diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index b75e49e..57d27ed 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -6609,20 +6609,26 @@
 }
 
 /// Create a shuffle mask that matches the PACKSS/PACKUS truncation.
+/// A multi-stage pack shuffle mask is created by specifying NumStages > 1.
 /// Note: This ignores saturation, so inputs must be checked first.
 static void createPackShuffleMask(MVT VT, SmallVectorImpl<int> &Mask,
-                                  bool Unary) {
+                                  bool Unary, unsigned NumStages = 1) {
   assert(Mask.empty() && "Expected an empty shuffle mask vector");
   unsigned NumElts = VT.getVectorNumElements();
   unsigned NumLanes = VT.getSizeInBits() / 128;
   unsigned NumEltsPerLane = 128 / VT.getScalarSizeInBits();
   unsigned Offset = Unary ? 0 : NumElts;
+  unsigned Repetitions = 1u << (NumStages - 1);
+  unsigned Increment = 1u << NumStages;
+  assert((NumEltsPerLane >> NumStages) > 0 && "Illegal packing compaction");
 
   for (unsigned Lane = 0; Lane != NumLanes; ++Lane) {
-    for (unsigned Elt = 0; Elt != NumEltsPerLane; Elt += 2)
-      Mask.push_back(Elt + (Lane * NumEltsPerLane));
-    for (unsigned Elt = 0; Elt != NumEltsPerLane; Elt += 2)
-      Mask.push_back(Elt + (Lane * NumEltsPerLane) + Offset);
+    for (unsigned Stage = 0; Stage != Repetitions; ++Stage) {
+      for (unsigned Elt = 0; Elt != NumEltsPerLane; Elt += Increment)
+        Mask.push_back(Elt + (Lane * NumEltsPerLane));
+      for (unsigned Elt = 0; Elt != NumEltsPerLane; Elt += Increment)
+        Mask.push_back(Elt + (Lane * NumEltsPerLane) + Offset);
+    }
   }
 }
 
@@ -11328,13 +11334,17 @@
 
 // X86 has dedicated pack instructions that can handle specific truncation
 // operations: PACKSS and PACKUS.
+// Checks for compaction shuffle masks if MaxStages > 1.
 // TODO: Add support for matching multiple PACKSS/PACKUS stages.
 static bool matchShuffleWithPACK(MVT VT, MVT &SrcVT, SDValue &V1, SDValue &V2,
                                  unsigned &PackOpcode, ArrayRef<int> TargetMask,
                                  SelectionDAG &DAG,
-                                 const X86Subtarget &Subtarget) {
+                                 const X86Subtarget &Subtarget,
+                                 unsigned MaxStages = 1) {
   unsigned NumElts = VT.getVectorNumElements();
   unsigned BitSize = VT.getScalarSizeInBits();
+  assert(0 < MaxStages && MaxStages <= 3 && (BitSize << MaxStages) <= 64 &&
+         "Illegal maximum compaction");
 
   auto MatchPACK = [&](SDValue N1, SDValue N2, MVT PackVT) {
     unsigned NumSrcBits = PackVT.getScalarSizeInBits();
@@ -11363,22 +11373,25 @@
     return false;
   };
 
-  MVT PackSVT = MVT::getIntegerVT(BitSize * 2);
-  MVT PackVT = MVT::getVectorVT(PackSVT, NumElts / 2);
+  // Attempt to match against wider and wider compaction patterns.
+  for (unsigned NumStages = 1; NumStages <= MaxStages; ++NumStages) {
+    MVT PackSVT = MVT::getIntegerVT(BitSize << NumStages);
+    MVT PackVT = MVT::getVectorVT(PackSVT, NumElts >> NumStages);
 
-  // Try binary shuffle.
-  SmallVector<int, 32> BinaryMask;
-  createPackShuffleMask(VT, BinaryMask, false);
-  if (isTargetShuffleEquivalent(TargetMask, BinaryMask, V1, V2))
-    if (MatchPACK(V1, V2, PackVT))
-      return true;
+    // Try binary shuffle.
+    SmallVector<int, 32> BinaryMask;
+    createPackShuffleMask(VT, BinaryMask, false, NumStages);
+    if (isTargetShuffleEquivalent(TargetMask, BinaryMask, V1, V2))
+      if (MatchPACK(V1, V2, PackVT))
+        return true;
 
-  // Try unary shuffle.
-  SmallVector<int, 32> UnaryMask;
-  createPackShuffleMask(VT, UnaryMask, true);
-  if (isTargetShuffleEquivalent(TargetMask, UnaryMask, V1))
-    if (MatchPACK(V1, V1, PackVT))
-      return true;
+    // Try unary shuffle.
+    SmallVector<int, 32> UnaryMask;
+    createPackShuffleMask(VT, UnaryMask, true, NumStages);
+    if (isTargetShuffleEquivalent(TargetMask, UnaryMask, V1))
+      if (MatchPACK(V1, V1, PackVT))
+        return true;
+  }
 
   return false;
 }
@@ -11388,12 +11401,44 @@
                                     const X86Subtarget &Subtarget) {
   MVT PackVT;
   unsigned PackOpcode;
-  if (matchShuffleWithPACK(VT, PackVT, V1, V2, PackOpcode, Mask, DAG,
-                           Subtarget))
-    return DAG.getNode(PackOpcode, DL, VT, DAG.getBitcast(PackVT, V1),
-                       DAG.getBitcast(PackVT, V2));
+  unsigned SizeBits = VT.getSizeInBits();
+  unsigned EltBits = VT.getScalarSizeInBits();
+  unsigned MaxStages = Log2_32(64 / EltBits);
+  if (!matchShuffleWithPACK(VT, PackVT, V1, V2, PackOpcode, Mask, DAG,
+                            Subtarget, MaxStages))
+    return SDValue();
 
-  return SDValue();
+  unsigned CurrentEltBits = PackVT.getScalarSizeInBits();
+  unsigned NumStages = Log2_32(CurrentEltBits / EltBits);
+
+  // Don't lower multi-stage packs on AVX512, truncation is better.
+  if (NumStages != 1 && SizeBits == 128 && Subtarget.hasVLX())
+    return SDValue();
+
+  // Pack to the largest type possible:
+  // vXi64/vXi32 -> PACK*SDW and vXi16 -> PACK*SWB.
+  unsigned MaxPackBits = 16;
+  if (CurrentEltBits > 16 &&
+      (PackOpcode == X86ISD::PACKSS || Subtarget.hasSSE41()))
+    MaxPackBits = 32;
+
+  // Repeatedly pack down to the target size.
+  SDValue Res;
+  for (unsigned i = 0; i != NumStages; ++i) {
+    unsigned SrcEltBits = std::min(MaxPackBits, CurrentEltBits);
+    unsigned NumSrcElts = SizeBits / SrcEltBits;
+    MVT SrcSVT = MVT::getIntegerVT(SrcEltBits);
+    MVT DstSVT = MVT::getIntegerVT(SrcEltBits / 2);
+    MVT SrcVT = MVT::getVectorVT(SrcSVT, NumSrcElts);
+    MVT DstVT = MVT::getVectorVT(DstSVT, NumSrcElts * 2);
+    Res = DAG.getNode(PackOpcode, DL, DstVT, DAG.getBitcast(SrcVT, V1),
+                      DAG.getBitcast(SrcVT, V2));
+    V1 = V2 = Res;
+    CurrentEltBits /= 2;
+  }
+  assert(Res && Res.getValueType() == VT &&
+         "Failed to lower compaction shuffle");
+  return Res;
 }
 
 /// Try to emit a bitmask instruction for a shuffle.
