEXT:    vpunpckhbw {{.*#+}} ymm1 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
+; CHECK-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm1, %ymm1
+; CHECK-NEXT:    vpsrlw $8, %ymm1, %ymm1
+; CHECK-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
+; CHECK-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
+; CHECK-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; CHECK-NEXT:    vpackuswb %ymm1, %ymm0, %ymm0
+; CHECK-NEXT:    retq
   %shl = shl <32 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1>
   %lshr = lshr <32 x i8> %a, <i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7>
   %or = or <32 x i8> %shl, %lshr

diff  --git a/llvm/test/CodeGen/X86/srem-seteq-vec-nonsplat.ll b/llvm/test/CodeGen/X86/srem-seteq-vec-nonsplat.ll
index c0d9287ba16ed..a80931bfaa836 100644
--- a/llvm/test/CodeGen/X86/srem-seteq-vec-nonsplat.ll
+++ b/llvm/test/CodeGen/X86/srem-seteq-vec-nonsplat.ll
@@ -2290,78 +2290,58 @@ define <32 x i1> @pr51133(<32 x i8> %x, <32 x i8> %y) {
 ; CHECK-SSE2-LABEL: pr51133:
 ; CHECK-SSE2:       # %bb.0:
 ; CHECK-SSE2-NEXT:    movq %rdi, %rax
-; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm4
-; CHECK-SSE2-NEXT:    punpckhbw {{.*#+}} xmm4 = xmm4[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; CHECK-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm4
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm5 = [255,255,255,255,255,255,255,255]
-; CHECK-SSE2-NEXT:    pand %xmm5, %xmm4
+; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm5
+; CHECK-SSE2-NEXT:    punpckhbw {{.*#+}} xmm5 = xmm5[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; CHECK-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm5
+; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [255,255,255,255,255,255,255,255]
+; CHECK-SSE2-NEXT:    pand %xmm4, %xmm5
 ; CHECK-SSE2-NEXT:    movdqa %xmm1, %xmm6
 ; CHECK-SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; CHECK-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm6
-; CHECK-SSE2-NEXT:    pand %xmm5, %xmm6
-; CHECK-SSE2-NEXT:    packuswb %xmm4, %xmm6
+; CHECK-SSE2-NEXT:    pand %xmm4, %xmm6
+; CHECK-SSE2-NEXT:    packuswb %xmm5, %xmm6
 ; CHECK-SSE2-NEXT:    paddb {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm6
-; CHECK-SSE2-NEXT:    pxor %xmm8, %xmm8
-; CHECK-SSE2-NEXT:    movdqa %xmm6, %xmm7
-; CHECK-SSE2-NEXT:    punpckhbw {{.*#+}} xmm7 = xmm7[8],xmm8[8],xmm7[9],xmm8[9],xmm7[10],xmm8[10],xmm7[11],xmm8[11],xmm7[12],xmm8[12],xmm7[13],xmm8[13],xmm7[14],xmm8[14],xmm7[15],xmm8[15]
-; CHECK-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm7
-; CHECK-SSE2-NEXT:    psrlw $8, %xmm7
-; CHECK-SSE2-NEXT:    movdqa %xmm6, %xmm4
-; CHECK-SSE2-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm8[0],xmm4[1],xmm8[1],xmm4[2],xmm8[2],xmm4[3],xmm8[3],xmm4[4],xmm8[4],xmm4[5],xmm8[5],xmm4[6],xmm8[6],xmm4[7],xmm8[7]
-; CHECK-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm4
-; CHECK-SSE2-NEXT:    psrlw $8, %xmm4
-; CHECK-SSE2-NEXT:    packuswb %xmm7, %xmm4
-; CHECK-SSE2-NEXT:    movdqa %xmm6, %xmm7
-; CHECK-SSE2-NEXT:    punpckhbw {{.*#+}} xmm7 = xmm7[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; CHECK-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm7
-; CHECK-SSE2-NEXT:    pand %xmm5, %xmm7
+; CHECK-SSE2-NEXT:    movdqa %xmm6, %xmm5
+; CHECK-SSE2-NEXT:    punpckhbw {{.*#+}} xmm5 = xmm5[8],xmm6[8],xmm5[9],xmm6[9],xmm5[10],xmm6[10],xmm5[11],xmm6[11],xmm5[12],xmm6[12],xmm5[13],xmm6[13],xmm5[14],xmm6[14],xmm5[15],xmm6[15]
+; CHECK-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm5
+; CHECK-SSE2-NEXT:    psrlw $8, %xmm5
 ; CHECK-SSE2-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; CHECK-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm6
-; CHECK-SSE2-NEXT:    pand %xmm5, %xmm6
-; CHECK-SSE2-NEXT:    packuswb %xmm7, %xmm6
-; CHECK-SSE2-NEXT:    por %xmm4, %xmm6
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [84,2,36,42,2,0,2,4,2,255,4,36,126,30,2,2]
-; CHECK-SSE2-NEXT:    pminub %xmm6, %xmm4
-; CHECK-SSE2-NEXT:    pcmpeqb %xmm6, %xmm4
-; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm6 = [255,255,255,255,255,0,255,255,255,255,255,255,255,255,255,255]
-; CHECK-SSE2-NEXT:    pandn %xmm6, %xmm4
+; CHECK-SSE2-NEXT:    psrlw $8, %xmm6
+; CHECK-SSE2-NEXT:    packuswb %xmm5, %xmm6
+; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm7 = [84,2,36,42,2,0,2,4,2,255,4,36,126,30,2,2]
+; CHECK-SSE2-NEXT:    pminub %xmm6, %xmm7
+; CHECK-SSE2-NEXT:    pcmpeqb %xmm6, %xmm7
+; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm5 = [255,255,255,255,255,0,255,255,255,255,255,255,255,255,255,255]
+; CHECK-SSE2-NEXT:    pandn %xmm5, %xmm7
 ; CHECK-SSE2-NEXT:    pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
-; CHECK-SSE2-NEXT:    pcmpgtb %xmm8, %xmm1
-; CHECK-SSE2-NEXT:    pandn %xmm1, %xmm6
-; CHECK-SSE2-NEXT:    por %xmm4, %xmm6
+; CHECK-SSE2-NEXT:    pxor %xmm6, %xmm6
+; CHECK-SSE2-NEXT:    pcmpgtb %xmm6, %xmm1
+; CHECK-SSE2-NEXT:    pandn %xmm1, %xmm5
+; CHECK-SSE2-NEXT:    por %xmm7, %xmm5
 ; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm1
 ; CHECK-SSE2-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
 ; CHECK-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
-; CHECK-SSE2-NEXT:    pand %xmm5, %xmm1
+; CHECK-SSE2-NEXT:    pand %xmm4, %xmm1
 ; CHECK-SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; CHECK-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; CHECK-SSE2-NEXT:    pand %xmm5, %xmm0
+; CHECK-SSE2-NEXT:    pand %xmm4, %xmm0
 ; CHECK-SSE2-NEXT:    packuswb %xmm1, %xmm0
 ; CHECK-SSE2-NEXT:    paddb {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
 ; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm1
-; CHECK-SSE2-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm8[8],xmm1[9],xmm8[9],xmm1[10],xmm8[10],xmm1[11],xmm8[11],xmm1[12],xmm8[12],xmm1[13],xmm8[13],xmm1[14],xmm8[14],xmm1[15],xmm8[15]
+; CHECK-SSE2-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; CHECK-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
 ; CHECK-SSE2-NEXT:    psrlw $8, %xmm1
-; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm4
-; CHECK-SSE2-NEXT:    punpcklbw {{.*#+}} xmm4 = xmm4[0],xmm8[0],xmm4[1],xmm8[1],xmm4[2],xmm8[2],xmm4[3],xmm8[3],xmm4[4],xmm8[4],xmm4[5],xmm8[5],xmm4[6],xmm8[6],xmm4[7],xmm8[7]
-; CHECK-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm4
-; CHECK-SSE2-NEXT:    psrlw $8, %xmm4
-; CHECK-SSE2-NEXT:    packuswb %xmm1, %xmm4
-; CHECK-SSE2-NEXT:    movdqa %xmm0, %xmm1
-; CHECK-SSE2-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; CHECK-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
-; CHECK-SSE2-NEXT:    pand %xmm5, %xmm1
 ; CHECK-SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; CHECK-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; CHECK-SSE2-NEXT:    pand %xmm5, %xmm0
+; CHECK-SSE2-NEXT:    psrlw $8, %xmm0
 ; CHECK-SSE2-NEXT:    packuswb %xmm1, %xmm0
-; CHECK-SSE2-NEXT:    por %xmm4, %xmm0
 ; CHECK-SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [19,51,13,7,127,31,127,3,5,5,51,37,3,127,85,5]
 ; CHECK-SSE2-NEXT:    pmaxub %xmm0, %xmm1
 ; CHECK-SSE2-NEXT:    pcmpeqb %xmm0, %xmm1
-; CHECK-SSE2-NEXT:    pcmpeqb %xmm8, %xmm3
-; CHECK-SSE2-NEXT:    pandn %xmm6, %xmm3
-; CHECK-SSE2-NEXT:    pcmpeqb %xmm8, %xmm2
+; CHECK-SSE2-NEXT:    pcmpeqb %xmm6, %xmm3
+; CHECK-SSE2-NEXT:    pandn %xmm5, %xmm3
+; CHECK-SSE2-NEXT:    pcmpeqb %xmm6, %xmm2
 ; CHECK-SSE2-NEXT:    pandn %xmm1, %xmm2
 ; CHECK-SSE2-NEXT:    pmovmskb %xmm2, %ecx
 ; CHECK-SSE2-NEXT:    pmovmskb %xmm3, %edx
@@ -2374,73 +2354,57 @@ define <32 x i1> @pr51133(<32 x i8> %x, <32 x i8> %y) {
 ; CHECK-SSE41:       # %bb.0:
 ; CHECK-SSE41-NEXT:    movdqa %xmm0, %xmm4
 ; CHECK-SSE41-NEXT:    movq %rdi, %rax
-; CHECK-SSE41-NEXT:    movdqa %xmm1, %xmm5
-; CHECK-SSE41-NEXT:    punpckhbw {{.*#+}} xmm5 = xmm5[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; CHECK-SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm5
-; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm9 = [255,255,255,255,255,255,255,255]
-; CHECK-SSE41-NEXT:    pand %xmm9, %xmm5
-; CHECK-SSE41-NEXT:    pmovzxbw {{.*#+}} xmm0 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
-; CHECK-SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; CHECK-SSE41-NEXT:    pand %xmm9, %xmm0
-; CHECK-SSE41-NEXT:    packuswb %xmm5, %xmm0
-; CHECK-SSE41-NEXT:    paddb {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; CHECK-SSE41-NEXT:    pxor %xmm8, %xmm8
-; CHECK-SSE41-NEXT:    movdqa %xmm0, %xmm7
-; CHECK-SSE41-NEXT:    punpckhbw {{.*#+}} xmm7 = xmm7[8],xmm8[8],xmm7[9],xmm8[9],xmm7[10],xmm8[10],xmm7[11],xmm8[11],xmm7[12],xmm8[12],xmm7[13],xmm8[13],xmm7[14],xmm8[14],xmm7[15],xmm8[15]
-; CHECK-SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm7
-; CHECK-SSE41-NEXT:    psrlw $8, %xmm7
-; CHECK-SSE41-NEXT:    pmovzxbw {{.*#+}} xmm5 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm6 = [256,256,256,128,64,2,256,32]
-; CHECK-SSE41-NEXT:    pmullw %xmm5, %xmm6
-; CHECK-SSE41-NEXT:    psrlw $8, %xmm6
-; CHECK-SSE41-NEXT:    packuswb %xmm7, %xmm6
+; CHECK-SSE41-NEXT:    movdqa %xmm1, %xmm0
 ; CHECK-SSE41-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
 ; CHECK-SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; CHECK-SSE41-NEXT:    pand %xmm9, %xmm0
-; CHECK-SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm5
-; CHECK-SSE41-NEXT:    pand %xmm9, %xmm5
-; CHECK-SSE41-NEXT:    packuswb %xmm0, %xmm5
-; CHECK-SSE41-NEXT:    por %xmm6, %xmm5
+; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm5 = [255,255,255,255,255,255,255,255]
+; CHECK-SSE41-NEXT:    pand %xmm5, %xmm0
+; CHECK-SSE41-NEXT:    pmovzxbw {{.*#+}} xmm6 = xmm1[0],zero,xmm1[1],zero,xmm1[2],zero,xmm1[3],zero,xmm1[4],zero,xmm1[5],zero,xmm1[6],zero,xmm1[7],zero
+; CHECK-SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm6
+; CHECK-SSE41-NEXT:    pand %xmm5, %xmm6
+; CHECK-SSE41-NEXT:    packuswb %xmm0, %xmm6
+; CHECK-SSE41-NEXT:    paddb {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm6
+; CHECK-SSE41-NEXT:    movdqa %xmm6, %xmm0
+; CHECK-SSE41-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm6[8],xmm0[9],xmm6[9],xmm0[10],xmm6[10],xmm0[11],xmm6[11],xmm0[12],xmm6[12],xmm0[13],xmm6[13],xmm0[14],xmm6[14],xmm0[15],xmm6[15]
+; CHECK-SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; CHECK-SSE41-NEXT:    psrlw $8, %xmm0
+; CHECK-SSE41-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; CHECK-SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm6
+; CHECK-SSE41-NEXT:    psrlw $8, %xmm6
+; CHECK-SSE41-NEXT:    packuswb %xmm0, %xmm6
 ; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [84,2,36,42,2,0,2,4,2,255,4,36,126,30,2,2]
-; CHECK-SSE41-NEXT:    pminub %xmm5, %xmm0
-; CHECK-SSE41-NEXT:    pcmpeqb %xmm5, %xmm0
-; CHECK-SSE41-NEXT:    pcmpeqd %xmm5, %xmm5
-; CHECK-SSE41-NEXT:    pxor %xmm0, %xmm5
+; CHECK-SSE41-NEXT:    pminub %xmm6, %xmm0
+; CHECK-SSE41-NEXT:    pcmpeqb %xmm6, %xmm0
+; CHECK-SSE41-NEXT:    pcmpeqd %xmm7, %xmm7
+; CHECK-SSE41-NEXT:    pxor %xmm0, %xmm7
 ; CHECK-SSE41-NEXT:    pand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
-; CHECK-SSE41-NEXT:    pcmpgtb %xmm8, %xmm1
+; CHECK-SSE41-NEXT:    pxor %xmm6, %xmm6
+; CHECK-SSE41-NEXT:    pcmpgtb %xmm6, %xmm1
 ; CHECK-SSE41-NEXT:    movaps {{.*#+}} xmm0 = [255,255,255,255,255,0,255,255,255,255,255,255,255,255,255,255]
-; CHECK-SSE41-NEXT:    pblendvb %xmm0, %xmm5, %xmm1
+; CHECK-SSE41-NEXT:    pblendvb %xmm0, %xmm7, %xmm1
 ; CHECK-SSE41-NEXT:    pmovzxbw {{.*#+}} xmm0 = xmm4[0],zero,xmm4[1],zero,xmm4[2],zero,xmm4[3],zero,xmm4[4],zero,xmm4[5],zero,xmm4[6],zero,xmm4[7],zero
 ; CHECK-SSE41-NEXT:    punpckhbw {{.*#+}} xmm4 = xmm4[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
 ; CHECK-SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm4
-; CHECK-SSE41-NEXT:    pand %xmm9, %xmm4
+; CHECK-SSE41-NEXT:    pand %xmm5, %xmm4
 ; CHECK-SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; CHECK-SSE41-NEXT:    pand %xmm9, %xmm0
+; CHECK-SSE41-NEXT:    pand %xmm5, %xmm0
 ; CHECK-SSE41-NEXT:    packuswb %xmm4, %xmm0
 ; CHECK-SSE41-NEXT:    paddb {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
 ; CHECK-SSE41-NEXT:    movdqa %xmm0, %xmm4
-; CHECK-SSE41-NEXT:    punpckhbw {{.*#+}} xmm4 = xmm4[8],xmm8[8],xmm4[9],xmm8[9],xmm4[10],xmm8[10],xmm4[11],xmm8[11],xmm4[12],xmm8[12],xmm4[13],xmm8[13],xmm4[14],xmm8[14],xmm4[15],xmm8[15]
+; CHECK-SSE41-NEXT:    punpckhbw {{.*#+}} xmm4 = xmm4[8],xmm0[8],xmm4[9],xmm0[9],xmm4[10],xmm0[10],xmm4[11],xmm0[11],xmm4[12],xmm0[12],xmm4[13],xmm0[13],xmm4[14],xmm0[14],xmm4[15],xmm0[15]
 ; CHECK-SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm4
 ; CHECK-SSE41-NEXT:    psrlw $8, %xmm4
-; CHECK-SSE41-NEXT:    pmovzxbw {{.*#+}} xmm5 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
-; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm6 = [256,256,256,128,128,32,128,32]
-; CHECK-SSE41-NEXT:    pmullw %xmm5, %xmm6
-; CHECK-SSE41-NEXT:    psrlw $8, %xmm6
-; CHECK-SSE41-NEXT:    packuswb %xmm4, %xmm6
-; CHECK-SSE41-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; CHECK-SSE41-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; CHECK-SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; CHECK-SSE41-NEXT:    pand %xmm9, %xmm0
-; CHECK-SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm5
-; CHECK-SSE41-NEXT:    pand %xmm9, %xmm5
-; CHECK-SSE41-NEXT:    packuswb %xmm0, %xmm5
-; CHECK-SSE41-NEXT:    por %xmm6, %xmm5
-; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm0 = [19,51,13,7,127,31,127,3,5,5,51,37,3,127,85,5]
-; CHECK-SSE41-NEXT:    pmaxub %xmm5, %xmm0
-; CHECK-SSE41-NEXT:    pcmpeqb %xmm5, %xmm0
-; CHECK-SSE41-NEXT:    pcmpeqb %xmm8, %xmm3
+; CHECK-SSE41-NEXT:    psrlw $8, %xmm0
+; CHECK-SSE41-NEXT:    packuswb %xmm4, %xmm0
+; CHECK-SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [19,51,13,7,127,31,127,3,5,5,51,37,3,127,85,5]
+; CHECK-SSE41-NEXT:    pmaxub %xmm0, %xmm4
+; CHECK-SSE41-NEXT:    pcmpeqb %xmm0, %xmm4
+; CHECK-SSE41-NEXT:    pcmpeqb %xmm6, %xmm3
 ; CHECK-SSE41-NEXT:    pandn %xmm1, %xmm3
-; CHECK-SSE41-NEXT:    pcmpeqb %xmm8, %xmm2
-; CHECK-SSE41-NEXT:    pandn %xmm0, %xmm2
+; CHECK-SSE41-NEXT:    pcmpeqb %xmm6, %xmm2
+; CHECK-SSE41-NEXT:    pandn %xmm4, %xmm2
 ; CHECK-SSE41-NEXT:    pmovmskb %xmm2, %ecx
 ; CHECK-SSE41-NEXT:    pmovmskb %xmm3, %edx
 ; CHECK-SSE41-NEXT:    shll $16, %edx
@@ -2583,30 +2547,22 @@ define <32 x i1> @pr51133(<32 x i8> %x, <32 x i8> %y) {
 ; CHECK-AVX512VL-NEXT:    vpand %ymm3, %ymm2, %ymm2
 ; CHECK-AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm4 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
 ; CHECK-AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm4, %ymm4
-; CHECK-AVX512VL-NEXT:    vpand %ymm3, %ymm4, %ymm4
-; CHECK-AVX512VL-NEXT:    vpackuswb %ymm2, %ymm4, %ymm2
+; CHECK-AVX512VL-NEXT:    vpand %ymm3, %ymm4, %ymm3
+; CHECK-AVX512VL-NEXT:    vpackuswb %ymm2, %ymm3, %ymm2
 ; CHECK-AVX512VL-NEXT:    vpaddb {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; CHECK-AVX512VL-NEXT:    vpxor %xmm4, %xmm4, %xmm4
-; CHECK-AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm5 = ymm2[8],ymm4[8],ymm2[9],ymm4[9],ymm2[10],ymm4[10],ymm2[11],ymm4[11],ymm2[12],ymm4[12],ymm2[13],ymm4[13],ymm2[14],ymm4[14],ymm2[15],ymm4[15],ymm2[24],ymm4[24],ymm2[25],ymm4[25],ymm2[26],ymm4[26],ymm2[27],ymm4[27],ymm2[28],ymm4[28],ymm2[29],ymm4[29],ymm2[30],ymm4[30],ymm2[31],ymm4[31]
-; CHECK-AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm5, %ymm5
-; CHECK-AVX512VL-NEXT:    vpsrlw $8, %ymm5, %ymm5
-; CHECK-AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm6 = ymm2[0],ymm4[0],ymm2[1],ymm4[1],ymm2[2],ymm4[2],ymm2[3],ymm4[3],ymm2[4],ymm4[4],ymm2[5],ymm4[5],ymm2[6],ymm4[6],ymm2[7],ymm4[7],ymm2[16],ymm4[16],ymm2[17],ymm4[17],ymm2[18],ymm4[18],ymm2[19],ymm4[19],ymm2[20],ymm4[20],ymm2[21],ymm4[21],ymm2[22],ymm4[22],ymm2[23],ymm4[23]
-; CHECK-AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm6, %ymm6
-; CHECK-AVX512VL-NEXT:    vpsrlw $8, %ymm6, %ymm6
-; CHECK-AVX512VL-NEXT:    vpackuswb %ymm5, %ymm6, %ymm5
-; CHECK-AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm6 = ymm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; CHECK-AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm6, %ymm6
-; CHECK-AVX512VL-NEXT:    vpand %ymm3, %ymm6, %ymm6
+; CHECK-AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
+; CHECK-AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm3, %ymm3
+; CHECK-AVX512VL-NEXT:    vpsrlw $8, %ymm3, %ymm3
 ; CHECK-AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm2 = ymm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
 ; CHECK-AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; CHECK-AVX512VL-NEXT:    vpand %ymm3, %ymm2, %ymm2
-; CHECK-AVX512VL-NEXT:    vpackuswb %ymm6, %ymm2, %ymm2
-; CHECK-AVX512VL-NEXT:    vpor %ymm5, %ymm2, %ymm2
+; CHECK-AVX512VL-NEXT:    vpsrlw $8, %ymm2, %ymm2
+; CHECK-AVX512VL-NEXT:    vpackuswb %ymm3, %ymm2, %ymm2
 ; CHECK-AVX512VL-NEXT:    vpminub {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm3
 ; CHECK-AVX512VL-NEXT:    vpcmpeqb %ymm3, %ymm2, %ymm2
 ; CHECK-AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,255,255,255,255,255,255,255,255,255,255]
 ; CHECK-AVX512VL-NEXT:    vpandn %ymm3, %ymm2, %ymm2
 ; CHECK-AVX512VL-NEXT:    vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
+; CHECK-AVX512VL-NEXT:    vpxor %xmm4, %xmm4, %xmm4
 ; CHECK-AVX512VL-NEXT:    vpcmpgtb %ymm4, %ymm0, %ymm0
 ; CHECK-AVX512VL-NEXT:    vpandn %ymm0, %ymm3, %ymm3
 ; CHECK-AVX512VL-NEXT:    vpcmpeqb %ymm4, %ymm1, %ymm0