diff --git a/llvm/test/CodeGen/X86/avx-fp2int.ll b/llvm/test/CodeGen/X86/avx-fp2int.ll
index ac5fcfe..195b6f6 100644
--- a/llvm/test/CodeGen/X86/avx-fp2int.ll
+++ b/llvm/test/CodeGen/X86/avx-fp2int.ll
@@ -7,7 +7,8 @@
 ; CHECK-LABEL: test1:
 ; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    vcvttpd2dq %ymm0, %xmm0
-; CHECK-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; CHECK-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; CHECK-NEXT:    vzeroupper
 ; CHECK-NEXT:    retl
   %c = fptoui <4 x double> %d to <4 x i8>
@@ -17,7 +18,8 @@
 ; CHECK-LABEL: test2:
 ; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    vcvttpd2dq %ymm0, %xmm0
-; CHECK-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; CHECK-NEXT:    vpackssdw %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
 ; CHECK-NEXT:    vzeroupper
 ; CHECK-NEXT:    retl
   %c = fptosi <4 x double> %d to <4 x i8>
diff --git a/llvm/test/CodeGen/X86/bitcast-and-setcc-512.ll b/llvm/test/CodeGen/X86/bitcast-and-setcc-512.ll
index 3852b0e..588eddc 100644
--- a/llvm/test/CodeGen/X86/bitcast-and-setcc-512.ll
+++ b/llvm/test/CodeGen/X86/bitcast-and-setcc-512.ll
@@ -13,38 +13,26 @@
 ; SSE-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm10
 ; SSE-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm11
 ; SSE-NEXT:    pcmpgtq %xmm7, %xmm3
-; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[0,1,0,2,4,5,6,7]
 ; SSE-NEXT:    pcmpgtq %xmm6, %xmm2
-; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,0,2,4,5,6,7]
-; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
+; SSE-NEXT:    packssdw %xmm3, %xmm2
+; SSE-NEXT:    packssdw %xmm2, %xmm2
 ; SSE-NEXT:    pcmpgtq %xmm5, %xmm1
-; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
 ; SSE-NEXT:    pcmpgtq %xmm4, %xmm0
-; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
-; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE-NEXT:    packssdw %xmm1, %xmm0
+; SSE-NEXT:    packssdw %xmm0, %xmm0
 ; SSE-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm2[4,5,6,7]
 ; SSE-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm11
-; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm11[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,1,0,2,4,5,6,7]
 ; SSE-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm10
-; SSE-NEXT:    pshufd {{.*#+}} xmm2 = xmm10[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,0,2,4,5,6,7]
-; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
+; SSE-NEXT:    packssdw %xmm11, %xmm10
+; SSE-NEXT:    packssdw %xmm10, %xmm1
 ; SSE-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm9
-; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm9[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
 ; SSE-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm8
-; SSE-NEXT:    pshufd {{.*#+}} xmm3 = xmm8[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[0,2,2,3,4,5,6,7]
-; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
-; SSE-NEXT:    pblendw {{.*#+}} xmm3 = xmm3[0,1,2,3],xmm2[4,5,6,7]
-; SSE-NEXT:    pand %xmm0, %xmm3
-; SSE-NEXT:    packsswb %xmm0, %xmm3
-; SSE-NEXT:    pmovmskb %xmm3, %eax
+; SSE-NEXT:    packssdw %xmm9, %xmm8
+; SSE-NEXT:    packssdw %xmm0, %xmm8
+; SSE-NEXT:    pblendw {{.*#+}} xmm8 = xmm8[0,1,2,3],xmm1[4,5,6,7]
+; SSE-NEXT:    pand %xmm0, %xmm8
+; SSE-NEXT:    packsswb %xmm0, %xmm8
+; SSE-NEXT:    pmovmskb %xmm8, %eax
 ; SSE-NEXT:    # kill: def $al killed $al killed $eax
 ; SSE-NEXT:    retq
 ;
@@ -125,38 +113,26 @@
 ; SSE-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm10
 ; SSE-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm11
 ; SSE-NEXT:    cmpltpd %xmm3, %xmm7
-; SSE-NEXT:    shufps {{.*#+}} xmm7 = xmm7[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm7[0,1,0,2,4,5,6,7]
 ; SSE-NEXT:    cmpltpd %xmm2, %xmm6
-; SSE-NEXT:    shufps {{.*#+}} xmm6 = xmm6[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm6[0,1,0,2,4,5,6,7]
-; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
+; SSE-NEXT:    packssdw %xmm7, %xmm6
+; SSE-NEXT:    packssdw %xmm6, %xmm2
 ; SSE-NEXT:    cmpltpd %xmm1, %xmm5
-; SSE-NEXT:    shufps {{.*#+}} xmm5 = xmm5[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm5[0,2,2,3,4,5,6,7]
 ; SSE-NEXT:    cmpltpd %xmm0, %xmm4
-; SSE-NEXT:    shufps {{.*#+}} xmm4 = xmm4[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm4[0,2,2,3,4,5,6,7]
-; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; SSE-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm2[4,5,6,7]
+; SSE-NEXT:    packssdw %xmm5, %xmm4
+; SSE-NEXT:    packssdw %xmm0, %xmm4
+; SSE-NEXT:    pblendw {{.*#+}} xmm4 = xmm4[0,1,2,3],xmm2[4,5,6,7]
 ; SSE-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm11
-; SSE-NEXT:    shufps {{.*#+}} xmm11 = xmm11[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm11[0,1,0,2,4,5,6,7]
 ; SSE-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm10
-; SSE-NEXT:    shufps {{.*#+}} xmm10 = xmm10[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm2 = xmm10[0,1,0,2,4,5,6,7]
-; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
+; SSE-NEXT:    packssdw %xmm11, %xmm10
+; SSE-NEXT:    packssdw %xmm10, %xmm0
 ; SSE-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm9
-; SSE-NEXT:    shufps {{.*#+}} xmm9 = xmm9[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm9[0,2,2,3,4,5,6,7]
 ; SSE-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm8
-; SSE-NEXT:    shufps {{.*#+}} xmm8 = xmm8[0,2,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm3 = xmm8[0,2,2,3,4,5,6,7]
-; SSE-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1]
-; SSE-NEXT:    pblendw {{.*#+}} xmm3 = xmm3[0,1,2,3],xmm2[4,5,6,7]
-; SSE-NEXT:    pand %xmm0, %xmm3
-; SSE-NEXT:    packsswb %xmm0, %xmm3
-; SSE-NEXT:    pmovmskb %xmm3, %eax
+; SSE-NEXT:    packssdw %xmm9, %xmm8
+; SSE-NEXT:    packssdw %xmm0, %xmm8
+; SSE-NEXT:    pblendw {{.*#+}} xmm8 = xmm8[0,1,2,3],xmm0[4,5,6,7]
+; SSE-NEXT:    pand %xmm4, %xmm8
+; SSE-NEXT:    packsswb %xmm0, %xmm8
+; SSE-NEXT:    pmovmskb %xmm8, %eax
 ; SSE-NEXT:    # kill: def $al killed $al killed $eax
 ; SSE-NEXT:    retq
 ;
@@ -337,33 +313,23 @@
 ; SSE-LABEL: v16i32:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm8
-; SSE-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm10
 ; SSE-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm9
+; SSE-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm10
 ; SSE-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm11
 ; SSE-NEXT:    pcmpgtd %xmm7, %xmm3
-; SSE-NEXT:    movdqa {{.*#+}} xmm7 = <u,u,u,u,0,4,8,12,u,u,u,u,u,u,u,u>
-; SSE-NEXT:    pshufb %xmm7, %xmm3
 ; SSE-NEXT:    pcmpgtd %xmm6, %xmm2
-; SSE-NEXT:    pshufb %xmm7, %xmm2
-; SSE-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
+; SSE-NEXT:    packssdw %xmm3, %xmm2
 ; SSE-NEXT:    pcmpgtd %xmm5, %xmm1
-; SSE-NEXT:    movdqa {{.*#+}} xmm3 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
-; SSE-NEXT:    pshufb %xmm3, %xmm1
 ; SSE-NEXT:    pcmpgtd %xmm4, %xmm0
-; SSE-NEXT:    pshufb %xmm3, %xmm0
-; SSE-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
-; SSE-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm2[4,5,6,7]
+; SSE-NEXT:    packssdw %xmm1, %xmm0
+; SSE-NEXT:    packsswb %xmm2, %xmm0
 ; SSE-NEXT:    pcmpgtd {{[0-9]+}}(%rsp), %xmm11
-; SSE-NEXT:    pshufb %xmm7, %xmm11
-; SSE-NEXT:    pcmpgtd {{[0-9]+}}(%rsp), %xmm9
-; SSE-NEXT:    pshufb %xmm7, %xmm9
-; SSE-NEXT:    punpckldq {{.*#+}} xmm9 = xmm9[0],xmm11[0],xmm9[1],xmm11[1]
 ; SSE-NEXT:    pcmpgtd {{[0-9]+}}(%rsp), %xmm10
-; SSE-NEXT:    pshufb %xmm3, %xmm10
+; SSE-NEXT:    packssdw %xmm11, %xmm10
+; SSE-NEXT:    pcmpgtd {{[0-9]+}}(%rsp), %xmm9
 ; SSE-NEXT:    pcmpgtd {{[0-9]+}}(%rsp), %xmm8
-; SSE-NEXT:    pshufb %xmm3, %xmm8
-; SSE-NEXT:    punpckldq {{.*#+}} xmm8 = xmm8[0],xmm10[0],xmm8[1],xmm10[1]
-; SSE-NEXT:    pblendw {{.*#+}} xmm8 = xmm8[0,1,2,3],xmm9[4,5,6,7]
+; SSE-NEXT:    packssdw %xmm9, %xmm8
+; SSE-NEXT:    packsswb %xmm10, %xmm8
 ; SSE-NEXT:    pand %xmm0, %xmm8
 ; SSE-NEXT:    pmovmskb %xmm8, %eax
 ; SSE-NEXT:    # kill: def $ax killed $ax killed $eax
@@ -449,33 +415,23 @@
 ; SSE-LABEL: v16f32:
 ; SSE:       # %bb.0:
 ; SSE-NEXT:    movaps {{[0-9]+}}(%rsp), %xmm8
-; SSE-NEXT:    movaps {{[0-9]+}}(%rsp), %xmm10
 ; SSE-NEXT:    movaps {{[0-9]+}}(%rsp), %xmm9
+; SSE-NEXT:    movaps {{[0-9]+}}(%rsp), %xmm10
 ; SSE-NEXT:    movaps {{[0-9]+}}(%rsp), %xmm11
 ; SSE-NEXT:    cmpltps %xmm3, %xmm7
-; SSE-NEXT:    movdqa {{.*#+}} xmm3 = <u,u,u,u,0,4,8,12,u,u,u,u,u,u,u,u>
-; SSE-NEXT:    pshufb %xmm3, %xmm7
 ; SSE-NEXT:    cmpltps %xmm2, %xmm6
-; SSE-NEXT:    pshufb %xmm3, %xmm6
-; SSE-NEXT:    punpckldq {{.*#+}} xmm6 = xmm6[0],xmm7[0],xmm6[1],xmm7[1]
+; SSE-NEXT:    packssdw %xmm7, %xmm6
 ; SSE-NEXT:    cmpltps %xmm1, %xmm5
-; SSE-NEXT:    movdqa {{.*#+}} xmm1 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
-; SSE-NEXT:    pshufb %xmm1, %xmm5
 ; SSE-NEXT:    cmpltps %xmm0, %xmm4
-; SSE-NEXT:    pshufb %xmm1, %xmm4
-; SSE-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm5[0],xmm4[1],xmm5[1]
-; SSE-NEXT:    pblendw {{.*#+}} xmm4 = xmm4[0,1,2,3],xmm6[4,5,6,7]
+; SSE-NEXT:    packssdw %xmm5, %xmm4
+; SSE-NEXT:    packsswb %xmm6, %xmm4
 ; SSE-NEXT:    cmpltps {{[0-9]+}}(%rsp), %xmm11
-; SSE-NEXT:    pshufb %xmm3, %xmm11
-; SSE-NEXT:    cmpltps {{[0-9]+}}(%rsp), %xmm9
-; SSE-NEXT:    pshufb %xmm3, %xmm9
-; SSE-NEXT:    punpckldq {{.*#+}} xmm9 = xmm9[0],xmm11[0],xmm9[1],xmm11[1]
 ; SSE-NEXT:    cmpltps {{[0-9]+}}(%rsp), %xmm10
-; SSE-NEXT:    pshufb %xmm1, %xmm10
+; SSE-NEXT:    packssdw %xmm11, %xmm10
+; SSE-NEXT:    cmpltps {{[0-9]+}}(%rsp), %xmm9
 ; SSE-NEXT:    cmpltps {{[0-9]+}}(%rsp), %xmm8
-; SSE-NEXT:    pshufb %xmm1, %xmm8
-; SSE-NEXT:    punpckldq {{.*#+}} xmm8 = xmm8[0],xmm10[0],xmm8[1],xmm10[1]
-; SSE-NEXT:    pblendw {{.*#+}} xmm8 = xmm8[0,1,2,3],xmm9[4,5,6,7]
+; SSE-NEXT:    packssdw %xmm9, %xmm8
+; SSE-NEXT:    packsswb %xmm10, %xmm8
 ; SSE-NEXT:    pand %xmm4, %xmm8
 ; SSE-NEXT:    pmovmskb %xmm8, %eax
 ; SSE-NEXT:    # kill: def $ax killed $ax killed $eax
diff --git a/llvm/test/CodeGen/X86/masked_store_trunc_ssat.ll b/llvm/test/CodeGen/X86/masked_store_trunc_ssat.ll
index b82d781..f2f9cc3 100644
--- a/llvm/test/CodeGen/X86/masked_store_trunc_ssat.ll
+++ b/llvm/test/CodeGen/X86/masked_store_trunc_ssat.ll
@@ -5263,7 +5263,8 @@
 ; SSE4-NEXT:    pxor %xmm2, %xmm2
 ; SSE4-NEXT:    pminsd {{.*}}(%rip), %xmm0
 ; SSE4-NEXT:    pmaxsd {{.*}}(%rip), %xmm0
-; SSE4-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE4-NEXT:    packssdw %xmm0, %xmm0
+; SSE4-NEXT:    packsswb %xmm0, %xmm0
 ; SSE4-NEXT:    pcmpeqd %xmm1, %xmm2
 ; SSE4-NEXT:    movmskps %xmm2, %eax
 ; SSE4-NEXT:    xorl $15, %eax
@@ -5301,7 +5302,8 @@
 ; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; AVX1-NEXT:    vpminsd {{.*}}(%rip), %xmm0, %xmm0
 ; AVX1-NEXT:    vpmaxsd {{.*}}(%rip), %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpackssdw %xmm0, %xmm0, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
 ; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm1, %xmm1
 ; AVX1-NEXT:    vmovmskps %xmm1, %eax
 ; AVX1-NEXT:    xorl $15, %eax
@@ -5341,7 +5343,8 @@
 ; AVX2-NEXT:    vpminsd %xmm3, %xmm0, %xmm0
 ; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm3 = [4294967168,4294967168,4294967168,4294967168]
 ; AVX2-NEXT:    vpmaxsd %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    vpackssdw %xmm0, %xmm0, %xmm0
+; AVX2-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
 ; AVX2-NEXT:    vpcmpeqd %xmm2, %xmm1, %xmm1
 ; AVX2-NEXT:    vmovmskps %xmm1, %eax
 ; AVX2-NEXT:    xorl $15, %eax
diff --git a/llvm/test/CodeGen/X86/masked_store_trunc_usat.ll b/llvm/test/CodeGen/X86/masked_store_trunc_usat.ll
index 84958fe..be5823d 100644
--- a/llvm/test/CodeGen/X86/masked_store_trunc_usat.ll
+++ b/llvm/test/CodeGen/X86/masked_store_trunc_usat.ll
@@ -4438,12 +4438,10 @@
 ; SSE4:       # %bb.0:
 ; SSE4-NEXT:    pxor %xmm4, %xmm4
 ; SSE4-NEXT:    movdqa {{.*#+}} xmm5 = [255,255,255,255]
-; SSE4-NEXT:    pminud %xmm5, %xmm0
 ; SSE4-NEXT:    pminud %xmm5, %xmm1
-; SSE4-NEXT:    movdqa {{.*#+}} xmm5 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
-; SSE4-NEXT:    pshufb %xmm5, %xmm1
-; SSE4-NEXT:    pshufb %xmm5, %xmm0
-; SSE4-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE4-NEXT:    pminud %xmm5, %xmm0
+; SSE4-NEXT:    packusdw %xmm1, %xmm0
+; SSE4-NEXT:    packuswb %xmm0, %xmm0
 ; SSE4-NEXT:    pcmpeqd %xmm4, %xmm3
 ; SSE4-NEXT:    pcmpeqd %xmm1, %xmm1
 ; SSE4-NEXT:    pxor %xmm1, %xmm3
@@ -4511,14 +4509,12 @@
 ;
 ; AVX1-LABEL: truncstore_v8i32_v8i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255]
-; AVX1-NEXT:    vpminud %xmm2, %xmm0, %xmm3
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; AVX1-NEXT:    vpminud %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
-; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb %xmm2, %xmm3, %xmm2
-; AVX1-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm2[0],xmm0[0],xmm2[1],xmm0[1]
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [255,255,255,255]
+; AVX1-NEXT:    vpminud %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vpminud %xmm3, %xmm0, %xmm0
+; AVX1-NEXT:    vpackusdw %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
 ; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
 ; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm2, %xmm2
@@ -4591,10 +4587,8 @@
 ; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm3 = [255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpminud %ymm3, %ymm0, %ymm0
 ; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm3
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm4 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
-; AVX2-NEXT:    vpshufb %xmm4, %xmm3, %xmm3
-; AVX2-NEXT:    vpshufb %xmm4, %xmm0, %xmm0
-; AVX2-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
+; AVX2-NEXT:    vpackusdw %xmm3, %xmm0, %xmm0
+; AVX2-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX2-NEXT:    vpcmpeqd %ymm2, %ymm1, %ymm1
 ; AVX2-NEXT:    vmovmskps %ymm1, %eax