diff  --git a/llvm/test/CodeGen/X86/vector-fshl-rot-128.ll b/llvm/test/CodeGen/X86/vector-fshl-rot-128.ll
index 595e50b0e4446..c0b7423051a68 100644
--- a/llvm/test/CodeGen/X86/vector-fshl-rot-128.ll
+++ b/llvm/test/CodeGen/X86/vector-fshl-rot-128.ll
@@ -1542,87 +1542,28 @@ define <8 x i16> @constant_funnnel_v8i16(<8 x i16> %x) nounwind {
 }
 
 define <16 x i8> @constant_funnnel_v16i8(<16 x i8> %x) nounwind {
-; SSE2-LABEL: constant_funnnel_v16i8:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    pxor %xmm1, %xmm1
-; SSE2-NEXT:    movdqa %xmm0, %xmm2
-; SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm1[8],xmm2[9],xmm1[9],xmm2[10],xmm1[10],xmm2[11],xmm1[11],xmm2[12],xmm1[12],xmm2[13],xmm1[13],xmm2[14],xmm1[14],xmm2[15],xmm1[15]
-; SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
-; SSE2-NEXT:    psrlw $8, %xmm2
-; SSE2-NEXT:    movdqa %xmm0, %xmm3
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3],xmm3[4],xmm1[4],xmm3[5],xmm1[5],xmm3[6],xmm1[6],xmm3[7],xmm1[7]
-; SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm3
-; SSE2-NEXT:    psrlw $8, %xmm3
-; SSE2-NEXT:    packuswb %xmm2, %xmm3
-; SSE2-NEXT:    movdqa %xmm0, %xmm1
-; SSE2-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
-; SSE2-NEXT:    pand %xmm2, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; SSE2-NEXT:    pand %xmm2, %xmm0
-; SSE2-NEXT:    packuswb %xmm1, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: constant_funnnel_v16i8:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    movdqa %xmm0, %xmm2
-; SSE41-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
-; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; SSE41-NEXT:    pand %xmm3, %xmm2
-; 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 {{.*#+}} xmm4 = [1,2,4,8,16,32,64,128]
-; SSE41-NEXT:    pmullw %xmm1, %xmm4
-; SSE41-NEXT:    pand %xmm3, %xmm4
-; SSE41-NEXT:    packuswb %xmm2, %xmm4
-; SSE41-NEXT:    pxor %xmm2, %xmm2
-; SSE41-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm2[8],xmm0[9],xmm2[9],xmm0[10],xmm2[10],xmm0[11],xmm2[11],xmm0[12],xmm2[12],xmm0[13],xmm2[13],xmm0[14],xmm2[14],xmm0[15],xmm2[15]
-; SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; SSE41-NEXT:    psrlw $8, %xmm0
-; SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
-; SSE41-NEXT:    psrlw $8, %xmm1
-; SSE41-NEXT:    packuswb %xmm0, %xmm1
-; SSE41-NEXT:    por %xmm4, %xmm1
-; SSE41-NEXT:    movdqa %xmm1, %xmm0
-; SSE41-NEXT:    retq
-;
-; AVX1-LABEL: constant_funnnel_v16i8:
-; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpand %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = 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 {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm3, %xmm4
-; AVX1-NEXT:    vpand %xmm2, %xmm4, %xmm2
-; AVX1-NEXT:    vpackuswb %xmm1, %xmm2, %xmm1
-; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm0 = xmm0[8],xmm2[8],xmm0[9],xmm2[9],xmm0[10],xmm2[10],xmm0[11],xmm2[11],xmm0[12],xmm2[12],xmm0[13],xmm2[13],xmm0[14],xmm2[14],xmm0[15],xmm2[15]
-; AVX1-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
-; AVX1-NEXT:    vpsrlw $8, %xmm0, %xmm0
-; AVX1-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm3, %xmm2
-; AVX1-NEXT:    vpsrlw $8, %xmm2, %xmm2
-; AVX1-NEXT:    vpackuswb %xmm0, %xmm2, %xmm0
-; AVX1-NEXT:    vpor %xmm0, %xmm1, %xmm0
-; AVX1-NEXT:    retq
+; SSE-LABEL: constant_funnnel_v16i8:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movdqa %xmm0, %xmm1
+; SSE-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
+; SSE-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
+; SSE-NEXT:    psrlw $8, %xmm1
+; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE-NEXT:    psrlw $8, %xmm0
+; SSE-NEXT:    packuswb %xmm1, %xmm0
+; SSE-NEXT:    retq
 ;
-; AVX2-LABEL: constant_funnnel_v16i8:
-; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} ymm0 = 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
-; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm1
-; AVX2-NEXT:    vpsrlw $8, %ymm1, %ymm1
-; AVX2-NEXT:    vextracti128 $1, %ymm1, %xmm2
-; AVX2-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
-; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm2
-; AVX2-NEXT:    vpackuswb %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vzeroupper
-; AVX2-NEXT:    retq
+; AVX-LABEL: constant_funnnel_v16i8:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
+; AVX-NEXT:    vpsrlw $8, %xmm1, %xmm1
+; AVX-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; AVX-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
+; AVX-NEXT:    vpsrlw $8, %xmm0, %xmm0
+; AVX-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    retq
 ;
 ; AVX512F-LABEL: constant_funnnel_v16i8:
 ; AVX512F:       # %bb.0:
@@ -1697,26 +1638,14 @@ define <16 x i8> @constant_funnnel_v16i8(<16 x i8> %x) nounwind {
 ;
 ; X86-SSE2-LABEL: constant_funnnel_v16i8:
 ; X86-SSE2:       # %bb.0:
-; X86-SSE2-NEXT:    pxor %xmm2, %xmm2
-; X86-SSE2-NEXT:    movdqa %xmm0, %xmm3
-; X86-SSE2-NEXT:    punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm2[8],xmm3[9],xmm2[9],xmm3[10],xmm2[10],xmm3[11],xmm2[11],xmm3[12],xmm2[12],xmm3[13],xmm2[13],xmm3[14],xmm2[14],xmm3[15],xmm2[15]
-; X86-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm3
-; X86-SSE2-NEXT:    psrlw $8, %xmm3
 ; X86-SSE2-NEXT:    movdqa %xmm0, %xmm1
-; X86-SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
+; X86-SSE2-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; X86-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1
 ; X86-SSE2-NEXT:    psrlw $8, %xmm1
-; X86-SSE2-NEXT:    packuswb %xmm3, %xmm1
-; X86-SSE2-NEXT:    movdqa %xmm0, %xmm2
-; X86-SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; X86-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm2
-; X86-SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; X86-SSE2-NEXT:    pand %xmm3, %xmm2
 ; X86-SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; X86-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE2-NEXT:    pand %xmm3, %xmm0
-; X86-SSE2-NEXT:    packuswb %xmm2, %xmm0
-; X86-SSE2-NEXT:    por %xmm1, %xmm0
+; X86-SSE2-NEXT:    psrlw $8, %xmm0
+; X86-SSE2-NEXT:    packuswb %xmm1, %xmm0
 ; X86-SSE2-NEXT:    retl
   %res = call <16 x i8> @llvm.fshl.v16i8(<16 x i8> %x, <16 x i8> %x, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1>)
   ret <16 x i8> %res