 ; AVX2-NEXT:    notl %eax
@@ -5030,7 +5024,8 @@
 ; SSE4:       # %bb.0:
 ; SSE4-NEXT:    pxor %xmm2, %xmm2
 ; SSE4-NEXT:    pminud {{.*}}(%rip), %xmm0
-; SSE4-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE4-NEXT:    packusdw %xmm0, %xmm0
+; SSE4-NEXT:    packuswb %xmm0, %xmm0
 ; SSE4-NEXT:    pcmpeqd %xmm1, %xmm2
 ; SSE4-NEXT:    movmskps %xmm2, %eax
 ; SSE4-NEXT:    xorl $15, %eax
@@ -5067,7 +5062,8 @@
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; AVX1-NEXT:    vpminud {{.*}}(%rip), %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm1, %xmm1
 ; AVX1-NEXT:    vmovmskps %xmm1, %eax
 ; AVX1-NEXT:    xorl $15, %eax
@@ -5105,7 +5101,8 @@
 ; AVX2-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm3 = [255,255,255,255]
 ; AVX2-NEXT:    vpminud %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX2-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX2-NEXT:    vpcmpeqd %xmm2, %xmm1, %xmm1
 ; AVX2-NEXT:    vmovmskps %xmm1, %eax
 ; AVX2-NEXT:    xorl $15, %eax
diff --git a/llvm/test/CodeGen/X86/psubus.ll b/llvm/test/CodeGen/X86/psubus.ll
index 2903cbd..1d45732 100644
--- a/llvm/test/CodeGen/X86/psubus.ll
+++ b/llvm/test/CodeGen/X86/psubus.ll
@@ -622,34 +622,30 @@
 ; SSE2-NEXT:    psubd %xmm5, %xmm4
 ; SSE2-NEXT:    por %xmm0, %xmm5
 ; SSE2-NEXT:    pcmpgtd %xmm9, %xmm5
-; SSE2-NEXT:    movdqa {{.*#+}} xmm9 = [255,0,255,0,255,0,255,0]
-; SSE2-NEXT:    pand %xmm9, %xmm5
 ; SSE2-NEXT:    movdqa %xmm3, %xmm7
 ; SSE2-NEXT:    pxor %xmm0, %xmm7
 ; SSE2-NEXT:    psubd %xmm10, %xmm3
 ; SSE2-NEXT:    por %xmm0, %xmm10
 ; SSE2-NEXT:    pcmpgtd %xmm7, %xmm10
-; SSE2-NEXT:    pand %xmm9, %xmm10
-; SSE2-NEXT:    packuswb %xmm5, %xmm10
+; SSE2-NEXT:    packssdw %xmm5, %xmm10
 ; SSE2-NEXT:    movdqa %xmm2, %xmm5
 ; SSE2-NEXT:    pxor %xmm0, %xmm5
 ; SSE2-NEXT:    psubd %xmm6, %xmm2
 ; SSE2-NEXT:    por %xmm0, %xmm6
 ; SSE2-NEXT:    pcmpgtd %xmm5, %xmm6
-; SSE2-NEXT:    pand %xmm9, %xmm6
 ; SSE2-NEXT:    movdqa %xmm1, %xmm5
 ; SSE2-NEXT:    pxor %xmm0, %xmm5
 ; SSE2-NEXT:    por %xmm8, %xmm0
 ; SSE2-NEXT:    pcmpgtd %xmm5, %xmm0
-; SSE2-NEXT:    pand %xmm9, %xmm0
-; SSE2-NEXT:    packuswb %xmm6, %xmm0
-; SSE2-NEXT:    packuswb %xmm10, %xmm0
+; SSE2-NEXT:    packssdw %xmm6, %xmm0
+; SSE2-NEXT:    packsswb %xmm10, %xmm0
 ; SSE2-NEXT:    psubd %xmm8, %xmm1
-; SSE2-NEXT:    pand %xmm9, %xmm4
-; SSE2-NEXT:    pand %xmm9, %xmm3
+; SSE2-NEXT:    movdqa {{.*#+}} xmm5 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
+; SSE2-NEXT:    pand %xmm5, %xmm4
+; SSE2-NEXT:    pand %xmm5, %xmm3
 ; SSE2-NEXT:    packuswb %xmm4, %xmm3
-; SSE2-NEXT:    pand %xmm9, %xmm2
-; SSE2-NEXT:    pand %xmm9, %xmm1
+; SSE2-NEXT:    pand %xmm5, %xmm2
+; SSE2-NEXT:    pand %xmm5, %xmm1
 ; SSE2-NEXT:    packuswb %xmm2, %xmm1
 ; SSE2-NEXT:    packuswb %xmm3, %xmm1
 ; SSE2-NEXT:    pandn %xmm1, %xmm0
@@ -657,48 +653,41 @@
 ;
 ; SSSE3-LABEL: test14:
 ; SSSE3:       # %bb.0: # %vector.ph
-; SSSE3-NEXT:    pxor %xmm7, %xmm7
-; SSSE3-NEXT:    movdqa %xmm0, %xmm11
-; SSSE3-NEXT:    punpckhbw {{.*#+}} xmm11 = xmm11[8],xmm7[8],xmm11[9],xmm7[9],xmm11[10],xmm7[10],xmm11[11],xmm7[11],xmm11[12],xmm7[12],xmm11[13],xmm7[13],xmm11[14],xmm7[14],xmm11[15],xmm7[15]
-; SSSE3-NEXT:    movdqa %xmm11, %xmm8
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm7[0],xmm8[1],xmm7[1],xmm8[2],xmm7[2],xmm8[3],xmm7[3]
-; SSSE3-NEXT:    punpckhwd {{.*#+}} xmm11 = xmm11[4],xmm7[4],xmm11[5],xmm7[5],xmm11[6],xmm7[6],xmm11[7],xmm7[7]
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0],xmm7[0],xmm0[1],xmm7[1],xmm0[2],xmm7[2],xmm0[3],xmm7[3],xmm0[4],xmm7[4],xmm0[5],xmm7[5],xmm0[6],xmm7[6],xmm0[7],xmm7[7]
-; SSSE3-NEXT:    movdqa %xmm0, %xmm10
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm10 = xmm10[0],xmm7[0],xmm10[1],xmm7[1],xmm10[2],xmm7[2],xmm10[3],xmm7[3]
-; SSSE3-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm7[4],xmm0[5],xmm7[5],xmm0[6],xmm7[6],xmm0[7],xmm7[7]
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm7 = [2147483648,2147483648,2147483648,2147483648]
-; SSSE3-NEXT:    movdqa %xmm2, %xmm9
-; SSSE3-NEXT:    pxor %xmm7, %xmm9
-; SSSE3-NEXT:    psubd %xmm0, %xmm2
 ; SSSE3-NEXT:    movdqa %xmm0, %xmm5
-; SSSE3-NEXT:    por %xmm7, %xmm5
+; SSSE3-NEXT:    pxor %xmm0, %xmm0
+; SSSE3-NEXT:    movdqa %xmm5, %xmm6
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm6 = xmm6[0],xmm0[0],xmm6[1],xmm0[1],xmm6[2],xmm0[2],xmm6[3],xmm0[3],xmm6[4],xmm0[4],xmm6[5],xmm0[5],xmm6[6],xmm0[6],xmm6[7],xmm0[7]
+; SSSE3-NEXT:    movdqa %xmm6, %xmm8
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm0[0],xmm8[1],xmm0[1],xmm8[2],xmm0[2],xmm8[3],xmm0[3]
+; SSSE3-NEXT:    punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm0[4],xmm6[5],xmm0[5],xmm6[6],xmm0[6],xmm6[7],xmm0[7]
+; SSSE3-NEXT:    punpckhbw {{.*#+}} xmm5 = xmm5[8],xmm0[8],xmm5[9],xmm0[9],xmm5[10],xmm0[10],xmm5[11],xmm0[11],xmm5[12],xmm0[12],xmm5[13],xmm0[13],xmm5[14],xmm0[14],xmm5[15],xmm0[15]
+; SSSE3-NEXT:    movdqa %xmm5, %xmm10
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm10 = xmm10[0],xmm0[0],xmm10[1],xmm0[1],xmm10[2],xmm0[2],xmm10[3],xmm0[3]
+; SSSE3-NEXT:    punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm0[4],xmm5[5],xmm0[5],xmm5[6],xmm0[6],xmm5[7],xmm0[7]
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,2147483648,2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm4, %xmm9
+; SSSE3-NEXT:    pxor %xmm0, %xmm9
+; SSSE3-NEXT:    psubd %xmm5, %xmm4
+; SSSE3-NEXT:    por %xmm0, %xmm5
 ; SSSE3-NEXT:    pcmpgtd %xmm9, %xmm5
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm9 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
-; SSSE3-NEXT:    pshufb %xmm9, %xmm5
-; SSSE3-NEXT:    movdqa %xmm1, %xmm6
-; SSSE3-NEXT:    pxor %xmm7, %xmm6
-; SSSE3-NEXT:    psubd %xmm10, %xmm1
-; SSSE3-NEXT:    movdqa %xmm10, %xmm0
-; SSSE3-NEXT:    por %xmm7, %xmm0
-; SSSE3-NEXT:    pcmpgtd %xmm6, %xmm0
-; SSSE3-NEXT:    pshufb %xmm9, %xmm0
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm5[0],xmm0[1],xmm5[1]
-; SSSE3-NEXT:    movdqa %xmm4, %xmm5
-; SSSE3-NEXT:    pxor %xmm7, %xmm5
-; SSSE3-NEXT:    psubd %xmm11, %xmm4
-; SSSE3-NEXT:    por %xmm7, %xmm11
-; SSSE3-NEXT:    pcmpgtd %xmm5, %xmm11
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm5 = <u,u,u,u,0,4,8,12,u,u,u,u,u,u,u,u>
-; SSSE3-NEXT:    pshufb %xmm5, %xmm11
-; SSSE3-NEXT:    movdqa %xmm3, %xmm6
-; SSSE3-NEXT:    pxor %xmm7, %xmm6
-; SSSE3-NEXT:    por %xmm8, %xmm7
-; SSSE3-NEXT:    pcmpgtd %xmm6, %xmm7
-; SSSE3-NEXT:    pshufb %xmm5, %xmm7
-; SSSE3-NEXT:    punpckldq {{.*#+}} xmm7 = xmm7[0],xmm11[0],xmm7[1],xmm11[1]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm7[2,3]
-; SSSE3-NEXT:    psubd %xmm8, %xmm3
+; SSSE3-NEXT:    movdqa %xmm3, %xmm7
+; SSSE3-NEXT:    pxor %xmm0, %xmm7
+; SSSE3-NEXT:    psubd %xmm10, %xmm3
+; SSSE3-NEXT:    por %xmm0, %xmm10
+; SSSE3-NEXT:    pcmpgtd %xmm7, %xmm10
+; SSSE3-NEXT:    packssdw %xmm5, %xmm10
+; SSSE3-NEXT:    movdqa %xmm2, %xmm5
+; SSSE3-NEXT:    pxor %xmm0, %xmm5
+; SSSE3-NEXT:    psubd %xmm6, %xmm2
+; SSSE3-NEXT:    por %xmm0, %xmm6
+; SSSE3-NEXT:    pcmpgtd %xmm5, %xmm6
+; SSSE3-NEXT:    movdqa %xmm1, %xmm5
+; SSSE3-NEXT:    pxor %xmm0, %xmm5
+; SSSE3-NEXT:    por %xmm8, %xmm0
+; SSSE3-NEXT:    pcmpgtd %xmm5, %xmm0
+; SSSE3-NEXT:    packssdw %xmm6, %xmm0
+; SSSE3-NEXT:    packsswb %xmm10, %xmm0
+; SSSE3-NEXT:    psubd %xmm8, %xmm1
 ; SSSE3-NEXT:    movdqa {{.*#+}} xmm5 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
 ; SSSE3-NEXT:    pand %xmm5, %xmm4
 ; SSSE3-NEXT:    pand %xmm5, %xmm3
@@ -724,27 +713,21 @@
 ; SSE41-NEXT:    pcmpeqd %xmm4, %xmm0
 ; SSE41-NEXT:    pcmpeqd %xmm6, %xmm6
 ; SSE41-NEXT:    pxor %xmm6, %xmm0
-; SSE41-NEXT:    movdqa {{.*#+}} xmm7 = <u,u,u,u,0,4,8,12,u,u,u,u,u,u,u,u>
-; SSE41-NEXT:    pshufb %xmm7, %xmm0
-; SSE41-NEXT:    movdqa %xmm3, %xmm5
-; SSE41-NEXT:    pmaxud %xmm9, %xmm5
-; SSE41-NEXT:    pcmpeqd %xmm3, %xmm5
-; SSE41-NEXT:    pxor %xmm6, %xmm5
-; SSE41-NEXT:    pshufb %xmm7, %xmm5
-; SSE41-NEXT:    punpckldq {{.*#+}} xmm5 = xmm5[0],xmm0[0],xmm5[1],xmm0[1]
+; SSE41-NEXT:    movdqa %xmm3, %xmm7
+; SSE41-NEXT:    pmaxud %xmm9, %xmm7
+; SSE41-NEXT:    pcmpeqd %xmm3, %xmm7
+; SSE41-NEXT:    pxor %xmm6, %xmm7
+; SSE41-NEXT:    packssdw %xmm0, %xmm7
 ; SSE41-NEXT:    movdqa %xmm1, %xmm0
 ; SSE41-NEXT:    pmaxud %xmm8, %xmm0
 ; SSE41-NEXT:    pcmpeqd %xmm1, %xmm0
 ; SSE41-NEXT:    pxor %xmm6, %xmm0
-; SSE41-NEXT:    movdqa {{.*#+}} xmm12 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
-; SSE41-NEXT:    pshufb %xmm12, %xmm0
-; SSE41-NEXT:    movdqa %xmm2, %xmm7
-; SSE41-NEXT:    pmaxud %xmm11, %xmm7
-; SSE41-NEXT:    pcmpeqd %xmm2, %xmm7
-; SSE41-NEXT:    pxor %xmm6, %xmm7
-; SSE41-NEXT:    pshufb %xmm12, %xmm7
-; SSE41-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm7[0],xmm0[1],xmm7[1]
-; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm5[4,5,6,7]
+; SSE41-NEXT:    movdqa %xmm2, %xmm5
+; SSE41-NEXT:    pmaxud %xmm11, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm2, %xmm5
+; SSE41-NEXT:    pxor %xmm6, %xmm5
+; SSE41-NEXT:    packssdw %xmm5, %xmm0
+; SSE41-NEXT:    packsswb %xmm7, %xmm0
 ; SSE41-NEXT:    psubd %xmm11, %xmm2
 ; SSE41-NEXT:    psubd %xmm8, %xmm1
 ; SSE41-NEXT:    psubd %xmm9, %xmm3
diff --git a/llvm/test/CodeGen/X86/vec-strict-fptoint-128.ll b/llvm/test/CodeGen/X86/vec-strict-fptoint-128.ll
index ed2db51..b3a91e6 100644
--- a/llvm/test/CodeGen/X86/vec-strict-fptoint-128.ll
+++ b/llvm/test/CodeGen/X86/vec-strict-fptoint-128.ll
@@ -2638,29 +2638,29 @@
 ; SSE-32-LABEL: strict_vector_fptosi_v4f32_to_v4i8:
 ; SSE-32:       # %bb.0:
 ; SSE-32-NEXT:    cvttps2dq %xmm0, %xmm0
-; SSE-32-NEXT:    pand {{\.LCPI.*}}, %xmm0
-; SSE-32-NEXT:    packuswb %xmm0, %xmm0
-; SSE-32-NEXT:    packuswb %xmm0, %xmm0
+; SSE-32-NEXT:    packssdw %xmm0, %xmm0
+; SSE-32-NEXT:    packsswb %xmm0, %xmm0
 ; SSE-32-NEXT:    retl
 ;
 ; SSE-64-LABEL: strict_vector_fptosi_v4f32_to_v4i8:
 ; SSE-64:       # %bb.0:
 ; SSE-64-NEXT:    cvttps2dq %xmm0, %xmm0
-; SSE-64-NEXT:    pand {{.*}}(%rip), %xmm0
-; SSE-64-NEXT:    packuswb %xmm0, %xmm0
-; SSE-64-NEXT:    packuswb %xmm0, %xmm0
+; SSE-64-NEXT:    packssdw %xmm0, %xmm0
+; SSE-64-NEXT:    packsswb %xmm0, %xmm0
 ; SSE-64-NEXT:    retq
 ;
 ; AVX-LABEL: strict_vector_fptosi_v4f32_to_v4i8:
 ; AVX:       # %bb.0:
 ; AVX-NEXT:    vcvttps2dq %xmm0, %xmm0
-; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX-NEXT:    vpackssdw %xmm0, %xmm0, %xmm0
+; AVX-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
 ; AVX-NEXT:    ret{{[l|q]}}
 ;
 ; AVX512F-LABEL: strict_vector_fptosi_v4f32_to_v4i8:
 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    vcvttps2dq %xmm0, %xmm0
-; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT:    vpackssdw %xmm0, %xmm0, %xmm0
+; AVX512F-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
 ; AVX512F-NEXT:    ret{{[l|q]}}
 ;
 ; AVX512VL-LABEL: strict_vector_fptosi_v4f32_to_v4i8:
@@ -2672,7 +2672,8 @@
 ; AVX512DQ-LABEL: strict_vector_fptosi_v4f32_to_v4i8:
 ; AVX512DQ:       # %bb.0:
 ; AVX512DQ-NEXT:    vcvttps2dq %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512DQ-NEXT:    vpackssdw %xmm0, %xmm0, %xmm0
+; AVX512DQ-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
 ; AVX512DQ-NEXT:    ret{{[l|q]}}
 ;
 ; AVX512VLDQ-LABEL: strict_vector_fptosi_v4f32_to_v4i8:
@@ -2689,7 +2690,6 @@
 ; SSE-32-LABEL: strict_vector_fptoui_v4f32_to_v4i8:
 ; SSE-32:       # %bb.0:
 ; SSE-32-NEXT:    cvttps2dq %xmm0, %xmm0
-; SSE-32-NEXT:    pand {{\.LCPI.*}}, %xmm0
 ; SSE-32-NEXT:    packuswb %xmm0, %xmm0
 ; SSE-32-NEXT:    packuswb %xmm0, %xmm0
 ; SSE-32-NEXT:    retl
@@ -2697,7 +2697,6 @@
 ; SSE-64-LABEL: strict_vector_fptoui_v4f32_to_v4i8:
 ; SSE-64:       # %bb.0:
 ; SSE-64-NEXT:    cvttps2dq %xmm0, %xmm0
-; SSE-64-NEXT:    pand {{.*}}(%rip), %xmm0
 ; SSE-64-NEXT:    packuswb %xmm0, %xmm0
 ; SSE-64-NEXT:    packuswb %xmm0, %xmm0
 ; SSE-64-NEXT:    retq
@@ -2705,13 +2704,15 @@
 ; AVX-LABEL: strict_vector_fptoui_v4f32_to_v4i8:
 ; AVX:       # %bb.0:
 ; AVX-NEXT:    vcvttps2dq %xmm0, %xmm0
-; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX-NEXT:    ret{{[l|q]}}
 ;
 ; AVX512F-LABEL: strict_vector_fptoui_v4f32_to_v4i8:
 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    vcvttps2dq %xmm0, %xmm0
-; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX512F-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX512F-NEXT:    ret{{[l|q]}}
 ;
 ; AVX512VL-LABEL: strict_vector_fptoui_v4f32_to_v4i8:
@@ -2723,7 +2724,8 @@
 ; AVX512DQ-LABEL: strict_vector_fptoui_v4f32_to_v4i8:
 ; AVX512DQ:       # %bb.0:
 ; AVX512DQ-NEXT:    vcvttps2dq %xmm0, %xmm0
-; AVX512DQ-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512DQ-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX512DQ-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX512DQ-NEXT:    ret{{[l|q]}}
 ;
 ; AVX512VLDQ-LABEL: strict_vector_fptoui_v4f32_to_v4i8:
diff --git a/llvm/test/CodeGen/X86/vec-strict-fptoint-256.ll b/llvm/test/CodeGen/X86/vec-strict-fptoint-256.ll
index 0f9fc8d..00aee49f 100644
--- a/llvm/test/CodeGen/X86/vec-strict-fptoint-256.ll
+++ b/llvm/test/CodeGen/X86/vec-strict-fptoint-256.ll
@@ -1250,14 +1250,16 @@
 ; AVX-LABEL: strict_vector_fptosi_v4f64_to_v4i8:
 ; AVX:       # %bb.0:
 ; AVX-NEXT:    vcvttpd2dq %ymm0, %xmm0
-; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX-NEXT:    vpackssdw %xmm0, %xmm0, %xmm0
+; AVX-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    ret{{[l|q]}}
 ;
 ; AVX512F-LABEL: strict_vector_fptosi_v4f64_to_v4i8:
 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    vcvttpd2dq %ymm0, %xmm0
-; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT:    vpackssdw %xmm0, %xmm0, %xmm0
+; AVX512F-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
 ; AVX512F-NEXT:    vzeroupper
 ; AVX512F-NEXT:    ret{{[l|q]}}
 ;
@@ -1271,7 +1273,8 @@
 ; AVX512DQ-LABEL: strict_vector_fptosi_v4f64_to_v4i8:
 ; AVX512DQ:       # %bb.0:
 ; AVX512DQ-NEXT:    vcvttpd2dq %ymm0, %xmm0
-; AVX512DQ-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512DQ-NEXT:    vpackssdw %xmm0, %xmm0, %xmm0