diff  --git a/llvm/test/CodeGen/X86/vector-fshl-rot-256.ll b/llvm/test/CodeGen/X86/vector-fshl-rot-256.ll
index 4c70b4b96ce2b..5550bbcd83b2a 100644
--- a/llvm/test/CodeGen/X86/vector-fshl-rot-256.ll
+++ b/llvm/test/CodeGen/X86/vector-fshl-rot-256.ll
@@ -1275,104 +1275,56 @@ define <32 x i8> @constant_funnnel_v32i8(<32 x i8> %x) nounwind {
 ; AVX1-LABEL: constant_funnnel_v32i8:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpxor %xmm8, %xmm8, %xmm8
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm8[8],xmm1[9],xmm8[9],xmm1[10],xmm8[10],xmm1[11],xmm8[11],xmm1[12],xmm8[12],xmm1[13],xmm8[13],xmm1[14],xmm8[14],xmm1[15],xmm8[15]
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm9 = [256,128,64,32,16,8,4,2]
-; AVX1-NEXT:    vpmullw %xmm3, %xmm9, %xmm3
-; AVX1-NEXT:    vpsrlw $8, %xmm3, %xmm3
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm5 = 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:    vmovdqa {{.*#+}} xmm6 = [256,2,4,8,16,32,64,128]
-; AVX1-NEXT:    vpmullw %xmm6, %xmm5, %xmm7
-; AVX1-NEXT:    vpsrlw $8, %xmm7, %xmm7
-; AVX1-NEXT:    vpackuswb %xmm3, %xmm7, %xmm3
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm7 = [1,128,64,32,16,8,4,2]
-; AVX1-NEXT:    vpmullw %xmm7, %xmm1, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpand %xmm2, %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 = [1,128,64,32,16,8,4,2]
+; AVX1-NEXT:    vpmullw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrlw $8, %xmm2, %xmm2
+; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [1,2,4,8,16,32,64,128]
-; AVX1-NEXT:    vpmullw %xmm4, %xmm5, %xmm5
-; AVX1-NEXT:    vpand %xmm2, %xmm5, %xmm5
-; AVX1-NEXT:    vpackuswb %xmm1, %xmm5, %xmm1
-; AVX1-NEXT:    vpor %xmm3, %xmm1, %xmm1
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm0[8],xmm8[8],xmm0[9],xmm8[9],xmm0[10],xmm8[10],xmm0[11],xmm8[11],xmm0[12],xmm8[12],xmm0[13],xmm8[13],xmm0[14],xmm8[14],xmm0[15],xmm8[15]
-; AVX1-NEXT:    vpmullw %xmm3, %xmm9, %xmm3
-; AVX1-NEXT:    vpsrlw $8, %xmm3, %xmm3
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm5 = 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 %xmm6, %xmm5, %xmm6
-; AVX1-NEXT:    vpsrlw $8, %xmm6, %xmm6
-; AVX1-NEXT:    vpackuswb %xmm3, %xmm6, %xmm3
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vpmullw %xmm7, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpmullw %xmm4, %xmm5, %xmm4
-; AVX1-NEXT:    vpand %xmm2, %xmm4, %xmm2
-; AVX1-NEXT:    vpackuswb %xmm0, %xmm2, %xmm0
-; AVX1-NEXT:    vpor %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpmullw %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrlw $8, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX1-NEXT:    vpmullw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrlw $8, %xmm2, %xmm2
+; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; AVX1-NEXT:    vpmullw %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpsrlw $8, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: constant_funnnel_v32i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX2-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31]
-; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX2-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX2-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23]
+; AVX2-NEXT:    vpunpckhbw {{.*#+}} ymm1 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
 ; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm1, %ymm1
 ; AVX2-NEXT:    vpsrlw $8, %ymm1, %ymm1
-; AVX2-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
-; AVX2-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX2-NEXT:    vpand %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
 ; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vpand %ymm3, %ymm0, %ymm0
-; AVX2-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
-; AVX2-NEXT:    vpor %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; AVX2-NEXT:    vpackuswb %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512F-LABEL: constant_funnnel_v32i8:
 ; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31]
-; AVX512F-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX512F-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23]
+; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm1 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
 ; AVX512F-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm1, %ymm1
 ; AVX512F-NEXT:    vpsrlw $8, %ymm1, %ymm1
-; AVX512F-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512F-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512F-NEXT:    vpand %ymm3, %ymm2, %ymm2
 ; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
 ; AVX512F-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX512F-NEXT:    vpand %ymm3, %ymm0, %ymm0
-; AVX512F-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
-; AVX512F-NEXT:    vpor %ymm1, %ymm0, %ymm0
+; AVX512F-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; AVX512F-NEXT:    vpackuswb %ymm1, %ymm0, %ymm0
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512VL-LABEL: constant_funnnel_v32i8:
 ; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31]
-; AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX512VL-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23]
+; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm1 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
 ; AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm1, %ymm1
 ; AVX512VL-NEXT:    vpsrlw $8, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512VL-NEXT:    vpand %ymm3, %ymm2, %ymm2
 ; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
 ; AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX512VL-NEXT:    vpand %ymm3, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpor %ymm1, %ymm0, %ymm0
+; AVX512VL-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; AVX512VL-NEXT:    vpackuswb %ymm1, %ymm0, %ymm0
 ; AVX512VL-NEXT:    retq
 ;
 ; AVX512BW-LABEL: constant_funnnel_v32i8:

diff  --git a/llvm/test/CodeGen/X86/vector-fshl-rot-512.ll b/llvm/test/CodeGen/X86/vector-fshl-rot-512.ll
index 2eb4d6aa53152..df579191c87fa 100644
--- a/llvm/test/CodeGen/X86/vector-fshl-rot-512.ll
+++ b/llvm/test/CodeGen/X86/vector-fshl-rot-512.ll
@@ -557,177 +557,93 @@ define <64 x i8> @constant_funnnel_v64i8(<64 x i8> %x) nounwind {
 ; AVX512F-LABEL: constant_funnnel_v64i8:
 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512F-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm1[8],ymm2[8],ymm1[9],ymm2[9],ymm1[10],ymm2[10],ymm1[11],ymm2[11],ymm1[12],ymm2[12],ymm1[13],ymm2[13],ymm1[14],ymm2[14],ymm1[15],ymm2[15],ymm1[24],ymm2[24],ymm1[25],ymm2[25],ymm1[26],ymm2[26],ymm1[27],ymm2[27],ymm1[28],ymm2[28],ymm1[29],ymm2[29],ymm1[30],ymm2[30],ymm1[31],ymm2[31]
-; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [256,128,64,32,16,8,4,2,256,128,64,32,16,8,4,2]
-; AVX512F-NEXT:    # ymm4 = mem[0,1,0,1]
-; AVX512F-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512F-NEXT:    vpsrlw $8, %ymm3, %ymm3
-; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm5 = ymm1[0],ymm2[0],ymm1[1],ymm2[1],ymm1[2],ymm2[2],ymm1[3],ymm2[3],ymm1[4],ymm2[4],ymm1[5],ymm2[5],ymm1[6],ymm2[6],ymm1[7],ymm2[7],ymm1[16],ymm2[16],ymm1[17],ymm2[17],ymm1[18],ymm2[18],ymm1[19],ymm2[19],ymm1[20],ymm2[20],ymm1[21],ymm2[21],ymm1[22],ymm2[22],ymm1[23],ymm2[23]
-; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm6 = [256,2,4,8,16,32,64,128,256,2,4,8,16,32,64,128]
-; AVX512F-NEXT:    # ymm6 = mem[0,1,0,1]
-; AVX512F-NEXT:    vpmullw %ymm6, %ymm5, %ymm5
-; AVX512F-NEXT:    vpsrlw $8, %ymm5, %ymm5
-; AVX512F-NEXT:    vpackuswb %ymm3, %ymm5, %ymm3
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm5 = ymm0[8],ymm2[8],ymm0[9],ymm2[9],ymm0[10],ymm2[10],ymm0[11],ymm2[11],ymm0[12],ymm2[12],ymm0[13],ymm2[13],ymm0[14],ymm2[14],ymm0[15],ymm2[15],ymm0[24],ymm2[24],ymm0[25],ymm2[25],ymm0[26],ymm2[26],ymm0[27],ymm2[27],ymm0[28],ymm2[28],ymm0[29],ymm2[29],ymm0[30],ymm2[30],ymm0[31],ymm2[31]
-; AVX512F-NEXT:    vpmullw %ymm4, %ymm5, %ymm4
-; AVX512F-NEXT:    vpsrlw $8, %ymm4, %ymm4
-; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm2 = ymm0[0],ymm2[0],ymm0[1],ymm2[1],ymm0[2],ymm2[2],ymm0[3],ymm2[3],ymm0[4],ymm2[4],ymm0[5],ymm2[5],ymm0[6],ymm2[6],ymm0[7],ymm2[7],ymm0[16],ymm2[16],ymm0[17],ymm2[17],ymm0[18],ymm2[18],ymm0[19],ymm2[19],ymm0[20],ymm2[20],ymm0[21],ymm2[21],ymm0[22],ymm2[22],ymm0[23],ymm2[23]
-; AVX512F-NEXT:    vpmullw %ymm6, %ymm2, %ymm2
+; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
+; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm3 = [1,128,64,32,16,8,4,2,1,128,64,32,16,8,4,2]
+; AVX512F-NEXT:    # ymm3 = mem[0,1,0,1]
+; AVX512F-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
 ; AVX512F-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX512F-NEXT:    vpackuswb %ymm4, %ymm2, %ymm2
-; AVX512F-NEXT:    vinserti64x4 $1, %ymm3, %zmm2, %zmm2
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [1,128,64,32,16,8,4,2,1,128,64,32,16,8,4,2]
-; AVX512F-NEXT:    # ymm4 = mem[0,1,0,1]
-; AVX512F-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm5 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512F-NEXT:    vpand %ymm5, %ymm3, %ymm3
 ; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
-; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm6 = [1,2,4,8,16,32,64,128,1,2,4,8,16,32,64,128]
-; AVX512F-NEXT:    # ymm6 = mem[0,1,0,1]
-; AVX512F-NEXT:    vpmullw %ymm6, %ymm1, %ymm1
-; AVX512F-NEXT:    vpand %ymm5, %ymm1, %ymm1
-; AVX512F-NEXT:    vpackuswb %ymm3, %ymm1, %ymm1
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512F-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512F-NEXT:    vpand %ymm5, %ymm3, %ymm3
+; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [1,2,4,8,16,32,64,128,1,2,4,8,16,32,64,128]
+; AVX512F-NEXT:    # ymm4 = mem[0,1,0,1]
+; AVX512F-NEXT:    vpmullw %ymm4, %ymm1, %ymm1
+; AVX512F-NEXT:    vpsrlw $8, %ymm1, %ymm1
+; AVX512F-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
+; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
+; AVX512F-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
+; AVX512F-NEXT:    vpsrlw $8, %ymm2, %ymm2
 ; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
-; AVX512F-NEXT:    vpmullw %ymm6, %ymm0, %ymm0
-; AVX512F-NEXT:    vpand %ymm5, %ymm0, %ymm0
-; AVX512F-NEXT:    vpackuswb %ymm3, %ymm0, %ymm0
+; AVX512F-NEXT:    vpmullw %ymm4, %ymm0, %ymm0
+; AVX512F-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; AVX512F-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
 ; AVX512F-NEXT:    vinserti64x4 $1, %ymm1, %zmm0, %zmm0
-; AVX512F-NEXT:    vporq %zmm2, %zmm0, %zmm0
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512VL-LABEL: constant_funnnel_v64i8:
 ; AVX512VL:       # %bb.0:
 ; AVX512VL-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512VL-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm1[8],ymm2[8],ymm1[9],ymm2[9],ymm1[10],ymm2[10],ymm1[11],ymm2[11],ymm1[12],ymm2[12],ymm1[13],ymm2[13],ymm1[14],ymm2[14],ymm1[15],ymm2[15],ymm1[24],ymm2[24],ymm1[25],ymm2[25],ymm1[26],ymm2[26],ymm1[27],ymm2[27],ymm1[28],ymm2[28],ymm1[29],ymm2[29],ymm1[30],ymm2[30],ymm1[31],ymm2[31]
-; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [256,128,64,32,16,8,4,2,256,128,64,32,16,8,4,2]
-; AVX512VL-NEXT:    # ymm4 = mem[0,1,0,1]
-; AVX512VL-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512VL-NEXT:    vpsrlw $8, %ymm3, %ymm3
-; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm5 = ymm1[0],ymm2[0],ymm1[1],ymm2[1],ymm1[2],ymm2[2],ymm1[3],ymm2[3],ymm1[4],ymm2[4],ymm1[5],ymm2[5],ymm1[6],ymm2[6],ymm1[7],ymm2[7],ymm1[16],ymm2[16],ymm1[17],ymm2[17],ymm1[18],ymm2[18],ymm1[19],ymm2[19],ymm1[20],ymm2[20],ymm1[21],ymm2[21],ymm1[22],ymm2[22],ymm1[23],ymm2[23]
-; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm6 = [256,2,4,8,16,32,64,128,256,2,4,8,16,32,64,128]
-; AVX512VL-NEXT:    # ymm6 = mem[0,1,0,1]
-; AVX512VL-NEXT:    vpmullw %ymm6, %ymm5, %ymm5
-; AVX512VL-NEXT:    vpsrlw $8, %ymm5, %ymm5
-; AVX512VL-NEXT:    vpackuswb %ymm3, %ymm5, %ymm3
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm5 = ymm0[8],ymm2[8],ymm0[9],ymm2[9],ymm0[10],ymm2[10],ymm0[11],ymm2[11],ymm0[12],ymm2[12],ymm0[13],ymm2[13],ymm0[14],ymm2[14],ymm0[15],ymm2[15],ymm0[24],ymm2[24],ymm0[25],ymm2[25],ymm0[26],ymm2[26],ymm0[27],ymm2[27],ymm0[28],ymm2[28],ymm0[29],ymm2[29],ymm0[30],ymm2[30],ymm0[31],ymm2[31]
-; AVX512VL-NEXT:    vpmullw %ymm4, %ymm5, %ymm4
-; AVX512VL-NEXT:    vpsrlw $8, %ymm4, %ymm4
-; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm2 = ymm0[0],ymm2[0],ymm0[1],ymm2[1],ymm0[2],ymm2[2],ymm0[3],ymm2[3],ymm0[4],ymm2[4],ymm0[5],ymm2[5],ymm0[6],ymm2[6],ymm0[7],ymm2[7],ymm0[16],ymm2[16],ymm0[17],ymm2[17],ymm0[18],ymm2[18],ymm0[19],ymm2[19],ymm0[20],ymm2[20],ymm0[21],ymm2[21],ymm0[22],ymm2[22],ymm0[23],ymm2[23]
-; AVX512VL-NEXT:    vpmullw %ymm6, %ymm2, %ymm2
+; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
+; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm3 = [1,128,64,32,16,8,4,2,1,128,64,32,16,8,4,2]
+; AVX512VL-NEXT:    # ymm3 = mem[0,1,0,1]
+; AVX512VL-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
 ; AVX512VL-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX512VL-NEXT:    vpackuswb %ymm4, %ymm2, %ymm2
-; AVX512VL-NEXT:    vinserti64x4 $1, %ymm3, %zmm2, %zmm2
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [1,128,64,32,16,8,4,2,1,128,64,32,16,8,4,2]
-; AVX512VL-NEXT:    # ymm4 = mem[0,1,0,1]
-; AVX512VL-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm5 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512VL-NEXT:    vpand %ymm5, %ymm3, %ymm3
 ; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
-; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm6 = [1,2,4,8,16,32,64,128,1,2,4,8,16,32,64,128]
-; AVX512VL-NEXT:    # ymm6 = mem[0,1,0,1]
-; AVX512VL-NEXT:    vpmullw %ymm6, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpand %ymm5, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpackuswb %ymm3, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512VL-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512VL-NEXT:    vpand %ymm5, %ymm3, %ymm3
+; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [1,2,4,8,16,32,64,128,1,2,4,8,16,32,64,128]
+; AVX512VL-NEXT:    # ymm4 = mem[0,1,0,1]
+; AVX512VL-NEXT:    vpmullw %ymm4, %ymm1, %ymm1
+; AVX512VL-NEXT:    vpsrlw $8, %ymm1, %ymm1
+; AVX512VL-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
+; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
+; AVX512VL-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
+; AVX512VL-NEXT:    vpsrlw $8, %ymm2, %ymm2
 ; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
-; AVX512VL-NEXT:    vpmullw %ymm6, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpand %ymm5, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpackuswb %ymm3, %ymm0, %ymm0
+; AVX512VL-NEXT:    vpmullw %ymm4, %ymm0, %ymm0
+; AVX512VL-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; AVX512VL-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
 ; AVX512VL-NEXT:    vinserti64x4 $1, %ymm1, %zmm0, %zmm0
-; AVX512VL-NEXT:    vporq %zmm2, %zmm0, %zmm0
 ; AVX512VL-NEXT:    retq
 ;
 ; AVX512BW-LABEL: constant_funnnel_v64i8:
 ; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
-; AVX512BW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512BW-NEXT:    vpsrlw $8, %zmm2, %zmm2
-; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
+; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm1 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512BW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpsrlw $8, %zmm1, %zmm1
-; AVX512BW-NEXT:    vpackuswb %zmm2, %zmm1, %zmm1
-; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512BW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512BW-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
-; AVX512BW-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpsrlw $8, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512VLBW-LABEL: constant_funnnel_v64i8:
 ; AVX512VLBW:       # %bb.0:
-; AVX512VLBW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512VLBW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
-; AVX512VLBW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VLBW-NEXT:    vpsrlw $8, %zmm2, %zmm2
-; AVX512VLBW-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
+; AVX512VLBW-NEXT:    vpunpckhbw {{.*#+}} zmm1 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512VLBW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm1, %zmm1
 ; AVX512VLBW-NEXT:    vpsrlw $8, %zmm1, %zmm1
-; AVX512VLBW-NEXT:    vpackuswb %zmm2, %zmm1, %zmm1
-; AVX512VLBW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512VLBW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VLBW-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512VLBW-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512VLBW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512VLBW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
-; AVX512VLBW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
-; AVX512VLBW-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
-; AVX512VLBW-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512VLBW-NEXT:    vpsrlw $8, %zmm0, %zmm0
+; AVX512VLBW-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512VLBW-NEXT:    retq
 ;
 ; AVX512VBMI2-LABEL: constant_funnnel_v64i8:
 ; AVX512VBMI2:       # %bb.0:
-; AVX512VBMI2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512VBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
-; AVX512VBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VBMI2-NEXT:    vpsrlw $8, %zmm2, %zmm2
-; AVX512VBMI2-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
+; AVX512VBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm1 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512VBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm1, %zmm1
 ; AVX512VBMI2-NEXT:    vpsrlw $8, %zmm1, %zmm1
-; AVX512VBMI2-NEXT:    vpackuswb %zmm2, %zmm1, %zmm1
-; AVX512VBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512VBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VBMI2-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512VBMI2-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512VBMI2-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512VBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
-; AVX512VBMI2-NEXT:    vpandq %zmm3, %zmm0, %zmm0
-; AVX512VBMI2-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
-; AVX512VBMI2-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512VBMI2-NEXT:    vpsrlw $8, %zmm0, %zmm0
+; AVX512VBMI2-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512VBMI2-NEXT:    retq
 ;
 ; AVX512VLVBMI2-LABEL: constant_funnnel_v64i8:
 ; AVX512VLVBMI2:       # %bb.0:
-; AVX512VLVBMI2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512VLVBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
-; AVX512VLVBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VLVBMI2-NEXT:    vpsrlw $8, %zmm2, %zmm2
-; AVX512VLVBMI2-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
+; AVX512VLVBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm1 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512VLVBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm1, %zmm1
 ; AVX512VLVBMI2-NEXT:    vpsrlw $8, %zmm1, %zmm1
-; AVX512VLVBMI2-NEXT:    vpackuswb %zmm2, %zmm1, %zmm1
-; AVX512VLVBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512VLVBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VLVBMI2-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512VLVBMI2-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512VLVBMI2-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512VLVBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
-; AVX512VLVBMI2-NEXT:    vpandq %zmm3, %zmm0, %zmm0
-; AVX512VLVBMI2-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
-; AVX512VLVBMI2-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512VLVBMI2-NEXT:    vpsrlw $8, %zmm0, %zmm0
+; AVX512VLVBMI2-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512VLVBMI2-NEXT:    retq
   %res = call <64 x i8> @llvm.fshl.v64i8(<64 x i8> %x, <64 x i8> %x, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1>)
   ret <64 x i8> %res

diff  --git a/llvm/test/CodeGen/X86/vector-fshr-rot-128.ll b/llvm/test/CodeGen/X86/vector-fshr-rot-128.ll
index 98fa47eb8b386..2762549b27f66 100644
--- a/llvm/test/CodeGen/X86/vector-fshr-rot-128.ll
+++ b/llvm/test/CodeGen/X86/vector-fshr-rot-128.ll
@@ -1623,87 +1623,28 @@ define <8 x i16> @constant_funnnel_v8i16(<8 x i16> %x) nounwind {
 }
 
 define <16 x i8> @constant_funnnel_v16i8(<16 x i8> %x) nounwind {
-; SSE2-LABEL: constant_funnnel_v16i8:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    pxor %xmm1, %xmm1
-; SSE2-NEXT:    movdqa %xmm0, %xmm2
-; SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm1[8],xmm2[9],xmm1[9],xmm2[10],xmm1[10],xmm2[11],xmm1[11],xmm2[12],xmm1[12],xmm2[13],xmm1[13],xmm2[14],xmm1[14],xmm2[15],xmm1[15]
-; SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
-; SSE2-NEXT:    psrlw $8, %xmm2
-; SSE2-NEXT:    movdqa %xmm0, %xmm3
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3],xmm3[4],xmm1[4],xmm3[5],xmm1[5],xmm3[6],xmm1[6],xmm3[7],xmm1[7]
-; SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm3
-; SSE2-NEXT:    psrlw $8, %xmm3
-; SSE2-NEXT:    packuswb %xmm2, %xmm3
-; SSE2-NEXT:    movdqa %xmm0, %xmm1
-; SSE2-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
-; SSE2-NEXT:    pand %xmm2, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; SSE2-NEXT:    pand %xmm2, %xmm0
-; SSE2-NEXT:    packuswb %xmm1, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: constant_funnnel_v16i8:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    movdqa %xmm0, %xmm2
-; SSE41-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
-; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; SSE41-NEXT:    pand %xmm3, %xmm2
-; 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 {{.*#+}} xmm4 = [1,128,64,32,16,8,4,2]
-; SSE41-NEXT:    pmullw %xmm1, %xmm4
-; SSE41-NEXT:    pand %xmm3, %xmm4
-; SSE41-NEXT:    packuswb %xmm2, %xmm4
-; SSE41-NEXT:    pxor %xmm2, %xmm2
-; SSE41-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm2[8],xmm0[9],xmm2[9],xmm0[10],xmm2[10],xmm0[11],xmm2[11],xmm0[12],xmm2[12],xmm0[13],xmm2[13],xmm0[14],xmm2[14],xmm0[15],xmm2[15]
-; SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; SSE41-NEXT:    psrlw $8, %xmm0
-; SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
-; SSE41-NEXT:    psrlw $8, %xmm1
-; SSE41-NEXT:    packuswb %xmm0, %xmm1
-; SSE41-NEXT:    por %xmm4, %xmm1
-; SSE41-NEXT:    movdqa %xmm1, %xmm0
-; SSE41-NEXT:    retq
-;
-; AVX1-LABEL: constant_funnnel_v16i8:
-; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpand %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = 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 {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm3, %xmm4
-; AVX1-NEXT:    vpand %xmm2, %xmm4, %xmm2
-; AVX1-NEXT:    vpackuswb %xmm1, %xmm2, %xmm1
-; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm0 = xmm0[8],xmm2[8],xmm0[9],xmm2[9],xmm0[10],xmm2[10],xmm0[11],xmm2[11],xmm0[12],xmm2[12],xmm0[13],xmm2[13],xmm0[14],xmm2[14],xmm0[15],xmm2[15]
-; AVX1-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
-; AVX1-NEXT:    vpsrlw $8, %xmm0, %xmm0
-; AVX1-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm3, %xmm2
-; AVX1-NEXT:    vpsrlw $8, %xmm2, %xmm2
-; AVX1-NEXT:    vpackuswb %xmm0, %xmm2, %xmm0
-; AVX1-NEXT:    vpor %xmm0, %xmm1, %xmm0
-; AVX1-NEXT:    retq
+; SSE-LABEL: constant_funnnel_v16i8:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movdqa %xmm0, %xmm1
+; SSE-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
+; SSE-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
+; SSE-NEXT:    psrlw $8, %xmm1
+; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE-NEXT:    psrlw $8, %xmm0
+; SSE-NEXT:    packuswb %xmm1, %xmm0
+; SSE-NEXT:    retq
 ;
-; AVX2-LABEL: constant_funnnel_v16i8:
-; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} ymm0 = 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
-; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm1
-; AVX2-NEXT:    vpsrlw $8, %ymm1, %ymm1
-; AVX2-NEXT:    vextracti128 $1, %ymm1, %xmm2
-; AVX2-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
-; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm2
-; AVX2-NEXT:    vpackuswb %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vzeroupper
-; AVX2-NEXT:    retq
+; AVX-LABEL: constant_funnnel_v16i8:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
+; AVX-NEXT:    vpsrlw $8, %xmm1, %xmm1
+; AVX-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; AVX-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
+; AVX-NEXT:    vpsrlw $8, %xmm0, %xmm0
+; AVX-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    retq
 ;
 ; AVX512F-LABEL: constant_funnnel_v16i8:
 ; AVX512F:       # %bb.0:
@@ -1778,26 +1719,14 @@ define <16 x i8> @constant_funnnel_v16i8(<16 x i8> %x) nounwind {
 ;
 ; X86-SSE2-LABEL: constant_funnnel_v16i8:
 ; X86-SSE2:       # %bb.0:
-; X86-SSE2-NEXT:    pxor %xmm2, %xmm2
-; X86-SSE2-NEXT:    movdqa %xmm0, %xmm3
-; X86-SSE2-NEXT:    punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm2[8],xmm3[9],xmm2[9],xmm3[10],xmm2[10],xmm3[11],xmm2[11],xmm3[12],xmm2[12],xmm3[13],xmm2[13],xmm3[14],xmm2[14],xmm3[15],xmm2[15]
-; X86-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm3
-; X86-SSE2-NEXT:    psrlw $8, %xmm3
 ; X86-SSE2-NEXT:    movdqa %xmm0, %xmm1
-; X86-SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
+; X86-SSE2-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; X86-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1
 ; X86-SSE2-NEXT:    psrlw $8, %xmm1
-; X86-SSE2-NEXT:    packuswb %xmm3, %xmm1
-; X86-SSE2-NEXT:    movdqa %xmm0, %xmm2
-; X86-SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; X86-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm2
-; X86-SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; X86-SSE2-NEXT:    pand %xmm3, %xmm2
 ; X86-SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; X86-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE2-NEXT:    pand %xmm3, %xmm0
-; X86-SSE2-NEXT:    packuswb %xmm2, %xmm0
-; X86-SSE2-NEXT:    por %xmm1, %xmm0
+; X86-SSE2-NEXT:    psrlw $8, %xmm0
+; X86-SSE2-NEXT:    packuswb %xmm1, %xmm0
 ; X86-SSE2-NEXT:    retl
   %res = call <16 x i8> @llvm.fshr.v16i8(<16 x i8> %x, <16 x i8> %x, <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1>)
   ret <16 x i8> %res

diff  --git a/llvm/test/CodeGen/X86/vector-fshr-rot-256.ll b/llvm/test/CodeGen/X86/vector-fshr-rot-256.ll
index 26b2711cc8b88..2b502631a251a 100644
--- a/llvm/test/CodeGen/X86/vector-fshr-rot-256.ll
+++ b/llvm/test/CodeGen/X86/vector-fshr-rot-256.ll
@@ -1351,104 +1351,56 @@ define <32 x i8> @constant_funnnel_v32i8(<32 x i8> %x) nounwind {
 ; AVX1-LABEL: constant_funnnel_v32i8:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpxor %xmm8, %xmm8, %xmm8
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm8[8],xmm1[9],xmm8[9],xmm1[10],xmm8[10],xmm1[11],xmm8[11],xmm1[12],xmm8[12],xmm1[13],xmm8[13],xmm1[14],xmm8[14],xmm1[15],xmm8[15]
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm9 = [256,2,4,8,16,32,64,128]
-; AVX1-NEXT:    vpmullw %xmm3, %xmm9, %xmm3
-; AVX1-NEXT:    vpsrlw $8, %xmm3, %xmm3
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm5 = 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:    vmovdqa {{.*#+}} xmm6 = [256,128,64,32,16,8,4,2]
-; AVX1-NEXT:    vpmullw %xmm6, %xmm5, %xmm7
-; AVX1-NEXT:    vpsrlw $8, %xmm7, %xmm7
-; AVX1-NEXT:    vpackuswb %xmm3, %xmm7, %xmm3
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm7 = [1,2,4,8,16,32,64,128]
-; AVX1-NEXT:    vpmullw %xmm7, %xmm1, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpand %xmm2, %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 = [1,2,4,8,16,32,64,128]
+; AVX1-NEXT:    vpmullw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrlw $8, %xmm2, %xmm2
+; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [1,128,64,32,16,8,4,2]
-; AVX1-NEXT:    vpmullw %xmm4, %xmm5, %xmm5
-; AVX1-NEXT:    vpand %xmm2, %xmm5, %xmm5
-; AVX1-NEXT:    vpackuswb %xmm1, %xmm5, %xmm1
-; AVX1-NEXT:    vpor %xmm3, %xmm1, %xmm1
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm0[8],xmm8[8],xmm0[9],xmm8[9],xmm0[10],xmm8[10],xmm0[11],xmm8[11],xmm0[12],xmm8[12],xmm0[13],xmm8[13],xmm0[14],xmm8[14],xmm0[15],xmm8[15]
-; AVX1-NEXT:    vpmullw %xmm3, %xmm9, %xmm3
-; AVX1-NEXT:    vpsrlw $8, %xmm3, %xmm3
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm5 = 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 %xmm6, %xmm5, %xmm6
-; AVX1-NEXT:    vpsrlw $8, %xmm6, %xmm6
-; AVX1-NEXT:    vpackuswb %xmm3, %xmm6, %xmm3
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vpmullw %xmm7, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpmullw %xmm4, %xmm5, %xmm4
-; AVX1-NEXT:    vpand %xmm2, %xmm4, %xmm2
-; AVX1-NEXT:    vpackuswb %xmm0, %xmm2, %xmm0
-; AVX1-NEXT:    vpor %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpmullw %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrlw $8, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX1-NEXT:    vpmullw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrlw $8, %xmm2, %xmm2
+; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; AVX1-NEXT:    vpmullw %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpsrlw $8, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: constant_funnnel_v32i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX2-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31]
-; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX2-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX2-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23]
+; AVX2-NEXT:    vpunpckhbw {{.*#+}} ymm1 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
 ; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm1, %ymm1
 ; AVX2-NEXT:    vpsrlw $8, %ymm1, %ymm1
-; AVX2-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
-; AVX2-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX2-NEXT:    vpand %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
 ; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vpand %ymm3, %ymm0, %ymm0
-; AVX2-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
-; AVX2-NEXT:    vpor %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; AVX2-NEXT:    vpackuswb %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512F-LABEL: constant_funnnel_v32i8:
 ; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31]
-; AVX512F-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX512F-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23]
+; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm1 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
 ; AVX512F-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm1, %ymm1
 ; AVX512F-NEXT:    vpsrlw $8, %ymm1, %ymm1
-; AVX512F-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512F-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512F-NEXT:    vpand %ymm3, %ymm2, %ymm2
 ; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
 ; AVX512F-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX512F-NEXT:    vpand %ymm3, %ymm0, %ymm0
-; AVX512F-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
-; AVX512F-NEXT:    vpor %ymm1, %ymm0, %ymm0
+; AVX512F-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; AVX512F-NEXT:    vpackuswb %ymm1, %ymm0, %ymm0
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512VL-LABEL: constant_funnnel_v32i8:
 ; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31]
-; AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX512VL-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23]
+; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm1 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
 ; AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm1, %ymm1
 ; AVX512VL-NEXT:    vpsrlw $8, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512VL-NEXT:    vpand %ymm3, %ymm2, %ymm2
 ; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
 ; AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX512VL-NEXT:    vpand %ymm3, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpor %ymm1, %ymm0, %ymm0
+; AVX512VL-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; AVX512VL-NEXT:    vpackuswb %ymm1, %ymm0, %ymm0
 ; AVX512VL-NEXT:    retq
 ;
 ; AVX512BW-LABEL: constant_funnnel_v32i8:

diff  --git a/llvm/test/CodeGen/X86/vector-fshr-rot-512.ll b/llvm/test/CodeGen/X86/vector-fshr-rot-512.ll
index 5da28c68fc78c..ce177f1a8a819 100644
--- a/llvm/test/CodeGen/X86/vector-fshr-rot-512.ll
+++ b/llvm/test/CodeGen/X86/vector-fshr-rot-512.ll
@@ -575,177 +575,93 @@ define <64 x i8> @constant_funnnel_v64i8(<64 x i8> %x) nounwind {
 ; AVX512F-LABEL: constant_funnnel_v64i8:
 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512F-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm1[8],ymm2[8],ymm1[9],ymm2[9],ymm1[10],ymm2[10],ymm1[11],ymm2[11],ymm1[12],ymm2[12],ymm1[13],ymm2[13],ymm1[14],ymm2[14],ymm1[15],ymm2[15],ymm1[24],ymm2[24],ymm1[25],ymm2[25],ymm1[26],ymm2[26],ymm1[27],ymm2[27],ymm1[28],ymm2[28],ymm1[29],ymm2[29],ymm1[30],ymm2[30],ymm1[31],ymm2[31]
-; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [256,2,4,8,16,32,64,128,256,2,4,8,16,32,64,128]
-; AVX512F-NEXT:    # ymm4 = mem[0,1,0,1]
-; AVX512F-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512F-NEXT:    vpsrlw $8, %ymm3, %ymm3
-; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm5 = ymm1[0],ymm2[0],ymm1[1],ymm2[1],ymm1[2],ymm2[2],ymm1[3],ymm2[3],ymm1[4],ymm2[4],ymm1[5],ymm2[5],ymm1[6],ymm2[6],ymm1[7],ymm2[7],ymm1[16],ymm2[16],ymm1[17],ymm2[17],ymm1[18],ymm2[18],ymm1[19],ymm2[19],ymm1[20],ymm2[20],ymm1[21],ymm2[21],ymm1[22],ymm2[22],ymm1[23],ymm2[23]
-; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm6 = [256,128,64,32,16,8,4,2,256,128,64,32,16,8,4,2]
-; AVX512F-NEXT:    # ymm6 = mem[0,1,0,1]
-; AVX512F-NEXT:    vpmullw %ymm6, %ymm5, %ymm5
-; AVX512F-NEXT:    vpsrlw $8, %ymm5, %ymm5
-; AVX512F-NEXT:    vpackuswb %ymm3, %ymm5, %ymm3
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm5 = ymm0[8],ymm2[8],ymm0[9],ymm2[9],ymm0[10],ymm2[10],ymm0[11],ymm2[11],ymm0[12],ymm2[12],ymm0[13],ymm2[13],ymm0[14],ymm2[14],ymm0[15],ymm2[15],ymm0[24],ymm2[24],ymm0[25],ymm2[25],ymm0[26],ymm2[26],ymm0[27],ymm2[27],ymm0[28],ymm2[28],ymm0[29],ymm2[29],ymm0[30],ymm2[30],ymm0[31],ymm2[31]
-; AVX512F-NEXT:    vpmullw %ymm4, %ymm5, %ymm4
-; AVX512F-NEXT:    vpsrlw $8, %ymm4, %ymm4
-; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm2 = ymm0[0],ymm2[0],ymm0[1],ymm2[1],ymm0[2],ymm2[2],ymm0[3],ymm2[3],ymm0[4],ymm2[4],ymm0[5],ymm2[5],ymm0[6],ymm2[6],ymm0[7],ymm2[7],ymm0[16],ymm2[16],ymm0[17],ymm2[17],ymm0[18],ymm2[18],ymm0[19],ymm2[19],ymm0[20],ymm2[20],ymm0[21],ymm2[21],ymm0[22],ymm2[22],ymm0[23],ymm2[23]
-; AVX512F-NEXT:    vpmullw %ymm6, %ymm2, %ymm2
+; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
+; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm3 = [1,2,4,8,16,32,64,128,1,2,4,8,16,32,64,128]
+; AVX512F-NEXT:    # ymm3 = mem[0,1,0,1]
+; AVX512F-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
 ; AVX512F-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX512F-NEXT:    vpackuswb %ymm4, %ymm2, %ymm2
-; AVX512F-NEXT:    vinserti64x4 $1, %ymm3, %zmm2, %zmm2
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [1,2,4,8,16,32,64,128,1,2,4,8,16,32,64,128]
-; AVX512F-NEXT:    # ymm4 = mem[0,1,0,1]
-; AVX512F-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm5 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512F-NEXT:    vpand %ymm5, %ymm3, %ymm3
 ; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
-; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm6 = [1,128,64,32,16,8,4,2,1,128,64,32,16,8,4,2]
-; AVX512F-NEXT:    # ymm6 = mem[0,1,0,1]
-; AVX512F-NEXT:    vpmullw %ymm6, %ymm1, %ymm1
-; AVX512F-NEXT:    vpand %ymm5, %ymm1, %ymm1
-; AVX512F-NEXT:    vpackuswb %ymm3, %ymm1, %ymm1
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512F-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512F-NEXT:    vpand %ymm5, %ymm3, %ymm3
+; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [1,128,64,32,16,8,4,2,1,128,64,32,16,8,4,2]
+; AVX512F-NEXT:    # ymm4 = mem[0,1,0,1]
+; AVX512F-NEXT:    vpmullw %ymm4, %ymm1, %ymm1
+; AVX512F-NEXT:    vpsrlw $8, %ymm1, %ymm1
+; AVX512F-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
+; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
+; AVX512F-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
+; AVX512F-NEXT:    vpsrlw $8, %ymm2, %ymm2
 ; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
-; AVX512F-NEXT:    vpmullw %ymm6, %ymm0, %ymm0
-; AVX512F-NEXT:    vpand %ymm5, %ymm0, %ymm0
-; AVX512F-NEXT:    vpackuswb %ymm3, %ymm0, %ymm0
+; AVX512F-NEXT:    vpmullw %ymm4, %ymm0, %ymm0
+; AVX512F-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; AVX512F-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
 ; AVX512F-NEXT:    vinserti64x4 $1, %ymm1, %zmm0, %zmm0
-; AVX512F-NEXT:    vporq %zmm2, %zmm0, %zmm0
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512VL-LABEL: constant_funnnel_v64i8:
 ; AVX512VL:       # %bb.0:
 ; AVX512VL-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512VL-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm1[8],ymm2[8],ymm1[9],ymm2[9],ymm1[10],ymm2[10],ymm1[11],ymm2[11],ymm1[12],ymm2[12],ymm1[13],ymm2[13],ymm1[14],ymm2[14],ymm1[15],ymm2[15],ymm1[24],ymm2[24],ymm1[25],ymm2[25],ymm1[26],ymm2[26],ymm1[27],ymm2[27],ymm1[28],ymm2[28],ymm1[29],ymm2[29],ymm1[30],ymm2[30],ymm1[31],ymm2[31]
-; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [256,2,4,8,16,32,64,128,256,2,4,8,16,32,64,128]
-; AVX512VL-NEXT:    # ymm4 = mem[0,1,0,1]
-; AVX512VL-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512VL-NEXT:    vpsrlw $8, %ymm3, %ymm3
-; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm5 = ymm1[0],ymm2[0],ymm1[1],ymm2[1],ymm1[2],ymm2[2],ymm1[3],ymm2[3],ymm1[4],ymm2[4],ymm1[5],ymm2[5],ymm1[6],ymm2[6],ymm1[7],ymm2[7],ymm1[16],ymm2[16],ymm1[17],ymm2[17],ymm1[18],ymm2[18],ymm1[19],ymm2[19],ymm1[20],ymm2[20],ymm1[21],ymm2[21],ymm1[22],ymm2[22],ymm1[23],ymm2[23]
-; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm6 = [256,128,64,32,16,8,4,2,256,128,64,32,16,8,4,2]
-; AVX512VL-NEXT:    # ymm6 = mem[0,1,0,1]
-; AVX512VL-NEXT:    vpmullw %ymm6, %ymm5, %ymm5
-; AVX512VL-NEXT:    vpsrlw $8, %ymm5, %ymm5
-; AVX512VL-NEXT:    vpackuswb %ymm3, %ymm5, %ymm3
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm5 = ymm0[8],ymm2[8],ymm0[9],ymm2[9],ymm0[10],ymm2[10],ymm0[11],ymm2[11],ymm0[12],ymm2[12],ymm0[13],ymm2[13],ymm0[14],ymm2[14],ymm0[15],ymm2[15],ymm0[24],ymm2[24],ymm0[25],ymm2[25],ymm0[26],ymm2[26],ymm0[27],ymm2[27],ymm0[28],ymm2[28],ymm0[29],ymm2[29],ymm0[30],ymm2[30],ymm0[31],ymm2[31]
-; AVX512VL-NEXT:    vpmullw %ymm4, %ymm5, %ymm4
-; AVX512VL-NEXT:    vpsrlw $8, %ymm4, %ymm4
-; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm2 = ymm0[0],ymm2[0],ymm0[1],ymm2[1],ymm0[2],ymm2[2],ymm0[3],ymm2[3],ymm0[4],ymm2[4],ymm0[5],ymm2[5],ymm0[6],ymm2[6],ymm0[7],ymm2[7],ymm0[16],ymm2[16],ymm0[17],ymm2[17],ymm0[18],ymm2[18],ymm0[19],ymm2[19],ymm0[20],ymm2[20],ymm0[21],ymm2[21],ymm0[22],ymm2[22],ymm0[23],ymm2[23]
-; AVX512VL-NEXT:    vpmullw %ymm6, %ymm2, %ymm2
+; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
+; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm3 = [1,2,4,8,16,32,64,128,1,2,4,8,16,32,64,128]
+; AVX512VL-NEXT:    # ymm3 = mem[0,1,0,1]
+; AVX512VL-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
 ; AVX512VL-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX512VL-NEXT:    vpackuswb %ymm4, %ymm2, %ymm2
-; AVX512VL-NEXT:    vinserti64x4 $1, %ymm3, %zmm2, %zmm2
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [1,2,4,8,16,32,64,128,1,2,4,8,16,32,64,128]
-; AVX512VL-NEXT:    # ymm4 = mem[0,1,0,1]
-; AVX512VL-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm5 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512VL-NEXT:    vpand %ymm5, %ymm3, %ymm3
 ; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
-; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm6 = [1,128,64,32,16,8,4,2,1,128,64,32,16,8,4,2]
-; AVX512VL-NEXT:    # ymm6 = mem[0,1,0,1]
-; AVX512VL-NEXT:    vpmullw %ymm6, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpand %ymm5, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpackuswb %ymm3, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512VL-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512VL-NEXT:    vpand %ymm5, %ymm3, %ymm3
+; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [1,128,64,32,16,8,4,2,1,128,64,32,16,8,4,2]
+; AVX512VL-NEXT:    # ymm4 = mem[0,1,0,1]
+; AVX512VL-NEXT:    vpmullw %ymm4, %ymm1, %ymm1
+; AVX512VL-NEXT:    vpsrlw $8, %ymm1, %ymm1
+; AVX512VL-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
+; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
+; AVX512VL-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
+; AVX512VL-NEXT:    vpsrlw $8, %ymm2, %ymm2
 ; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
-; AVX512VL-NEXT:    vpmullw %ymm6, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpand %ymm5, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpackuswb %ymm3, %ymm0, %ymm0
+; AVX512VL-NEXT:    vpmullw %ymm4, %ymm0, %ymm0
+; AVX512VL-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; AVX512VL-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
 ; AVX512VL-NEXT:    vinserti64x4 $1, %ymm1, %zmm0, %zmm0
-; AVX512VL-NEXT:    vporq %zmm2, %zmm0, %zmm0
 ; AVX512VL-NEXT:    retq
 ;
 ; AVX512BW-LABEL: constant_funnnel_v64i8:
 ; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
-; AVX512BW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512BW-NEXT:    vpsrlw $8, %zmm2, %zmm2
-; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
+; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm1 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512BW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpsrlw $8, %zmm1, %zmm1
-; AVX512BW-NEXT:    vpackuswb %zmm2, %zmm1, %zmm1
-; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512BW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512BW-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
-; AVX512BW-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpsrlw $8, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512VLBW-LABEL: constant_funnnel_v64i8:
 ; AVX512VLBW:       # %bb.0:
-; AVX512VLBW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512VLBW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
-; AVX512VLBW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VLBW-NEXT:    vpsrlw $8, %zmm2, %zmm2
-; AVX512VLBW-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
+; AVX512VLBW-NEXT:    vpunpckhbw {{.*#+}} zmm1 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512VLBW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm1, %zmm1
 ; AVX512VLBW-NEXT:    vpsrlw $8, %zmm1, %zmm1
-; AVX512VLBW-NEXT:    vpackuswb %zmm2, %zmm1, %zmm1
-; AVX512VLBW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512VLBW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VLBW-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512VLBW-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512VLBW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512VLBW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
-; AVX512VLBW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
-; AVX512VLBW-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
-; AVX512VLBW-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512VLBW-NEXT:    vpsrlw $8, %zmm0, %zmm0
+; AVX512VLBW-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512VLBW-NEXT:    retq
 ;
 ; AVX512VBMI2-LABEL: constant_funnnel_v64i8:
 ; AVX512VBMI2:       # %bb.0:
-; AVX512VBMI2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512VBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
-; AVX512VBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VBMI2-NEXT:    vpsrlw $8, %zmm2, %zmm2
-; AVX512VBMI2-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
+; AVX512VBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm1 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512VBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm1, %zmm1
 ; AVX512VBMI2-NEXT:    vpsrlw $8, %zmm1, %zmm1
-; AVX512VBMI2-NEXT:    vpackuswb %zmm2, %zmm1, %zmm1
-; AVX512VBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512VBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VBMI2-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512VBMI2-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512VBMI2-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512VBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
-; AVX512VBMI2-NEXT:    vpandq %zmm3, %zmm0, %zmm0
-; AVX512VBMI2-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
-; AVX512VBMI2-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512VBMI2-NEXT:    vpsrlw $8, %zmm0, %zmm0
+; AVX512VBMI2-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512VBMI2-NEXT:    retq
 ;
 ; AVX512VLVBMI2-LABEL: constant_funnnel_v64i8:
 ; AVX512VLVBMI2:       # %bb.0:
-; AVX512VLVBMI2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512VLVBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
-; AVX512VLVBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VLVBMI2-NEXT:    vpsrlw $8, %zmm2, %zmm2
-; AVX512VLVBMI2-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
+; AVX512VLVBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm1 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512VLVBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm1, %zmm1
 ; AVX512VLVBMI2-NEXT:    vpsrlw $8, %zmm1, %zmm1
-; AVX512VLVBMI2-NEXT:    vpackuswb %zmm2, %zmm1, %zmm1
-; AVX512VLVBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512VLVBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VLVBMI2-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512VLVBMI2-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512VLVBMI2-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512VLVBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
-; AVX512VLVBMI2-NEXT:    vpandq %zmm3, %zmm0, %zmm0
-; AVX512VLVBMI2-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
-; AVX512VLVBMI2-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512VLVBMI2-NEXT:    vpsrlw $8, %zmm0, %zmm0
+; AVX512VLVBMI2-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512VLVBMI2-NEXT:    retq
   %res = call <64 x i8> @llvm.fshr.v64i8(<64 x i8> %x, <64 x i8> %x, <64 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1>)
   ret <64 x i8> %res

diff  --git a/llvm/test/CodeGen/X86/vector-rotate-128.ll b/llvm/test/CodeGen/X86/vector-rotate-128.ll
index 96bc224445186..572a007de6446 100644
--- a/llvm/test/CodeGen/X86/vector-rotate-128.ll
+++ b/llvm/test/CodeGen/X86/vector-rotate-128.ll
@@ -1479,87 +1479,28 @@ define <8 x i16> @constant_rotate_v8i16(<8 x i16> %a) nounwind {
 }
 
 define <16 x i8> @constant_rotate_v16i8(<16 x i8> %a) nounwind {
-; SSE2-LABEL: constant_rotate_v16i8:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    pxor %xmm1, %xmm1
-; SSE2-NEXT:    movdqa %xmm0, %xmm2
-; SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8],xmm1[8],xmm2[9],xmm1[9],xmm2[10],xmm1[10],xmm2[11],xmm1[11],xmm2[12],xmm1[12],xmm2[13],xmm1[13],xmm2[14],xmm1[14],xmm2[15],xmm1[15]
-; SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
-; SSE2-NEXT:    psrlw $8, %xmm2
-; SSE2-NEXT:    movdqa %xmm0, %xmm3
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3],xmm3[4],xmm1[4],xmm3[5],xmm1[5],xmm3[6],xmm1[6],xmm3[7],xmm1[7]
-; SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm3
-; SSE2-NEXT:    psrlw $8, %xmm3
-; SSE2-NEXT:    packuswb %xmm2, %xmm3
-; SSE2-NEXT:    movdqa %xmm0, %xmm1
-; SSE2-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
-; SSE2-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
-; SSE2-NEXT:    pand %xmm2, %xmm1
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; SSE2-NEXT:    pand %xmm2, %xmm0
-; SSE2-NEXT:    packuswb %xmm1, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSE41-LABEL: constant_rotate_v16i8:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    movdqa %xmm0, %xmm2
-; SSE41-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm2
-; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; SSE41-NEXT:    pand %xmm3, %xmm2
-; 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 {{.*#+}} xmm4 = [1,2,4,8,16,32,64,128]
-; SSE41-NEXT:    pmullw %xmm1, %xmm4
-; SSE41-NEXT:    pand %xmm3, %xmm4
-; SSE41-NEXT:    packuswb %xmm2, %xmm4
-; SSE41-NEXT:    pxor %xmm2, %xmm2
-; SSE41-NEXT:    punpckhbw {{.*#+}} xmm0 = xmm0[8],xmm2[8],xmm0[9],xmm2[9],xmm0[10],xmm2[10],xmm0[11],xmm2[11],xmm0[12],xmm2[12],xmm0[13],xmm2[13],xmm0[14],xmm2[14],xmm0[15],xmm2[15]
-; SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
-; SSE41-NEXT:    psrlw $8, %xmm0
-; SSE41-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
-; SSE41-NEXT:    psrlw $8, %xmm1
-; SSE41-NEXT:    packuswb %xmm0, %xmm1
-; SSE41-NEXT:    por %xmm4, %xmm1
-; SSE41-NEXT:    movdqa %xmm1, %xmm0
-; SSE41-NEXT:    retq
-;
-; AVX1-LABEL: constant_rotate_v16i8:
-; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpand %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = 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 {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm3, %xmm4
-; AVX1-NEXT:    vpand %xmm2, %xmm4, %xmm2
-; AVX1-NEXT:    vpackuswb %xmm1, %xmm2, %xmm1
-; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm0 = xmm0[8],xmm2[8],xmm0[9],xmm2[9],xmm0[10],xmm2[10],xmm0[11],xmm2[11],xmm0[12],xmm2[12],xmm0[13],xmm2[13],xmm0[14],xmm2[14],xmm0[15],xmm2[15]
-; AVX1-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
-; AVX1-NEXT:    vpsrlw $8, %xmm0, %xmm0
-; AVX1-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm3, %xmm2
-; AVX1-NEXT:    vpsrlw $8, %xmm2, %xmm2
-; AVX1-NEXT:    vpackuswb %xmm0, %xmm2, %xmm0
-; AVX1-NEXT:    vpor %xmm0, %xmm1, %xmm0
-; AVX1-NEXT:    retq
+; SSE-LABEL: constant_rotate_v16i8:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movdqa %xmm0, %xmm1
+; SSE-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
+; SSE-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1
+; SSE-NEXT:    psrlw $8, %xmm1
+; SSE-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0
+; SSE-NEXT:    psrlw $8, %xmm0
+; SSE-NEXT:    packuswb %xmm1, %xmm0
+; SSE-NEXT:    retq
 ;
-; AVX2-LABEL: constant_rotate_v16i8:
-; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} ymm0 = 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
-; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm1
-; AVX2-NEXT:    vpsrlw $8, %ymm1, %ymm1
-; AVX2-NEXT:    vextracti128 $1, %ymm1, %xmm2
-; AVX2-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
-; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vpand {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm2
-; AVX2-NEXT:    vpackuswb %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpor %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vzeroupper
-; AVX2-NEXT:    retq
+; AVX-LABEL: constant_rotate_v16i8:
+; AVX:       # %bb.0:
+; AVX-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm1, %xmm1
+; AVX-NEXT:    vpsrlw $8, %xmm1, %xmm1
+; AVX-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; AVX-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0
+; AVX-NEXT:    vpsrlw $8, %xmm0, %xmm0
+; AVX-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    retq
 ;
 ; AVX512F-LABEL: constant_rotate_v16i8:
 ; AVX512F:       # %bb.0:
@@ -1634,26 +1575,14 @@ define <16 x i8> @constant_rotate_v16i8(<16 x i8> %a) nounwind {
 ;
 ; X86-SSE2-LABEL: constant_rotate_v16i8:
 ; X86-SSE2:       # %bb.0:
-; X86-SSE2-NEXT:    pxor %xmm2, %xmm2
-; X86-SSE2-NEXT:    movdqa %xmm0, %xmm3
-; X86-SSE2-NEXT:    punpckhbw {{.*#+}} xmm3 = xmm3[8],xmm2[8],xmm3[9],xmm2[9],xmm3[10],xmm2[10],xmm3[11],xmm2[11],xmm3[12],xmm2[12],xmm3[13],xmm2[13],xmm3[14],xmm2[14],xmm3[15],xmm2[15]
-; X86-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm3
-; X86-SSE2-NEXT:    psrlw $8, %xmm3
 ; X86-SSE2-NEXT:    movdqa %xmm0, %xmm1
-; X86-SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
+; X86-SSE2-NEXT:    punpckhbw {{.*#+}} xmm1 = xmm1[8],xmm0[8],xmm1[9],xmm0[9],xmm1[10],xmm0[10],xmm1[11],xmm0[11],xmm1[12],xmm0[12],xmm1[13],xmm0[13],xmm1[14],xmm0[14],xmm1[15],xmm0[15]
 ; X86-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm1
 ; X86-SSE2-NEXT:    psrlw $8, %xmm1
-; X86-SSE2-NEXT:    packuswb %xmm3, %xmm1
-; X86-SSE2-NEXT:    movdqa %xmm0, %xmm2
-; X86-SSE2-NEXT:    punpckhbw {{.*#+}} xmm2 = xmm2[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; X86-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm2
-; X86-SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; X86-SSE2-NEXT:    pand %xmm3, %xmm2
 ; X86-SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; X86-SSE2-NEXT:    pmullw {{\.?LCPI[0-9]+_[0-9]+}}, %xmm0
-; X86-SSE2-NEXT:    pand %xmm3, %xmm0
-; X86-SSE2-NEXT:    packuswb %xmm2, %xmm0
-; X86-SSE2-NEXT:    por %xmm1, %xmm0
+; X86-SSE2-NEXT:    psrlw $8, %xmm0
+; X86-SSE2-NEXT:    packuswb %xmm1, %xmm0
 ; X86-SSE2-NEXT:    retl
   %shl = shl <16 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1>
   %lshr = lshr <16 x i8> %a, <i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7>