+; AVX512DQ-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
 ; AVX512DQ-NEXT:    vzeroupper
 ; AVX512DQ-NEXT:    ret{{[l|q]}}
 ;
@@ -1290,14 +1293,16 @@
 ; AVX-LABEL: strict_vector_fptoui_v4f64_to_v4i8:
 ; AVX:       # %bb.0:
 ; AVX-NEXT:    vcvttpd2dq %ymm0, %xmm0
-; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX-NEXT:    vzeroupper
 ; AVX-NEXT:    ret{{[l|q]}}
 ;
 ; AVX512F-LABEL: strict_vector_fptoui_v4f64_to_v4i8:
 ; AVX512F:       # %bb.0:
 ; AVX512F-NEXT:    vcvttpd2dq %ymm0, %xmm0
-; AVX512F-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512F-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX512F-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX512F-NEXT:    vzeroupper
 ; AVX512F-NEXT:    ret{{[l|q]}}
 ;
@@ -1311,7 +1316,8 @@
 ; AVX512DQ-LABEL: strict_vector_fptoui_v4f64_to_v4i8:
 ; AVX512DQ:       # %bb.0:
 ; AVX512DQ-NEXT:    vcvttpd2dq %ymm0, %xmm0
-; AVX512DQ-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512DQ-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX512DQ-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX512DQ-NEXT:    vzeroupper
 ; AVX512DQ-NEXT:    ret{{[l|q]}}
 ;
diff --git a/llvm/test/CodeGen/X86/vec_cast2.ll b/llvm/test/CodeGen/X86/vec_cast2.ll
index e43216d..ed703f1 100644
--- a/llvm/test/CodeGen/X86/vec_cast2.ll
+++ b/llvm/test/CodeGen/X86/vec_cast2.ll
@@ -122,7 +122,8 @@
 ; CHECK-LABEL: cvt_v4f32_v4i8:
 ; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    vcvttps2dq %xmm0, %xmm0
-; CHECK-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; CHECK-NEXT:    vpackssdw %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
 ; CHECK-NEXT:    retl
   %res = fptosi <4 x float> %src to <4 x i8>
   ret <4 x i8> %res
@@ -167,7 +168,8 @@
 ; CHECK-LABEL: cvt_v4f32_v4u8:
 ; CHECK:       ## %bb.0:
 ; CHECK-NEXT:    vcvttps2dq %xmm0, %xmm0
-; CHECK-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; CHECK-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; CHECK-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; CHECK-NEXT:    retl
   %res = fptoui <4 x float> %src to <4 x i8>
   ret <4 x i8> %res
diff --git a/llvm/test/CodeGen/X86/vector-shuffle-512-v64.ll b/llvm/test/CodeGen/X86/vector-shuffle-512-v64.ll
index 30574e4..e048f6e 100644
--- a/llvm/test/CodeGen/X86/vector-shuffle-512-v64.ll
+++ b/llvm/test/CodeGen/X86/vector-shuffle-512-v64.ll
@@ -597,14 +597,10 @@
 ; AVX512F-NEXT:    vpsrad $25, %zmm1, %zmm1
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm2
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm1, %ymm3
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm4 = <u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12>
-; AVX512F-NEXT:    vpshufb %ymm4, %ymm3, %ymm3
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm5 = <0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u>
-; AVX512F-NEXT:    vpshufb %ymm5, %ymm2, %ymm2
-; AVX512F-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0],ymm3[1],ymm2[2],ymm3[3],ymm2[4],ymm3[5],ymm2[6],ymm3[7]
-; AVX512F-NEXT:    vpshufb %ymm4, %ymm1, %ymm1
-; AVX512F-NEXT:    vpshufb %ymm5, %ymm0, %ymm0
-; AVX512F-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
+; AVX512F-NEXT:    vpackssdw %ymm3, %ymm2, %ymm2
+; AVX512F-NEXT:    vpacksswb %ymm2, %ymm2, %ymm2
+; AVX512F-NEXT:    vpackssdw %ymm1, %ymm0, %ymm0
+; AVX512F-NEXT:    vpacksswb %ymm0, %ymm0, %ymm0
 ; AVX512F-NEXT:    vinserti64x4 $1, %ymm2, %zmm0, %zmm0
 ; AVX512F-NEXT:    retq
 ;
@@ -612,17 +608,8 @@
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vpsrad $25, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vpsrad $25, %zmm1, %zmm1
-; AVX512BW-NEXT:    vextracti64x4 $1, %zmm1, %ymm2
-; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm3 = <u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12>
-; AVX512BW-NEXT:    vpshufb %ymm3, %ymm2, %ymm2
-; AVX512BW-NEXT:    vextracti64x4 $1, %zmm0, %ymm4
-; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm5 = <0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u>
-; AVX512BW-NEXT:    vpshufb %ymm5, %ymm4, %ymm4
-; AVX512BW-NEXT:    vpblendd {{.*#+}} ymm2 = ymm4[0],ymm2[1],ymm4[2],ymm2[3],ymm4[4],ymm2[5],ymm4[6],ymm2[7]
-; AVX512BW-NEXT:    vpshufb %ymm3, %ymm1, %ymm1
-; AVX512BW-NEXT:    vpshufb %ymm5, %ymm0, %ymm0
-; AVX512BW-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
-; AVX512BW-NEXT:    vinserti64x4 $1, %ymm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpackssdw %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpacksswb %zmm0, %zmm0, %zmm0
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512DQ-LABEL: shuffle_v64i8_ashr_00_04_08_12_64_68_72_76_00_04_08_12_64_68_72_76_16_20_24_28_80_84_88_92_16_20_24_28_80_84_88_92_32_36_40_44_96_100_104_108_32_36_40_44_96_100_104_108_48_52_56_60_112_116_120_124_48_52_56_60_112_116_120_124:
@@ -631,23 +618,19 @@
 ; AVX512DQ-NEXT:    vpsrad $25, %zmm1, %zmm1
 ; AVX512DQ-NEXT:    vextracti64x4 $1, %zmm0, %ymm2
 ; AVX512DQ-NEXT:    vextracti64x4 $1, %zmm1, %ymm3
-; AVX512DQ-NEXT:    vmovdqa {{.*#+}} ymm4 = <u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12>
-; AVX512DQ-NEXT:    vpshufb %ymm4, %ymm3, %ymm3
-; AVX512DQ-NEXT:    vmovdqa {{.*#+}} ymm5 = <0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u>
-; AVX512DQ-NEXT:    vpshufb %ymm5, %ymm2, %ymm2
-; AVX512DQ-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0],ymm3[1],ymm2[2],ymm3[3],ymm2[4],ymm3[5],ymm2[6],ymm3[7]
-; AVX512DQ-NEXT:    vpshufb %ymm4, %ymm1, %ymm1
-; AVX512DQ-NEXT:    vpshufb %ymm5, %ymm0, %ymm0
-; AVX512DQ-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
+; AVX512DQ-NEXT:    vpackssdw %ymm3, %ymm2, %ymm2
+; AVX512DQ-NEXT:    vpacksswb %ymm2, %ymm2, %ymm2
+; AVX512DQ-NEXT:    vpackssdw %ymm1, %ymm0, %ymm0
+; AVX512DQ-NEXT:    vpacksswb %ymm0, %ymm0, %ymm0
 ; AVX512DQ-NEXT:    vinserti64x4 $1, %ymm2, %zmm0, %zmm0
 ; AVX512DQ-NEXT:    retq
 ;
 ; AVX512VBMI-LABEL: shuffle_v64i8_ashr_00_04_08_12_64_68_72_76_00_04_08_12_64_68_72_76_16_20_24_28_80_84_88_92_16_20_24_28_80_84_88_92_32_36_40_44_96_100_104_108_32_36_40_44_96_100_104_108_48_52_56_60_112_116_120_124_48_52_56_60_112_116_120_124:
 ; AVX512VBMI:       # %bb.0:
-; AVX512VBMI-NEXT:    vpsrad $25, %zmm0, %zmm2
+; AVX512VBMI-NEXT:    vpsrad $25, %zmm0, %zmm0
 ; AVX512VBMI-NEXT:    vpsrad $25, %zmm1, %zmm1
-; AVX512VBMI-NEXT:    vmovdqa64 {{.*#+}} zmm0 = [0,4,8,12,64,68,72,76,0,4,8,12,64,68,72,76,16,20,24,28,80,84,88,92,16,20,24,28,80,84,88,92,32,36,40,44,96,100,104,108,32,36,40,44,96,100,104,108,48,52,56,60,112,116,120,124,48,52,56,60,112,116,120,124]
-; AVX512VBMI-NEXT:    vpermi2b %zmm1, %zmm2, %zmm0
+; AVX512VBMI-NEXT:    vpackssdw %zmm1, %zmm0, %zmm0
+; AVX512VBMI-NEXT:    vpacksswb %zmm0, %zmm0, %zmm0
 ; AVX512VBMI-NEXT:    retq
   %1 = ashr <16 x i32> %a0, <i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25>
   %2 = ashr <16 x i32> %a1, <i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25>
@@ -708,14 +691,10 @@
 ; AVX512F-NEXT:    vpsrld $25, %zmm1, %zmm1