diff  --git a/llvm/test/CodeGen/X86/vector-rotate-256.ll b/llvm/test/CodeGen/X86/vector-rotate-256.ll
index ce2bff99fc8da..2e2861e30cc0b 100644
--- a/llvm/test/CodeGen/X86/vector-rotate-256.ll
+++ b/llvm/test/CodeGen/X86/vector-rotate-256.ll
@@ -1215,104 +1215,56 @@ define <32 x i8> @constant_rotate_v32i8(<32 x i8> %a) nounwind {
 ; AVX1-LABEL: constant_rotate_v32i8:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpxor %xmm8, %xmm8, %xmm8
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm1[8],xmm8[8],xmm1[9],xmm8[9],xmm1[10],xmm8[10],xmm1[11],xmm8[11],xmm1[12],xmm8[12],xmm1[13],xmm8[13],xmm1[14],xmm8[14],xmm1[15],xmm8[15]
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm9 = [256,128,64,32,16,8,4,2]
-; AVX1-NEXT:    vpmullw %xmm3, %xmm9, %xmm3
-; AVX1-NEXT:    vpsrlw $8, %xmm3, %xmm3
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm5 = 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:    vmovdqa {{.*#+}} xmm6 = [256,2,4,8,16,32,64,128]
-; AVX1-NEXT:    vpmullw %xmm6, %xmm5, %xmm7
-; AVX1-NEXT:    vpsrlw $8, %xmm7, %xmm7
-; AVX1-NEXT:    vpackuswb %xmm3, %xmm7, %xmm3
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm1 = xmm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm7 = [1,128,64,32,16,8,4,2]
-; AVX1-NEXT:    vpmullw %xmm7, %xmm1, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
-; AVX1-NEXT:    vpand %xmm2, %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 = [1,128,64,32,16,8,4,2]
+; AVX1-NEXT:    vpmullw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrlw $8, %xmm2, %xmm2
+; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm1 = xmm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [1,2,4,8,16,32,64,128]
-; AVX1-NEXT:    vpmullw %xmm4, %xmm5, %xmm5
-; AVX1-NEXT:    vpand %xmm2, %xmm5, %xmm5
-; AVX1-NEXT:    vpackuswb %xmm1, %xmm5, %xmm1
-; AVX1-NEXT:    vpor %xmm3, %xmm1, %xmm1
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm3 = xmm0[8],xmm8[8],xmm0[9],xmm8[9],xmm0[10],xmm8[10],xmm0[11],xmm8[11],xmm0[12],xmm8[12],xmm0[13],xmm8[13],xmm0[14],xmm8[14],xmm0[15],xmm8[15]
-; AVX1-NEXT:    vpmullw %xmm3, %xmm9, %xmm3
-; AVX1-NEXT:    vpsrlw $8, %xmm3, %xmm3
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm5 = 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 %xmm6, %xmm5, %xmm6
-; AVX1-NEXT:    vpsrlw $8, %xmm6, %xmm6
-; AVX1-NEXT:    vpackuswb %xmm3, %xmm6, %xmm3
-; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm0 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
-; AVX1-NEXT:    vpmullw %xmm7, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpmullw %xmm4, %xmm5, %xmm4
-; AVX1-NEXT:    vpand %xmm2, %xmm4, %xmm2
-; AVX1-NEXT:    vpackuswb %xmm0, %xmm2, %xmm0
-; AVX1-NEXT:    vpor %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpmullw %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpsrlw $8, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpckhbw {{.*#+}} xmm2 = xmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15]
+; AVX1-NEXT:    vpmullw %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpsrlw $8, %xmm2, %xmm2
+; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; AVX1-NEXT:    vpmullw %xmm4, %xmm0, %xmm0
+; AVX1-NEXT:    vpsrlw $8, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: constant_rotate_v32i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX2-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31]
-; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX2-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX2-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23]
+; AVX2-NEXT:    vpunpckhbw {{.*#+}} ymm1 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
 ; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm1, %ymm1
 ; AVX2-NEXT:    vpsrlw $8, %ymm1, %ymm1
-; AVX2-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
-; AVX2-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX2-NEXT:    vpand %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
 ; AVX2-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vpand %ymm3, %ymm0, %ymm0
-; AVX2-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
-; AVX2-NEXT:    vpor %ymm1, %ymm0, %ymm0
+; AVX2-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; AVX2-NEXT:    vpackuswb %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512F-LABEL: constant_rotate_v32i8:
 ; AVX512F:       # %bb.0:
-; AVX512F-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31]
-; AVX512F-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX512F-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23]
+; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm1 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
 ; AVX512F-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm1, %ymm1
 ; AVX512F-NEXT:    vpsrlw $8, %ymm1, %ymm1
-; AVX512F-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512F-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512F-NEXT:    vpand %ymm3, %ymm2, %ymm2
 ; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
 ; AVX512F-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX512F-NEXT:    vpand %ymm3, %ymm0, %ymm0
-; AVX512F-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
-; AVX512F-NEXT:    vpor %ymm1, %ymm0, %ymm0
+; AVX512F-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; AVX512F-NEXT:    vpackuswb %ymm1, %ymm0, %ymm0
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512VL-LABEL: constant_rotate_v32i8:
 ; AVX512VL:       # %bb.0:
-; AVX512VL-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15],ymm0[24],ymm1[24],ymm0[25],ymm1[25],ymm0[26],ymm1[26],ymm0[27],ymm1[27],ymm0[28],ymm1[28],ymm0[29],ymm1[29],ymm0[30],ymm1[30],ymm0[31],ymm1[31]
-; AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX512VL-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[16],ymm1[16],ymm0[17],ymm1[17],ymm0[18],ymm1[18],ymm0[19],ymm1[19],ymm0[20],ymm1[20],ymm0[21],ymm1[21],ymm0[22],ymm1[22],ymm0[23],ymm1[23]
+; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm1 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
 ; AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm1, %ymm1
 ; AVX512VL-NEXT:    vpsrlw $8, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm2, %ymm2
-; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512VL-NEXT:    vpand %ymm3, %ymm2, %ymm2
 ; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
 ; AVX512VL-NEXT:    vpmullw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0
-; AVX512VL-NEXT:    vpand %ymm3, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpor %ymm1, %ymm0, %ymm0
+; AVX512VL-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; AVX512VL-NEXT:    vpackuswb %ymm1, %ymm0, %ymm0
 ; AVX512VL-NEXT:    retq
 ;
 ; AVX512BW-LABEL: constant_rotate_v32i8:

diff  --git a/llvm/test/CodeGen/X86/vector-rotate-512.ll b/llvm/test/CodeGen/X86/vector-rotate-512.ll
index 5d6c4136e2dc4..a54988c0870da 100644
--- a/llvm/test/CodeGen/X86/vector-rotate-512.ll
+++ b/llvm/test/CodeGen/X86/vector-rotate-512.ll
@@ -570,177 +570,93 @@ define <64 x i8> @constant_rotate_v64i8(<64 x i8> %a) nounwind {
 ; AVX512F-LABEL: constant_rotate_v64i8:
 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512F-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm1[8],ymm2[8],ymm1[9],ymm2[9],ymm1[10],ymm2[10],ymm1[11],ymm2[11],ymm1[12],ymm2[12],ymm1[13],ymm2[13],ymm1[14],ymm2[14],ymm1[15],ymm2[15],ymm1[24],ymm2[24],ymm1[25],ymm2[25],ymm1[26],ymm2[26],ymm1[27],ymm2[27],ymm1[28],ymm2[28],ymm1[29],ymm2[29],ymm1[30],ymm2[30],ymm1[31],ymm2[31]
-; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [256,128,64,32,16,8,4,2,256,128,64,32,16,8,4,2]
-; AVX512F-NEXT:    # ymm4 = mem[0,1,0,1]
-; AVX512F-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512F-NEXT:    vpsrlw $8, %ymm3, %ymm3
-; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm5 = ymm1[0],ymm2[0],ymm1[1],ymm2[1],ymm1[2],ymm2[2],ymm1[3],ymm2[3],ymm1[4],ymm2[4],ymm1[5],ymm2[5],ymm1[6],ymm2[6],ymm1[7],ymm2[7],ymm1[16],ymm2[16],ymm1[17],ymm2[17],ymm1[18],ymm2[18],ymm1[19],ymm2[19],ymm1[20],ymm2[20],ymm1[21],ymm2[21],ymm1[22],ymm2[22],ymm1[23],ymm2[23]
-; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm6 = [256,2,4,8,16,32,64,128,256,2,4,8,16,32,64,128]
-; AVX512F-NEXT:    # ymm6 = mem[0,1,0,1]
-; AVX512F-NEXT:    vpmullw %ymm6, %ymm5, %ymm5
-; AVX512F-NEXT:    vpsrlw $8, %ymm5, %ymm5
-; AVX512F-NEXT:    vpackuswb %ymm3, %ymm5, %ymm3
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm5 = ymm0[8],ymm2[8],ymm0[9],ymm2[9],ymm0[10],ymm2[10],ymm0[11],ymm2[11],ymm0[12],ymm2[12],ymm0[13],ymm2[13],ymm0[14],ymm2[14],ymm0[15],ymm2[15],ymm0[24],ymm2[24],ymm0[25],ymm2[25],ymm0[26],ymm2[26],ymm0[27],ymm2[27],ymm0[28],ymm2[28],ymm0[29],ymm2[29],ymm0[30],ymm2[30],ymm0[31],ymm2[31]
-; AVX512F-NEXT:    vpmullw %ymm4, %ymm5, %ymm4
-; AVX512F-NEXT:    vpsrlw $8, %ymm4, %ymm4
-; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm2 = ymm0[0],ymm2[0],ymm0[1],ymm2[1],ymm0[2],ymm2[2],ymm0[3],ymm2[3],ymm0[4],ymm2[4],ymm0[5],ymm2[5],ymm0[6],ymm2[6],ymm0[7],ymm2[7],ymm0[16],ymm2[16],ymm0[17],ymm2[17],ymm0[18],ymm2[18],ymm0[19],ymm2[19],ymm0[20],ymm2[20],ymm0[21],ymm2[21],ymm0[22],ymm2[22],ymm0[23],ymm2[23]
-; AVX512F-NEXT:    vpmullw %ymm6, %ymm2, %ymm2
+; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
+; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm3 = [1,128,64,32,16,8,4,2,1,128,64,32,16,8,4,2]
+; AVX512F-NEXT:    # ymm3 = mem[0,1,0,1]
+; AVX512F-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
 ; AVX512F-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX512F-NEXT:    vpackuswb %ymm4, %ymm2, %ymm2
-; AVX512F-NEXT:    vinserti64x4 $1, %ymm3, %zmm2, %zmm2
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [1,128,64,32,16,8,4,2,1,128,64,32,16,8,4,2]
-; AVX512F-NEXT:    # ymm4 = mem[0,1,0,1]
-; AVX512F-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm5 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512F-NEXT:    vpand %ymm5, %ymm3, %ymm3
 ; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
-; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm6 = [1,2,4,8,16,32,64,128,1,2,4,8,16,32,64,128]
-; AVX512F-NEXT:    # ymm6 = mem[0,1,0,1]
-; AVX512F-NEXT:    vpmullw %ymm6, %ymm1, %ymm1
-; AVX512F-NEXT:    vpand %ymm5, %ymm1, %ymm1
-; AVX512F-NEXT:    vpackuswb %ymm3, %ymm1, %ymm1
-; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512F-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512F-NEXT:    vpand %ymm5, %ymm3, %ymm3
+; AVX512F-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [1,2,4,8,16,32,64,128,1,2,4,8,16,32,64,128]
+; AVX512F-NEXT:    # ymm4 = mem[0,1,0,1]
+; AVX512F-NEXT:    vpmullw %ymm4, %ymm1, %ymm1
+; AVX512F-NEXT:    vpsrlw $8, %ymm1, %ymm1
+; AVX512F-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
+; AVX512F-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
+; AVX512F-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
+; AVX512F-NEXT:    vpsrlw $8, %ymm2, %ymm2
 ; AVX512F-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
-; AVX512F-NEXT:    vpmullw %ymm6, %ymm0, %ymm0
-; AVX512F-NEXT:    vpand %ymm5, %ymm0, %ymm0
-; AVX512F-NEXT:    vpackuswb %ymm3, %ymm0, %ymm0
+; AVX512F-NEXT:    vpmullw %ymm4, %ymm0, %ymm0
+; AVX512F-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; AVX512F-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
 ; AVX512F-NEXT:    vinserti64x4 $1, %ymm1, %zmm0, %zmm0
-; AVX512F-NEXT:    vporq %zmm2, %zmm0, %zmm0
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512VL-LABEL: constant_rotate_v64i8:
 ; AVX512VL:       # %bb.0:
 ; AVX512VL-NEXT:    vextracti64x4 $1, %zmm0, %ymm1
-; AVX512VL-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm1[8],ymm2[8],ymm1[9],ymm2[9],ymm1[10],ymm2[10],ymm1[11],ymm2[11],ymm1[12],ymm2[12],ymm1[13],ymm2[13],ymm1[14],ymm2[14],ymm1[15],ymm2[15],ymm1[24],ymm2[24],ymm1[25],ymm2[25],ymm1[26],ymm2[26],ymm1[27],ymm2[27],ymm1[28],ymm2[28],ymm1[29],ymm2[29],ymm1[30],ymm2[30],ymm1[31],ymm2[31]
-; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [256,128,64,32,16,8,4,2,256,128,64,32,16,8,4,2]
-; AVX512VL-NEXT:    # ymm4 = mem[0,1,0,1]
-; AVX512VL-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512VL-NEXT:    vpsrlw $8, %ymm3, %ymm3
-; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm5 = ymm1[0],ymm2[0],ymm1[1],ymm2[1],ymm1[2],ymm2[2],ymm1[3],ymm2[3],ymm1[4],ymm2[4],ymm1[5],ymm2[5],ymm1[6],ymm2[6],ymm1[7],ymm2[7],ymm1[16],ymm2[16],ymm1[17],ymm2[17],ymm1[18],ymm2[18],ymm1[19],ymm2[19],ymm1[20],ymm2[20],ymm1[21],ymm2[21],ymm1[22],ymm2[22],ymm1[23],ymm2[23]
-; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm6 = [256,2,4,8,16,32,64,128,256,2,4,8,16,32,64,128]
-; AVX512VL-NEXT:    # ymm6 = mem[0,1,0,1]
-; AVX512VL-NEXT:    vpmullw %ymm6, %ymm5, %ymm5
-; AVX512VL-NEXT:    vpsrlw $8, %ymm5, %ymm5
-; AVX512VL-NEXT:    vpackuswb %ymm3, %ymm5, %ymm3
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm5 = ymm0[8],ymm2[8],ymm0[9],ymm2[9],ymm0[10],ymm2[10],ymm0[11],ymm2[11],ymm0[12],ymm2[12],ymm0[13],ymm2[13],ymm0[14],ymm2[14],ymm0[15],ymm2[15],ymm0[24],ymm2[24],ymm0[25],ymm2[25],ymm0[26],ymm2[26],ymm0[27],ymm2[27],ymm0[28],ymm2[28],ymm0[29],ymm2[29],ymm0[30],ymm2[30],ymm0[31],ymm2[31]
-; AVX512VL-NEXT:    vpmullw %ymm4, %ymm5, %ymm4
-; AVX512VL-NEXT:    vpsrlw $8, %ymm4, %ymm4
-; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm2 = ymm0[0],ymm2[0],ymm0[1],ymm2[1],ymm0[2],ymm2[2],ymm0[3],ymm2[3],ymm0[4],ymm2[4],ymm0[5],ymm2[5],ymm0[6],ymm2[6],ymm0[7],ymm2[7],ymm0[16],ymm2[16],ymm0[17],ymm2[17],ymm0[18],ymm2[18],ymm0[19],ymm2[19],ymm0[20],ymm2[20],ymm0[21],ymm2[21],ymm0[22],ymm2[22],ymm0[23],ymm2[23]
-; AVX512VL-NEXT:    vpmullw %ymm6, %ymm2, %ymm2
+; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
+; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm3 = [1,128,64,32,16,8,4,2,1,128,64,32,16,8,4,2]
+; AVX512VL-NEXT:    # ymm3 = mem[0,1,0,1]
+; AVX512VL-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
 ; AVX512VL-NEXT:    vpsrlw $8, %ymm2, %ymm2
-; AVX512VL-NEXT:    vpackuswb %ymm4, %ymm2, %ymm2
-; AVX512VL-NEXT:    vinserti64x4 $1, %ymm3, %zmm2, %zmm2
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm1[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [1,128,64,32,16,8,4,2,1,128,64,32,16,8,4,2]
-; AVX512VL-NEXT:    # ymm4 = mem[0,1,0,1]
-; AVX512VL-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm5 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512VL-NEXT:    vpand %ymm5, %ymm3, %ymm3
 ; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm1 = ymm1[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
-; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm6 = [1,2,4,8,16,32,64,128,1,2,4,8,16,32,64,128]
-; AVX512VL-NEXT:    # ymm6 = mem[0,1,0,1]
-; AVX512VL-NEXT:    vpmullw %ymm6, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpand %ymm5, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpackuswb %ymm3, %ymm1, %ymm1
-; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm3 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
-; AVX512VL-NEXT:    vpmullw %ymm4, %ymm3, %ymm3
-; AVX512VL-NEXT:    vpand %ymm5, %ymm3, %ymm3
+; AVX512VL-NEXT:    vbroadcasti128 {{.*#+}} ymm4 = [1,2,4,8,16,32,64,128,1,2,4,8,16,32,64,128]
+; AVX512VL-NEXT:    # ymm4 = mem[0,1,0,1]
+; AVX512VL-NEXT:    vpmullw %ymm4, %ymm1, %ymm1
+; AVX512VL-NEXT:    vpsrlw $8, %ymm1, %ymm1
+; AVX512VL-NEXT:    vpackuswb %ymm2, %ymm1, %ymm1
+; AVX512VL-NEXT:    vpunpckhbw {{.*#+}} ymm2 = ymm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31]
+; AVX512VL-NEXT:    vpmullw %ymm3, %ymm2, %ymm2
+; AVX512VL-NEXT:    vpsrlw $8, %ymm2, %ymm2
 ; AVX512VL-NEXT:    vpunpcklbw {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23]
-; AVX512VL-NEXT:    vpmullw %ymm6, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpand %ymm5, %ymm0, %ymm0
-; AVX512VL-NEXT:    vpackuswb %ymm3, %ymm0, %ymm0
+; AVX512VL-NEXT:    vpmullw %ymm4, %ymm0, %ymm0
+; AVX512VL-NEXT:    vpsrlw $8, %ymm0, %ymm0
+; AVX512VL-NEXT:    vpackuswb %ymm2, %ymm0, %ymm0
 ; AVX512VL-NEXT:    vinserti64x4 $1, %ymm1, %zmm0, %zmm0
-; AVX512VL-NEXT:    vporq %zmm2, %zmm0, %zmm0
 ; AVX512VL-NEXT:    retq
 ;
 ; AVX512BW-LABEL: constant_rotate_v64i8:
 ; AVX512BW:       # %bb.0:
-; AVX512BW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
-; AVX512BW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512BW-NEXT:    vpsrlw $8, %zmm2, %zmm2
-; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
+; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm1 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512BW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm1, %zmm1
 ; AVX512BW-NEXT:    vpsrlw $8, %zmm1, %zmm1
-; AVX512BW-NEXT:    vpackuswb %zmm2, %zmm1, %zmm1
-; AVX512BW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512BW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512BW-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512BW-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512BW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512BW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
-; AVX512BW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
-; AVX512BW-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
-; AVX512BW-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpsrlw $8, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512VLBW-LABEL: constant_rotate_v64i8:
 ; AVX512VLBW:       # %bb.0:
-; AVX512VLBW-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512VLBW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
-; AVX512VLBW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VLBW-NEXT:    vpsrlw $8, %zmm2, %zmm2
-; AVX512VLBW-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
+; AVX512VLBW-NEXT:    vpunpckhbw {{.*#+}} zmm1 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512VLBW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm1, %zmm1
 ; AVX512VLBW-NEXT:    vpsrlw $8, %zmm1, %zmm1
-; AVX512VLBW-NEXT:    vpackuswb %zmm2, %zmm1, %zmm1
-; AVX512VLBW-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512VLBW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VLBW-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512VLBW-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512VLBW-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512VLBW-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
-; AVX512VLBW-NEXT:    vpandq %zmm3, %zmm0, %zmm0
-; AVX512VLBW-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
-; AVX512VLBW-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512VLBW-NEXT:    vpsrlw $8, %zmm0, %zmm0
+; AVX512VLBW-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512VLBW-NEXT:    retq
 ;
 ; AVX512VBMI2-LABEL: constant_rotate_v64i8:
 ; AVX512VBMI2:       # %bb.0:
-; AVX512VBMI2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512VBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
-; AVX512VBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VBMI2-NEXT:    vpsrlw $8, %zmm2, %zmm2
-; AVX512VBMI2-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
+; AVX512VBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm1 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512VBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm1, %zmm1
 ; AVX512VBMI2-NEXT:    vpsrlw $8, %zmm1, %zmm1
-; AVX512VBMI2-NEXT:    vpackuswb %zmm2, %zmm1, %zmm1
-; AVX512VBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512VBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VBMI2-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512VBMI2-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512VBMI2-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512VBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
-; AVX512VBMI2-NEXT:    vpandq %zmm3, %zmm0, %zmm0
-; AVX512VBMI2-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
-; AVX512VBMI2-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512VBMI2-NEXT:    vpsrlw $8, %zmm0, %zmm0
+; AVX512VBMI2-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512VBMI2-NEXT:    retq
 ;
 ; AVX512VLVBMI2-LABEL: constant_rotate_v64i8:
 ; AVX512VLVBMI2:       # %bb.0:
-; AVX512VLVBMI2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512VLVBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8],zmm1[8],zmm0[9],zmm1[9],zmm0[10],zmm1[10],zmm0[11],zmm1[11],zmm0[12],zmm1[12],zmm0[13],zmm1[13],zmm0[14],zmm1[14],zmm0[15],zmm1[15],zmm0[24],zmm1[24],zmm0[25],zmm1[25],zmm0[26],zmm1[26],zmm0[27],zmm1[27],zmm0[28],zmm1[28],zmm0[29],zmm1[29],zmm0[30],zmm1[30],zmm0[31],zmm1[31],zmm0[40],zmm1[40],zmm0[41],zmm1[41],zmm0[42],zmm1[42],zmm0[43],zmm1[43],zmm0[44],zmm1[44],zmm0[45],zmm1[45],zmm0[46],zmm1[46],zmm0[47],zmm1[47],zmm0[56],zmm1[56],zmm0[57],zmm1[57],zmm0[58],zmm1[58],zmm0[59],zmm1[59],zmm0[60],zmm1[60],zmm0[61],zmm1[61],zmm0[62],zmm1[62],zmm0[63],zmm1[63]
-; AVX512VLVBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VLVBMI2-NEXT:    vpsrlw $8, %zmm2, %zmm2
-; AVX512VLVBMI2-NEXT:    vpunpcklbw {{.*#+}} zmm1 = zmm0[0],zmm1[0],zmm0[1],zmm1[1],zmm0[2],zmm1[2],zmm0[3],zmm1[3],zmm0[4],zmm1[4],zmm0[5],zmm1[5],zmm0[6],zmm1[6],zmm0[7],zmm1[7],zmm0[16],zmm1[16],zmm0[17],zmm1[17],zmm0[18],zmm1[18],zmm0[19],zmm1[19],zmm0[20],zmm1[20],zmm0[21],zmm1[21],zmm0[22],zmm1[22],zmm0[23],zmm1[23],zmm0[32],zmm1[32],zmm0[33],zmm1[33],zmm0[34],zmm1[34],zmm0[35],zmm1[35],zmm0[36],zmm1[36],zmm0[37],zmm1[37],zmm0[38],zmm1[38],zmm0[39],zmm1[39],zmm0[48],zmm1[48],zmm0[49],zmm1[49],zmm0[50],zmm1[50],zmm0[51],zmm1[51],zmm0[52],zmm1[52],zmm0[53],zmm1[53],zmm0[54],zmm1[54],zmm0[55],zmm1[55]
+; AVX512VLVBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm1 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
 ; AVX512VLVBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm1, %zmm1
 ; AVX512VLVBMI2-NEXT:    vpsrlw $8, %zmm1, %zmm1
-; AVX512VLVBMI2-NEXT:    vpackuswb %zmm2, %zmm1, %zmm1
-; AVX512VLVBMI2-NEXT:    vpunpckhbw {{.*#+}} zmm2 = zmm0[8,8,9,9,10,10,11,11,12,12,13,13,14,14,15,15,24,24,25,25,26,26,27,27,28,28,29,29,30,30,31,31,40,40,41,41,42,42,43,43,44,44,45,45,46,46,47,47,56,56,57,57,58,58,59,59,60,60,61,61,62,62,63,63]
-; AVX512VLVBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm2, %zmm2
-; AVX512VLVBMI2-NEXT:    vmovdqa64 {{.*#+}} zmm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX512VLVBMI2-NEXT:    vpandq %zmm3, %zmm2, %zmm2
 ; AVX512VLVBMI2-NEXT:    vpunpcklbw {{.*#+}} zmm0 = zmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7,16,16,17,17,18,18,19,19,20,20,21,21,22,22,23,23,32,32,33,33,34,34,35,35,36,36,37,37,38,38,39,39,48,48,49,49,50,50,51,51,52,52,53,53,54,54,55,55]
 ; AVX512VLVBMI2-NEXT:    vpsllvw {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %zmm0, %zmm0
-; AVX512VLVBMI2-NEXT:    vpandq %zmm3, %zmm0, %zmm0
-; AVX512VLVBMI2-NEXT:    vpackuswb %zmm2, %zmm0, %zmm0
-; AVX512VLVBMI2-NEXT:    vporq %zmm1, %zmm0, %zmm0
+; AVX512VLVBMI2-NEXT:    vpsrlw $8, %zmm0, %zmm0
+; AVX512VLVBMI2-NEXT:    vpackuswb %zmm1, %zmm0, %zmm0
 ; AVX512VLVBMI2-NEXT:    retq
   %shl = shl <64 x i8> %a, <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1>
   %lshr = lshr <64 x i8> %a, <i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 7, i8 6, i8 5, i8 4, i8 3, i8 2, i8 1, i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7>


        


More information about the llvm-commits mailing list