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm0, %ymm2
 ; AVX512F-NEXT:    vextracti64x4 $1, %zmm1, %ymm3
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm4 = <u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12>
-; AVX512F-NEXT:    vpshufb %ymm4, %ymm3, %ymm3
-; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm5 = <0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u>
-; AVX512F-NEXT:    vpshufb %ymm5, %ymm2, %ymm2
-; AVX512F-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0],ymm3[1],ymm2[2],ymm3[3],ymm2[4],ymm3[5],ymm2[6],ymm3[7]
-; AVX512F-NEXT:    vpshufb %ymm4, %ymm1, %ymm1
-; AVX512F-NEXT:    vpshufb %ymm5, %ymm0, %ymm0
-; AVX512F-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
+; AVX512F-NEXT:    vpackusdw %ymm3, %ymm2, %ymm2
+; AVX512F-NEXT:    vpackuswb %ymm2, %ymm2, %ymm2
+; AVX512F-NEXT:    vpackusdw %ymm1, %ymm0, %ymm0
+; AVX512F-NEXT:    vpackuswb %ymm0, %ymm0, %ymm0
 ; AVX512F-NEXT:    vinserti64x4 $1, %ymm2, %zmm0, %zmm0
 ; AVX512F-NEXT:    retq
 ;
@@ -723,17 +702,8 @@
 ; AVX512BW:       # %bb.0:
 ; AVX512BW-NEXT:    vpsrld $25, %zmm0, %zmm0
 ; AVX512BW-NEXT:    vpsrld $25, %zmm1, %zmm1
-; AVX512BW-NEXT:    vextracti64x4 $1, %zmm1, %ymm2
-; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm3 = <u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12>
-; AVX512BW-NEXT:    vpshufb %ymm3, %ymm2, %ymm2
-; AVX512BW-NEXT:    vextracti64x4 $1, %zmm0, %ymm4
-; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm5 = <0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u>
-; AVX512BW-NEXT:    vpshufb %ymm5, %ymm4, %ymm4
-; AVX512BW-NEXT:    vpblendd {{.*#+}} ymm2 = ymm4[0],ymm2[1],ymm4[2],ymm2[3],ymm4[4],ymm2[5],ymm4[6],ymm2[7]
-; AVX512BW-NEXT:    vpshufb %ymm3, %ymm1, %ymm1
-; AVX512BW-NEXT:    vpshufb %ymm5, %ymm0, %ymm0
-; AVX512BW-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
-; AVX512BW-NEXT:    vinserti64x4 $1, %ymm2, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpackusdw %zmm1, %zmm0, %zmm0
+; AVX512BW-NEXT:    vpackuswb %zmm0, %zmm0, %zmm0
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512DQ-LABEL: shuffle_v64i8_lshr_00_04_08_12_64_68_72_76_00_04_08_12_64_68_72_76_16_20_24_28_80_84_88_092_16_20_24_28_80_84_88_92_32_36_40_44_96_100_104_108_32_36_40_44_96_100_104_108_48_52_56_60_112_116_120_124_48_52_56_60_112_116_120_124:
@@ -742,23 +712,19 @@
 ; AVX512DQ-NEXT:    vpsrld $25, %zmm1, %zmm1
 ; AVX512DQ-NEXT:    vextracti64x4 $1, %zmm0, %ymm2
 ; AVX512DQ-NEXT:    vextracti64x4 $1, %zmm1, %ymm3
-; AVX512DQ-NEXT:    vmovdqa {{.*#+}} ymm4 = <u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12>
-; AVX512DQ-NEXT:    vpshufb %ymm4, %ymm3, %ymm3
-; AVX512DQ-NEXT:    vmovdqa {{.*#+}} ymm5 = <0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u,0,4,8,12,u,u,u,u>
-; AVX512DQ-NEXT:    vpshufb %ymm5, %ymm2, %ymm2
-; AVX512DQ-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0],ymm3[1],ymm2[2],ymm3[3],ymm2[4],ymm3[5],ymm2[6],ymm3[7]
-; AVX512DQ-NEXT:    vpshufb %ymm4, %ymm1, %ymm1
-; AVX512DQ-NEXT:    vpshufb %ymm5, %ymm0, %ymm0
-; AVX512DQ-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
+; AVX512DQ-NEXT:    vpackusdw %ymm3, %ymm2, %ymm2
+; AVX512DQ-NEXT:    vpackuswb %ymm2, %ymm2, %ymm2
+; AVX512DQ-NEXT:    vpackusdw %ymm1, %ymm0, %ymm0
+; AVX512DQ-NEXT:    vpackuswb %ymm0, %ymm0, %ymm0
 ; AVX512DQ-NEXT:    vinserti64x4 $1, %ymm2, %zmm0, %zmm0
 ; AVX512DQ-NEXT:    retq
 ;
 ; AVX512VBMI-LABEL: shuffle_v64i8_lshr_00_04_08_12_64_68_72_76_00_04_08_12_64_68_72_76_16_20_24_28_80_84_88_092_16_20_24_28_80_84_88_92_32_36_40_44_96_100_104_108_32_36_40_44_96_100_104_108_48_52_56_60_112_116_120_124_48_52_56_60_112_116_120_124:
 ; AVX512VBMI:       # %bb.0:
-; AVX512VBMI-NEXT:    vpsrld $25, %zmm0, %zmm2
+; AVX512VBMI-NEXT:    vpsrld $25, %zmm0, %zmm0
 ; AVX512VBMI-NEXT:    vpsrld $25, %zmm1, %zmm1
-; AVX512VBMI-NEXT:    vmovdqa64 {{.*#+}} zmm0 = [0,4,8,12,64,68,72,76,0,4,8,12,64,68,72,76,16,20,24,28,80,84,88,92,16,20,24,28,80,84,88,92,32,36,40,44,96,100,104,108,32,36,40,44,96,100,104,108,48,52,56,60,112,116,120,124,48,52,56,60,112,116,120,124]
-; AVX512VBMI-NEXT:    vpermi2b %zmm1, %zmm2, %zmm0
+; AVX512VBMI-NEXT:    vpackusdw %zmm1, %zmm0, %zmm0
+; AVX512VBMI-NEXT:    vpackuswb %zmm0, %zmm0, %zmm0
 ; AVX512VBMI-NEXT:    retq
   %1 = lshr <16 x i32> %a0, <i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25>
   %2 = lshr <16 x i32> %a1, <i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25, i32 25>
diff --git a/llvm/test/CodeGen/X86/vector-trunc-packus.ll b/llvm/test/CodeGen/X86/vector-trunc-packus.ll
index c74b879..ed26bd8 100644
--- a/llvm/test/CodeGen/X86/vector-trunc-packus.ll
+++ b/llvm/test/CodeGen/X86/vector-trunc-packus.ll
@@ -5211,8 +5211,10 @@
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pminsd {{.*}}(%rip), %xmm0
 ; SSE41-NEXT:    pxor %xmm1, %xmm1
-; SSE41-NEXT:    pmaxsd %xmm1, %xmm0
-; SSE41-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE41-NEXT:    pmaxsd %xmm0, %xmm1
+; SSE41-NEXT:    packusdw %xmm0, %xmm1
+; SSE41-NEXT:    packuswb %xmm1, %xmm1
+; SSE41-NEXT:    movdqa %xmm1, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_packus_v4i32_v4i8:
@@ -5220,7 +5222,8 @@
 ; AVX1-NEXT:    vpminsd {{.*}}(%rip), %xmm0, %xmm0
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpmaxsd %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: trunc_packus_v4i32_v4i8:
@@ -5229,7 +5232,8 @@
 ; AVX2-NEXT:    vpminsd %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX2-NEXT:    vpmaxsd %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX2-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512F-LABEL: trunc_packus_v4i32_v4i8:
@@ -5316,7 +5320,8 @@
 ; SSE41-NEXT:    pminsd {{.*}}(%rip), %xmm0
 ; SSE41-NEXT:    pxor %xmm1, %xmm1
 ; SSE41-NEXT:    pmaxsd %xmm0, %xmm1
-; SSE41-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE41-NEXT:    packusdw %xmm0, %xmm1
+; SSE41-NEXT:    packuswb %xmm0, %xmm1
 ; SSE41-NEXT:    movd %xmm1, (%rdi)
 ; SSE41-NEXT:    retq
 ;
@@ -5325,7 +5330,8 @@
 ; AVX1-NEXT:    vpminsd {{.*}}(%rip), %xmm0, %xmm0
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpmaxsd %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovd %xmm0, (%rdi)
 ; AVX1-NEXT:    retq
 ;
@@ -5335,7 +5341,8 @@
 ; AVX2-NEXT:    vpminsd %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX2-NEXT:    vpmaxsd %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX2-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX2-NEXT:    vmovd %xmm0, (%rdi)
 ; AVX2-NEXT:    retq
 ;
diff --git a/llvm/test/CodeGen/X86/vector-trunc-ssat.ll b/llvm/test/CodeGen/X86/vector-trunc-ssat.ll
index 665e3fe..bed4553 100644
--- a/llvm/test/CodeGen/X86/vector-trunc-ssat.ll
+++ b/llvm/test/CodeGen/X86/vector-trunc-ssat.ll
@@ -5080,14 +5080,16 @@
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pminsd {{.*}}(%rip), %xmm0
 ; SSE41-NEXT:    pmaxsd {{.*}}(%rip), %xmm0
-; SSE41-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE41-NEXT:    packssdw %xmm0, %xmm0
+; SSE41-NEXT:    packsswb %xmm0, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_ssat_v4i32_v4i8:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vpminsd {{.*}}(%rip), %xmm0, %xmm0
 ; AVX1-NEXT:    vpmaxsd {{.*}}(%rip), %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpackssdw %xmm0, %xmm0, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: trunc_ssat_v4i32_v4i8:
@@ -5096,7 +5098,8 @@
 ; AVX2-NEXT:    vpminsd %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [4294967168,4294967168,4294967168,4294967168]
 ; AVX2-NEXT:    vpmaxsd %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    vpackssdw %xmm0, %xmm0, %xmm0
+; AVX2-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512F-LABEL: trunc_ssat_v4i32_v4i8:
@@ -5178,7 +5181,8 @@
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pminsd {{.*}}(%rip), %xmm0
 ; SSE41-NEXT:    pmaxsd {{.*}}(%rip), %xmm0
-; SSE41-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE41-NEXT:    packssdw %xmm0, %xmm0
+; SSE41-NEXT:    packsswb %xmm0, %xmm0
 ; SSE41-NEXT:    movd %xmm0, (%rdi)
 ; SSE41-NEXT:    retq
 ;
@@ -5186,7 +5190,8 @@
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vpminsd {{.*}}(%rip), %xmm0, %xmm0
 ; AVX1-NEXT:    vpmaxsd {{.*}}(%rip), %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpackssdw %xmm0, %xmm0, %xmm0
+; AVX1-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovd %xmm0, (%rdi)
 ; AVX1-NEXT:    retq
 ;
@@ -5196,7 +5201,8 @@
 ; AVX2-NEXT:    vpminsd %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [4294967168,4294967168,4294967168,4294967168]
 ; AVX2-NEXT:    vpmaxsd %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    vpackssdw %xmm0, %xmm0, %xmm0
+; AVX2-NEXT:    vpacksswb %xmm0, %xmm0, %xmm0
 ; AVX2-NEXT:    vmovd %xmm0, (%rdi)
 ; AVX2-NEXT:    retq
 ;
diff --git a/llvm/test/CodeGen/X86/vector-trunc-usat.ll b/llvm/test/CodeGen/X86/vector-trunc-usat.ll
index 8434bac..4e5f833 100644
--- a/llvm/test/CodeGen/X86/vector-trunc-usat.ll
+++ b/llvm/test/CodeGen/X86/vector-trunc-usat.ll
@@ -3694,20 +3694,23 @@
 ; SSE41-LABEL: trunc_usat_v4i32_v4i8:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pminud {{.*}}(%rip), %xmm0
-; SSE41-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE41-NEXT:    packusdw %xmm0, %xmm0
+; SSE41-NEXT:    packuswb %xmm0, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_usat_v4i32_v4i8:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vpminud {{.*}}(%rip), %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: trunc_usat_v4i32_v4i8:
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [255,255,255,255]
 ; AVX2-NEXT:    vpminud %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX2-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512F-LABEL: trunc_usat_v4i32_v4i8:
@@ -3776,14 +3779,16 @@
 ; SSE41-LABEL: trunc_usat_v4i32_v4i8_store:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    pminud {{.*}}(%rip), %xmm0
-; SSE41-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE41-NEXT:    packusdw %xmm0, %xmm0
+; SSE41-NEXT:    packuswb %xmm0, %xmm0
 ; SSE41-NEXT:    movd %xmm0, (%rdi)
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_usat_v4i32_v4i8_store:
 ; AVX1:       # %bb.0:
 ; AVX1-NEXT:    vpminud {{.*}}(%rip), %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovd %xmm0, (%rdi)
 ; AVX1-NEXT:    retq
 ;
@@ -3791,7 +3796,8 @@
 ; AVX2:       # %bb.0:
 ; AVX2-NEXT:    vpbroadcastd {{.*#+}} xmm1 = [255,255,255,255]
 ; AVX2-NEXT:    vpminud %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    vpackusdw %xmm0, %xmm0, %xmm0
+; AVX2-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX2-NEXT:    vmovd %xmm0, (%rdi)
 ; AVX2-NEXT:    retq
 ;
@@ -3881,24 +3887,20 @@
 ; SSE41-LABEL: trunc_usat_v8i32_v8i8:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255]
-; SSE41-NEXT:    pminud %xmm2, %xmm0
 ; SSE41-NEXT:    pminud %xmm2, %xmm1
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
-; SSE41-NEXT:    pshufb %xmm2, %xmm1
-; SSE41-NEXT:    pshufb %xmm2, %xmm0
-; SSE41-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE41-NEXT:    pminud %xmm2, %xmm0
+; SSE41-NEXT:    packusdw %xmm1, %xmm0
+; SSE41-NEXT:    packuswb %xmm0, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_usat_v8i32_v8i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255,255,255]
-; AVX1-NEXT:    vpminud %xmm1, %xmm0, %xmm2
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; AVX1-NEXT:    vpminud %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
-; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb %xmm1, %xmm2, %xmm1
-; AVX1-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255]
+; AVX1-NEXT:    vpminud %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpminud %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpackusdw %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
@@ -3907,10 +3909,8 @@
 ; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpminud %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
-; AVX2-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX2-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; AVX2-NEXT:    vpackusdw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
@@ -4001,25 +4001,21 @@
 ; SSE41-LABEL: trunc_usat_v8i32_v8i8_store:
 ; SSE41:       # %bb.0:
 ; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255]
-; SSE41-NEXT:    pminud %xmm2, %xmm0
 ; SSE41-NEXT:    pminud %xmm2, %xmm1
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
-; SSE41-NEXT:    pshufb %xmm2, %xmm1
-; SSE41-NEXT:    pshufb %xmm2, %xmm0
-; SSE41-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE41-NEXT:    pminud %xmm2, %xmm0
+; SSE41-NEXT:    packusdw %xmm1, %xmm0
+; SSE41-NEXT:    packuswb %xmm0, %xmm0
 ; SSE41-NEXT:    movq %xmm0, (%rdi)
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: trunc_usat_v8i32_v8i8_store:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,255,255,255]
-; AVX1-NEXT:    vpminud %xmm1, %xmm0, %xmm2
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
-; AVX1-NEXT:    vpminud %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
-; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb %xmm1, %xmm2, %xmm1
-; AVX1-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,255,255]
+; AVX1-NEXT:    vpminud %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpminud %xmm2, %xmm0, %xmm0
+; AVX1-NEXT:    vpackusdw %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovq %xmm0, (%rdi)
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
@@ -4029,10 +4025,8 @@
 ; AVX2-NEXT:    vpbroadcastd {{.*#+}} ymm1 = [255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpminud %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = <0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u>
-; AVX2-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX2-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; AVX2-NEXT:    vpackusdw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpackuswb %xmm0, %xmm0, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, (%rdi)
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq