Index: llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
===================================================================
--- llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
+++ llvm/trunk/lib/Target/X86/X86ISelLowering.cpp
@@ -25756,8 +25756,10 @@
   switch (N->getOpcode()) {
   default:
     llvm_unreachable("Do not know how to custom type legalize this operation!");
+  case X86ISD::ADDUS:
+  case X86ISD::SUBUS:
   case X86ISD::AVG: {
-    // Legalize types for X86ISD::AVG by expanding vectors.
+    // Legalize types for X86ISD::AVG/ADDUS/SUBUS by widening.
     assert(Subtarget.hasSSE2() && "Requires at least SSE2!");
 
     auto InVT = N->getValueType(0);
@@ -25775,7 +25777,7 @@
     Ops[0] = N->getOperand(1);
     SDValue InVec1 = DAG.getNode(ISD::CONCAT_VECTORS, dl, RegVT, Ops);
 
-    SDValue Res = DAG.getNode(X86ISD::AVG, dl, RegVT, InVec0, InVec1);
+    SDValue Res = DAG.getNode(N->getOpcode(), dl, RegVT, InVec0, InVec1);
     if (getTypeAction(*DAG.getContext(), InVT) != TypeWidenVector)
       Res = DAG.getNode(ISD::EXTRACT_SUBVECTOR, dl, InVT, Res,
                         DAG.getIntPtrConstant(0, dl));
@@ -33147,14 +33149,11 @@
     }
   }
 
-  // Early exit check
-  if (!TLI.isTypeLegal(VT))
-    return SDValue();
-
   // Match VSELECTs into subs with unsigned saturation.
   if (N->getOpcode() == ISD::VSELECT && Cond.getOpcode() == ISD::SETCC &&
       // psubus is available in SSE2 for i8 and i16 vectors.
-      Subtarget.hasSSE2() &&
+      Subtarget.hasSSE2() && VT.getVectorNumElements() >= 2 &&
+      isPowerOf2_32(VT.getVectorNumElements()) &&
       (VT.getVectorElementType() == MVT::i8 ||
        VT.getVectorElementType() == MVT::i16)) {
     ISD::CondCode CC = cast<CondCodeSDNode>(Cond.getOperand(2))->get();
@@ -33227,7 +33226,8 @@
   // Match VSELECTs into add with unsigned saturation.
   if (N->getOpcode() == ISD::VSELECT && Cond.getOpcode() == ISD::SETCC &&
       // paddus is available in SSE2 for i8 and i16 vectors.
-      Subtarget.hasSSE2() &&
+      Subtarget.hasSSE2() && VT.getVectorNumElements() >= 2 &&
+      isPowerOf2_32(VT.getVectorNumElements()) &&
       (VT.getVectorElementType() == MVT::i8 ||
        VT.getVectorElementType() == MVT::i16)) {
     ISD::CondCode CC = cast<CondCodeSDNode>(Cond.getOperand(2))->get();
@@ -33283,6 +33283,10 @@
     }
   }
 
+  // Early exit check
+  if (!TLI.isTypeLegal(VT))
+    return SDValue();
+
   if (SDValue V = combineVSelectWithAllOnesOrZeros(N, DAG, DCI, Subtarget))
     return V;
 
Index: llvm/trunk/test/CodeGen/X86/paddus.ll
===================================================================
--- llvm/trunk/test/CodeGen/X86/paddus.ll
+++ llvm/trunk/test/CodeGen/X86/paddus.ll
@@ -1533,99 +1533,36 @@
 }
 
 define void @addus_v8i8(<8 x i8>* %p1, <8 x i8>* %p2) {
-; SSE2-LABEL: addus_v8i8:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
-; SSE2-NEXT:    movq {{.*#+}} xmm2 = mem[0],zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE2-NEXT:    paddw %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    movdqa %xmm2, %xmm3
-; SSE2-NEXT:    pand %xmm1, %xmm3
-; SSE2-NEXT:    pcmpgtw %xmm3, %xmm0
-; SSE2-NEXT:    movdqa %xmm1, %xmm3
-; SSE2-NEXT:    pand %xmm0, %xmm3
-; SSE2-NEXT:    pandn %xmm2, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    packuswb %xmm0, %xmm0
-; SSE2-NEXT:    movq %xmm0, (%rdi)
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: addus_v8i8:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
-; SSSE3-NEXT:    movq {{.*#+}} xmm2 = mem[0],zero
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSSE3-NEXT:    paddw %xmm0, %xmm2
-; SSSE3-NEXT:    pand %xmm1, %xmm0
-; SSSE3-NEXT:    pand %xmm2, %xmm1
-; SSSE3-NEXT:    pcmpgtw %xmm1, %xmm0
-; SSSE3-NEXT:    movdqa %xmm0, %xmm1
-; SSSE3-NEXT:    pandn %xmm2, %xmm1
-; SSSE3-NEXT:    pand {{.*}}(%rip), %xmm0
-; SSSE3-NEXT:    por %xmm1, %xmm0
-; SSSE3-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
-; SSSE3-NEXT:    movq %xmm0, (%rdi)
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: addus_v8i8:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pmovzxbw {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
-; SSE41-NEXT:    pmovzxbw {{.*#+}} xmm2 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; SSE41-NEXT:    paddw %xmm0, %xmm2
-; SSE41-NEXT:    pand %xmm1, %xmm0
-; SSE41-NEXT:    movdqa %xmm2, %xmm3
-; SSE41-NEXT:    pand %xmm1, %xmm3
-; SSE41-NEXT:    pcmpgtw %xmm3, %xmm0
-; SSE41-NEXT:    pblendvb %xmm0, %xmm1, %xmm2
-; SSE41-NEXT:    pshufb {{.*#+}} xmm2 = xmm2[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
-; SSE41-NEXT:    movq %xmm2, (%rdi)
-; SSE41-NEXT:    retq
+; SSE-LABEL: addus_v8i8:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
+; SSE-NEXT:    movq {{.*#+}} xmm1 = mem[0],zero
+; SSE-NEXT:    paddusb %xmm0, %xmm1
+; SSE-NEXT:    movq %xmm1, (%rdi)
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: addus_v8i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
-; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm2
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; AVX1-NEXT:    vpaddw %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm3
-; AVX1-NEXT:    vpcmpgtw %xmm3, %xmm2, %xmm2
-; AVX1-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX1-NEXT:    vpaddusb %xmm0, %xmm1, %xmm0
 ; AVX1-NEXT:    vmovq %xmm0, (%rdi)
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: addus_v8i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
-; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm2
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; AVX2-NEXT:    vpaddw %xmm0, %xmm3, %xmm0
-; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm3
-; AVX2-NEXT:    vpcmpgtw %xmm3, %xmm2, %xmm2
-; AVX2-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX2-NEXT:    vpaddusb %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, (%rdi)
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: addus_v8i8:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpmovzxbw {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; AVX512-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
-; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm2
-; AVX512-NEXT:    vpmovzxbw {{.*#+}} xmm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; AVX512-NEXT:    vpaddw %xmm0, %xmm3, %xmm0
-; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm1
-; AVX512-NEXT:    vpcmpnleuw %xmm1, %xmm2, %k1
-; AVX512-NEXT:    vmovdqu16 {{.*}}(%rip), %xmm0 {%k1}
+; AVX512-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX512-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX512-NEXT:    vpaddusb %xmm0, %xmm1, %xmm0
+; AVX512-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX512-NEXT:    vpmovwb %xmm0, (%rdi)
 ; AVX512-NEXT:    retq
   %ld1 = load <8 x i8>, <8 x i8>* %p1, align 8
@@ -1638,105 +1575,36 @@
 }
 
 define void @addus_v4i8(<4 x i8>* %p1, <4 x i8>* %p2) {
-; SSE2-LABEL: addus_v4i8:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; SSE2-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSE2-NEXT:    paddd %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    movdqa %xmm2, %xmm3
-; SSE2-NEXT:    pand %xmm1, %xmm3
-; SSE2-NEXT:    pcmpgtd %xmm3, %xmm0
-; SSE2-NEXT:    movdqa %xmm1, %xmm3
-; SSE2-NEXT:    pand %xmm0, %xmm3
-; SSE2-NEXT:    pandn %xmm2, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    packuswb %xmm0, %xmm0
-; SSE2-NEXT:    packuswb %xmm0, %xmm0
-; SSE2-NEXT:    movd %xmm0, (%rdi)
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: addus_v4i8:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; SSSE3-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSSE3-NEXT:    paddd %xmm0, %xmm2
-; SSSE3-NEXT:    pand %xmm1, %xmm0
-; SSSE3-NEXT:    pand %xmm2, %xmm1
-; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm0
-; SSSE3-NEXT:    movdqa %xmm0, %xmm1
-; SSSE3-NEXT:    pandn %xmm2, %xmm1
-; SSSE3-NEXT:    pand {{.*}}(%rip), %xmm0
-; SSSE3-NEXT:    por %xmm1, %xmm0
-; SSSE3-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
-; SSSE3-NEXT:    movd %xmm0, (%rdi)
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: addus_v4i8:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm2 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; SSE41-NEXT:    paddd %xmm0, %xmm2
-; SSE41-NEXT:    pand %xmm1, %xmm0
-; SSE41-NEXT:    movdqa %xmm2, %xmm3
-; SSE41-NEXT:    pand %xmm1, %xmm3
-; SSE41-NEXT:    pcmpgtd %xmm3, %xmm0
-; SSE41-NEXT:    blendvps %xmm0, %xmm1, %xmm2
-; SSE41-NEXT:    pshufb {{.*#+}} xmm2 = xmm2[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
-; SSE41-NEXT:    movd %xmm2, (%rdi)
-; SSE41-NEXT:    retq
+; SSE-LABEL: addus_v4i8:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE-NEXT:    paddusb %xmm0, %xmm1
+; SSE-NEXT:    movd %xmm1, (%rdi)
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: addus_v4i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm2
-; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX1-NEXT:    vpaddd %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm3
-; AVX1-NEXT:    vpcmpgtd %xmm3, %xmm2, %xmm2
-; AVX1-NEXT:    vblendvps %xmm2, %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:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX1-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX1-NEXT:    vpaddusb %xmm0, %xmm1, %xmm0
 ; AVX1-NEXT:    vmovd %xmm0, (%rdi)
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: addus_v4i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm2
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX2-NEXT:    vpaddd %xmm0, %xmm3, %xmm0
-; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm1
-; AVX2-NEXT:    vpcmpgtd %xmm1, %xmm2, %xmm1
-; AVX2-NEXT:    vbroadcastss {{.*#+}} xmm2 = [255,255,255,255]
-; AVX2-NEXT:    vblendvps %xmm1, %xmm2, %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:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX2-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX2-NEXT:    vpaddusb %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovd %xmm0, (%rdi)
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: addus_v4i8:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX512-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm2
-; AVX512-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX512-NEXT:    vpaddd %xmm0, %xmm3, %xmm0
-; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm1
-; AVX512-NEXT:    vpcmpnleud %xmm1, %xmm2, %k1
-; AVX512-NEXT:    vpbroadcastd {{.*}}(%rip), %xmm0 {%k1}
+; AVX512-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX512-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX512-NEXT:    vpaddusb %xmm0, %xmm1, %xmm0
+; AVX512-NEXT:    vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
 ; AVX512-NEXT:    vpmovdb %xmm0, (%rdi)
 ; AVX512-NEXT:    retq
   %ld1 = load <4 x i8>, <4 x i8>* %p1, align 4
@@ -1753,39 +1621,10 @@
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    movzwl (%rdi), %eax
 ; SSE2-NEXT:    movd %eax, %xmm0
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[0,1,1,3]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [255,0,0,0,0,0,0,0,255,0,0,0,0,0,0,0]
 ; SSE2-NEXT:    movzwl (%rsi), %eax
-; SSE2-NEXT:    movd %eax, %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,1,1,3]
-; SSE2-NEXT:    paddq %xmm1, %xmm2
-; SSE2-NEXT:    pand %xmm0, %xmm1
-; SSE2-NEXT:    movdqa %xmm2, %xmm3
-; SSE2-NEXT:    pand %xmm0, %xmm3
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [2147483648,0,2147483648,0]
-; SSE2-NEXT:    por %xmm4, %xmm3
-; SSE2-NEXT:    por %xmm4, %xmm1
-; SSE2-NEXT:    movdqa %xmm1, %xmm4
-; SSE2-NEXT:    pcmpgtd %xmm3, %xmm4
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm3, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; SSE2-NEXT:    pand %xmm5, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
-; SSE2-NEXT:    por %xmm1, %xmm3
-; SSE2-NEXT:    movdqa %xmm3, %xmm1
-; SSE2-NEXT:    pandn %xmm2, %xmm1
-; SSE2-NEXT:    pand %xmm0, %xmm3
-; SSE2-NEXT:    por %xmm1, %xmm3
-; SSE2-NEXT:    pand %xmm0, %xmm3
-; SSE2-NEXT:    packuswb %xmm3, %xmm3
-; SSE2-NEXT:    packuswb %xmm3, %xmm3
-; SSE2-NEXT:    packuswb %xmm3, %xmm3
-; SSE2-NEXT:    movd %xmm3, %eax
+; SSE2-NEXT:    movd %eax, %xmm1
+; SSE2-NEXT:    paddusb %xmm0, %xmm1
+; SSE2-NEXT:    movd %xmm1, %eax
 ; SSE2-NEXT:    movw %ax, (%rdi)
 ; SSE2-NEXT:    retq
 ;
@@ -1793,91 +1632,52 @@
 ; SSSE3:       # %bb.0:
 ; SSSE3-NEXT:    movzwl (%rdi), %eax
 ; SSSE3-NEXT:    movd %eax, %xmm0
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [0,128,128,128,128,128,128,128,1,128,128,128,128,128,128,128]
-; SSSE3-NEXT:    pshufb %xmm1, %xmm0
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,0,0,0,0,0,0,255,0,0,0,0,0,0,0]
 ; SSSE3-NEXT:    movzwl (%rsi), %eax
-; SSSE3-NEXT:    movd %eax, %xmm3
-; SSSE3-NEXT:    pshufb %xmm1, %xmm3
-; SSSE3-NEXT:    paddq %xmm0, %xmm3
-; SSSE3-NEXT:    pand %xmm2, %xmm0
-; SSSE3-NEXT:    pand %xmm3, %xmm2
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,0,2147483648,0]
-; SSSE3-NEXT:    por %xmm1, %xmm2
-; SSSE3-NEXT:    por %xmm1, %xmm0
-; SSSE3-NEXT:    movdqa %xmm0, %xmm1
-; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm1
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[0,0,2,2]
-; SSSE3-NEXT:    pcmpeqd %xmm2, %xmm0
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
-; SSSE3-NEXT:    pand %xmm4, %xmm0
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; SSSE3-NEXT:    por %xmm0, %xmm1
-; SSSE3-NEXT:    movdqa %xmm1, %xmm0
-; SSSE3-NEXT:    pandn %xmm3, %xmm0
-; SSSE3-NEXT:    pand {{.*}}(%rip), %xmm1
-; SSSE3-NEXT:    por %xmm0, %xmm1
-; SSSE3-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSSE3-NEXT:    movd %eax, %xmm1
+; SSSE3-NEXT:    paddusb %xmm0, %xmm1
 ; SSSE3-NEXT:    movd %xmm1, %eax
 ; SSSE3-NEXT:    movw %ax, (%rdi)
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: addus_v2i8:
 ; SSE41:       # %bb.0:
-; SSE41-NEXT:    pmovzxbq {{.*#+}} xmm0 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; SSE41-NEXT:    pmovzxbq {{.*#+}} xmm1 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; SSE41-NEXT:    paddq %xmm0, %xmm1
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,0,0,0,0,0,0,255,0,0,0,0,0,0,0]
-; SSE41-NEXT:    pand %xmm1, %xmm2
-; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,0,2147483648,0]
-; SSE41-NEXT:    por %xmm3, %xmm0
-; SSE41-NEXT:    por %xmm3, %xmm2
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    pcmpgtd %xmm2, %xmm3
-; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm0, %xmm2
-; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
-; SSE41-NEXT:    pand %xmm4, %xmm2
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[1,1,3,3]
-; SSE41-NEXT:    por %xmm2, %xmm0
-; SSE41-NEXT:    blendvpd %xmm0, {{.*}}(%rip), %xmm1
-; SSE41-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; SSE41-NEXT:    movzwl (%rdi), %eax
+; SSE41-NEXT:    movd %eax, %xmm0
+; SSE41-NEXT:    movzwl (%rsi), %eax
+; SSE41-NEXT:    movd %eax, %xmm1
+; SSE41-NEXT:    paddusb %xmm0, %xmm1
 ; SSE41-NEXT:    pextrw $0, %xmm1, (%rdi)
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: addus_v2i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpmovzxbq {{.*#+}} xmm0 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX1-NEXT:    vpmovzxbq {{.*#+}} xmm1 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX1-NEXT:    vpaddq %xmm0, %xmm1, %xmm1
-; AVX1-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm2
-; AVX1-NEXT:    vpcmpgtq %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vblendvpd %xmm0, {{.*}}(%rip), %xmm1, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    movzwl (%rdi), %eax
+; AVX1-NEXT:    vmovd %eax, %xmm0
+; AVX1-NEXT:    movzwl (%rsi), %eax
+; AVX1-NEXT:    vmovd %eax, %xmm1
+; AVX1-NEXT:    vpaddusb %xmm0, %xmm1, %xmm0
 ; AVX1-NEXT:    vpextrw $0, %xmm0, (%rdi)
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: addus_v2i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpmovzxbq {{.*#+}} xmm0 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX2-NEXT:    vpmovzxbq {{.*#+}} xmm1 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX2-NEXT:    vpaddq %xmm0, %xmm1, %xmm1
-; AVX2-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm2
-; AVX2-NEXT:    vpcmpgtq %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vblendvpd %xmm0, {{.*}}(%rip), %xmm1, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    movzwl (%rdi), %eax
+; AVX2-NEXT:    vmovd %eax, %xmm0
+; AVX2-NEXT:    movzwl (%rsi), %eax
+; AVX2-NEXT:    vmovd %eax, %xmm1
+; AVX2-NEXT:    vpaddusb %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vpextrw $0, %xmm0, (%rdi)
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: addus_v2i8:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpmovzxbq {{.*#+}} xmm0 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vpmovzxbq {{.*#+}} xmm1 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vpaddq %xmm0, %xmm1, %xmm1
-; AVX512-NEXT:    vpand {{.*}}(%rip), %xmm1, %xmm2
-; AVX512-NEXT:    vpcmpnleuq %xmm2, %xmm0, %k1
-; AVX512-NEXT:    vmovdqa64 {{.*}}(%rip), %xmm1 {%k1}
-; AVX512-NEXT:    vpmovqb %xmm1, (%rdi)
+; AVX512-NEXT:    movzwl (%rdi), %eax
+; AVX512-NEXT:    vmovd %eax, %xmm0
+; AVX512-NEXT:    movzwl (%rsi), %eax
+; AVX512-NEXT:    vmovd %eax, %xmm1
+; AVX512-NEXT:    vpaddusb %xmm0, %xmm1, %xmm0
+; AVX512-NEXT:    vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
+; AVX512-NEXT:    vpmovqb %xmm0, (%rdi)
 ; AVX512-NEXT:    retq
   %ld1 = load <2 x i8>, <2 x i8>* %p1, align 2
   %ld2 = load <2 x i8>, <2 x i8>* %p2, align 2
@@ -1889,97 +1689,36 @@
 }
 
 define void @addus_v4i16(<4 x i16>* %p1, <4 x i16>* %p2) {
-; SSE2-LABEL: addus_v4i16:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [65535,0,65535,0,65535,0,65535,0]
-; SSE2-NEXT:    movq {{.*#+}} xmm2 = mem[0],zero
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSE2-NEXT:    paddd %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    pand %xmm2, %xmm1
-; SSE2-NEXT:    pcmpgtd %xmm1, %xmm0
-; SSE2-NEXT:    movdqa %xmm0, %xmm1
-; SSE2-NEXT:    pandn %xmm2, %xmm1
-; SSE2-NEXT:    pand {{.*}}(%rip), %xmm0
-; SSE2-NEXT:    por %xmm1, %xmm0
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
-; SSE2-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,6,7]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; SSE2-NEXT:    movq %xmm0, (%rdi)
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: addus_v4i16:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [65535,0,65535,0,65535,0,65535,0]
-; SSSE3-NEXT:    movq {{.*#+}} xmm2 = mem[0],zero
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSSE3-NEXT:    paddd %xmm0, %xmm2
-; SSSE3-NEXT:    pand %xmm1, %xmm0
-; SSSE3-NEXT:    pand %xmm2, %xmm1
-; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm0
-; SSSE3-NEXT:    movdqa %xmm0, %xmm1
-; SSSE3-NEXT:    pandn %xmm2, %xmm1
-; SSSE3-NEXT:    pand {{.*}}(%rip), %xmm0
-; SSSE3-NEXT:    por %xmm1, %xmm0
-; SSSE3-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; SSSE3-NEXT:    movq %xmm0, (%rdi)
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: addus_v4i16:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
-; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
-; SSE41-NEXT:    paddd %xmm0, %xmm1
-; SSE41-NEXT:    pxor %xmm2, %xmm2
-; SSE41-NEXT:    pblendw {{.*#+}} xmm2 = xmm1[0],xmm2[1],xmm1[2],xmm2[3],xmm1[4],xmm2[5],xmm1[6],xmm2[7]
-; SSE41-NEXT:    pcmpgtd %xmm2, %xmm0
-; SSE41-NEXT:    blendvps %xmm0, {{.*}}(%rip), %xmm1
-; SSE41-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; SSE41-NEXT:    movq %xmm1, (%rdi)
-; SSE41-NEXT:    retq
+; SSE-LABEL: addus_v4i16:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
+; SSE-NEXT:    movq {{.*#+}} xmm1 = mem[0],zero
+; SSE-NEXT:    paddusw %xmm0, %xmm1
+; SSE-NEXT:    movq %xmm1, (%rdi)
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: addus_v4i16:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
-; AVX1-NEXT:    vpaddd %xmm0, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm1[0],xmm2[1],xmm1[2],xmm2[3],xmm1[4],xmm2[5],xmm1[6],xmm2[7]
-; AVX1-NEXT:    vpcmpgtd %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vblendvps %xmm0, {{.*}}(%rip), %xmm1, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX1-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX1-NEXT:    vpaddusw %xmm0, %xmm1, %xmm0
 ; AVX1-NEXT:    vmovq %xmm0, (%rdi)
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: addus_v4i16:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpmovzxwd {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
-; AVX2-NEXT:    vpmovzxwd {{.*#+}} xmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
-; AVX2-NEXT:    vpaddd %xmm0, %xmm1, %xmm1
-; AVX2-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX2-NEXT:    vpblendw {{.*#+}} xmm2 = xmm1[0],xmm2[1],xmm1[2],xmm2[3],xmm1[4],xmm2[5],xmm1[6],xmm2[7]
-; AVX2-NEXT:    vpcmpgtd %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vbroadcastss {{.*#+}} xmm2 = [65535,65535,65535,65535]
-; AVX2-NEXT:    vblendvps %xmm0, %xmm2, %xmm1, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX2-NEXT:    vpaddusw %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, (%rdi)
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: addus_v4i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
-; AVX512-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512-NEXT:    vpunpcklwd {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; AVX512-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX512-NEXT:    vpaddusw %xmm0, %xmm1, %xmm0
 ; AVX512-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX512-NEXT:    vpmovzxwd {{.*#+}} xmm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
-; AVX512-NEXT:    vpaddd %xmm0, %xmm3, %xmm0
-; AVX512-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0],xmm1[1],xmm0[2],xmm1[3],xmm0[4],xmm1[5],xmm0[6],xmm1[7]
-; AVX512-NEXT:    vpcmpnleud %xmm1, %xmm2, %k1
-; AVX512-NEXT:    vpbroadcastd {{.*}}(%rip), %xmm0 {%k1}
 ; AVX512-NEXT:    vpmovdw %xmm0, (%rdi)
 ; AVX512-NEXT:    retq
   %ld1 = load <4 x i16>, <4 x i16>* %p1, align 4
@@ -1992,132 +1731,36 @@
 }
 
 define void @addus_v2i16(<2 x i16>* %p1, <2 x i16>* %p2) {
-; SSE2-LABEL: addus_v2i16:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
-; SSE2-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,5,6,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [65535,0,0,0,65535,0,0,0]
-; SSE2-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,1,0,3]
-; SSE2-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,5,5,6,7]
-; SSE2-NEXT:    paddq %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm1, %xmm0
-; SSE2-NEXT:    pand %xmm2, %xmm1
-; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,0,2147483648,0]
-; SSE2-NEXT:    por %xmm3, %xmm0
-; SSE2-NEXT:    por %xmm3, %xmm1
-; SSE2-NEXT:    movdqa %xmm0, %xmm3
-; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
-; SSE2-NEXT:    pand %xmm4, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[1,1,3,3]
-; SSE2-NEXT:    por %xmm0, %xmm1
-; SSE2-NEXT:    movdqa %xmm1, %xmm0
-; SSE2-NEXT:    pandn %xmm2, %xmm0
-; SSE2-NEXT:    pand {{.*}}(%rip), %xmm1
-; SSE2-NEXT:    por %xmm0, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
-; SSE2-NEXT:    movd %xmm0, (%rdi)
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: addus_v2i16:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
-; SSSE3-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,5,6,7]
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [65535,0,0,0,65535,0,0,0]
-; SSSE3-NEXT:    movd {{.*#+}} xmm2 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[0,1,0,3]
-; SSSE3-NEXT:    pshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,5,5,6,7]
-; SSSE3-NEXT:    paddq %xmm0, %xmm2
-; SSSE3-NEXT:    pand %xmm1, %xmm0
-; SSSE3-NEXT:    pand %xmm2, %xmm1
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,0,2147483648,0]
-; SSSE3-NEXT:    por %xmm3, %xmm0
-; SSSE3-NEXT:    por %xmm3, %xmm1
-; SSSE3-NEXT:    movdqa %xmm0, %xmm3
-; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm3
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
-; SSSE3-NEXT:    pcmpeqd %xmm0, %xmm1
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
-; SSSE3-NEXT:    pand %xmm4, %xmm0
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[1,1,3,3]
-; SSSE3-NEXT:    por %xmm0, %xmm1
-; SSSE3-NEXT:    movdqa %xmm1, %xmm0
-; SSSE3-NEXT:    pandn %xmm2, %xmm0
-; SSSE3-NEXT:    pand {{.*}}(%rip), %xmm1
-; SSSE3-NEXT:    por %xmm0, %xmm1
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
-; SSSE3-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
-; SSSE3-NEXT:    movd %xmm0, (%rdi)
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: addus_v2i16:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pmovzxwq {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; SSE41-NEXT:    pmovzxwq {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; SSE41-NEXT:    paddq %xmm0, %xmm1
-; SSE41-NEXT:    pxor %xmm2, %xmm2
-; SSE41-NEXT:    pblendw {{.*#+}} xmm2 = xmm1[0],xmm2[1,2,3],xmm1[4],xmm2[5,6,7]
-; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,0,2147483648,0]
-; SSE41-NEXT:    por %xmm3, %xmm0
-; SSE41-NEXT:    por %xmm3, %xmm2
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    pcmpgtd %xmm2, %xmm3
-; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm0, %xmm2
-; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
-; SSE41-NEXT:    pand %xmm4, %xmm2
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm3[1,1,3,3]
-; SSE41-NEXT:    por %xmm2, %xmm0
-; SSE41-NEXT:    blendvpd %xmm0, {{.*}}(%rip), %xmm1
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
-; SSE41-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
-; SSE41-NEXT:    movd %xmm0, (%rdi)
-; SSE41-NEXT:    retq
+; SSE-LABEL: addus_v2i16:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE-NEXT:    paddusw %xmm0, %xmm1
+; SSE-NEXT:    movd %xmm1, (%rdi)
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: addus_v2i16:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpmovzxwq {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX1-NEXT:    vpmovzxwq {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX1-NEXT:    vpaddq %xmm0, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm1[0],xmm2[1,2,3],xmm1[4],xmm2[5,6,7]
-; AVX1-NEXT:    vpcmpgtq %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vblendvpd %xmm0, {{.*}}(%rip), %xmm1, %xmm0
-; AVX1-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX1-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX1-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX1-NEXT:    vpaddusw %xmm0, %xmm1, %xmm0
 ; AVX1-NEXT:    vmovd %xmm0, (%rdi)
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: addus_v2i16:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpmovzxwq {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX2-NEXT:    vpmovzxwq {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX2-NEXT:    vpaddq %xmm0, %xmm1, %xmm1
-; AVX2-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX2-NEXT:    vpblendw {{.*#+}} xmm2 = xmm1[0],xmm2[1,2,3],xmm1[4],xmm2[5,6,7]
-; AVX2-NEXT:    vpcmpgtq %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vblendvpd %xmm0, {{.*}}(%rip), %xmm1, %xmm0
-; AVX2-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; AVX2-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX2-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX2-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX2-NEXT:    vpaddusw %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovd %xmm0, (%rdi)
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: addus_v2i16:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpmovzxwq {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX512-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512-NEXT:    vpblendw {{.*#+}} xmm2 = xmm0[0],xmm1[1,2,3],xmm0[4],xmm1[5,6,7]
-; AVX512-NEXT:    vpmovzxwq {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX512-NEXT:    vpaddq %xmm0, %xmm3, %xmm0
-; AVX512-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3],xmm0[4],xmm1[5,6,7]
-; AVX512-NEXT:    vpcmpnleuq %xmm1, %xmm2, %k1
-; AVX512-NEXT:    vmovdqa64 {{.*}}(%rip), %xmm0 {%k1}
+; AVX512-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX512-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX512-NEXT:    vpaddusw %xmm0, %xmm1, %xmm0
+; AVX512-NEXT:    vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
 ; AVX512-NEXT:    vpmovqw %xmm0, (%rdi)
 ; AVX512-NEXT:    retq
   %ld1 = load <2 x i16>, <2 x i16>* %p1, align 2
Index: llvm/trunk/test/CodeGen/X86/psubus.ll
===================================================================
--- llvm/trunk/test/CodeGen/X86/psubus.ll
+++ llvm/trunk/test/CodeGen/X86/psubus.ll
@@ -2213,93 +2213,36 @@
 }
 
 define void @subus_v8i8(<8 x i8>* %p1, <8 x i8>* %p2) {
-; SSE2-LABEL: subus_v8i8:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
-; SSE2-NEXT:    movdqa %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm1, %xmm2
-; SSE2-NEXT:    movq {{.*#+}} xmm3 = mem[0],zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSE2-NEXT:    pand %xmm3, %xmm1
-; SSE2-NEXT:    pcmpgtw %xmm1, %xmm2
-; SSE2-NEXT:    psubw %xmm3, %xmm0
-; SSE2-NEXT:    pand %xmm2, %xmm0
-; SSE2-NEXT:    pand {{.*}}(%rip), %xmm0
-; SSE2-NEXT:    packuswb %xmm0, %xmm0
-; SSE2-NEXT:    movq %xmm0, (%rdi)
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: subus_v8i8:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
-; SSSE3-NEXT:    movdqa %xmm0, %xmm2
-; SSSE3-NEXT:    pand %xmm1, %xmm2
-; SSSE3-NEXT:    movq {{.*#+}} xmm3 = mem[0],zero
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSSE3-NEXT:    pand %xmm3, %xmm1
-; SSSE3-NEXT:    pcmpgtw %xmm1, %xmm2
-; SSSE3-NEXT:    psubw %xmm3, %xmm0
-; SSSE3-NEXT:    pand %xmm2, %xmm0
-; SSSE3-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
-; SSSE3-NEXT:    movq %xmm0, (%rdi)
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: subus_v8i8:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pmovzxbw {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
-; SSE41-NEXT:    movdqa %xmm0, %xmm2
-; SSE41-NEXT:    pand %xmm1, %xmm2
-; SSE41-NEXT:    pmovzxbw {{.*#+}} xmm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; SSE41-NEXT:    pand %xmm3, %xmm1
-; SSE41-NEXT:    pcmpgtw %xmm1, %xmm2
-; SSE41-NEXT:    psubw %xmm3, %xmm0
-; SSE41-NEXT:    pand %xmm2, %xmm0
-; SSE41-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
-; SSE41-NEXT:    movq %xmm0, (%rdi)
-; SSE41-NEXT:    retq
+; SSE-LABEL: subus_v8i8:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
+; SSE-NEXT:    movq {{.*#+}} xmm1 = mem[0],zero
+; SSE-NEXT:    psubusb %xmm1, %xmm0
+; SSE-NEXT:    movq %xmm0, (%rdi)
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: subus_v8i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
-; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm2
-; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; AVX1-NEXT:    vpand %xmm1, %xmm3, %xmm1
-; AVX1-NEXT:    vpcmpgtw %xmm1, %xmm2, %xmm1
-; AVX1-NEXT:    vpsubw %xmm3, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm0, %xmm1, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX1-NEXT:    vpsubusb %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovq %xmm0, (%rdi)
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: subus_v8i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
-; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm2
-; AVX2-NEXT:    vpmovzxbw {{.*#+}} xmm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; AVX2-NEXT:    vpand %xmm1, %xmm3, %xmm1
-; AVX2-NEXT:    vpcmpgtw %xmm1, %xmm2, %xmm1
-; AVX2-NEXT:    vpsubw %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm0, %xmm1, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX2-NEXT:    vpsubusb %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, (%rdi)
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: subus_v8i8:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpmovzxbw {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; AVX512-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
-; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm2
-; AVX512-NEXT:    vpmovzxbw {{.*#+}} xmm3 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero,mem[4],zero,mem[5],zero,mem[6],zero,mem[7],zero
-; AVX512-NEXT:    vpand %xmm1, %xmm3, %xmm1
-; AVX512-NEXT:    vpcmpnleuw %xmm1, %xmm2, %k1
-; AVX512-NEXT:    vpsubw %xmm3, %xmm0, %xmm0 {%k1} {z}
+; AVX512-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX512-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX512-NEXT:    vpsubusb %xmm1, %xmm0, %xmm0
+; AVX512-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX512-NEXT:    vpmovwb %xmm0, (%rdi)
 ; AVX512-NEXT:    retq
   %ld1 = load <8 x i8>, <8 x i8>* %p1, align 8
@@ -2312,98 +2255,36 @@
 }
 
 define void @subus_v4i8(<4 x i8>* %p1, <4 x i8>* %p2) {
-; SSE2-LABEL: subus_v4i8:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; SSE2-NEXT:    movdqa %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm1, %xmm2
-; SSE2-NEXT:    movd {{.*#+}} xmm3 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
-; SSE2-NEXT:    pand %xmm3, %xmm1
-; SSE2-NEXT:    pcmpgtd %xmm1, %xmm2
-; SSE2-NEXT:    psubd %xmm3, %xmm0
-; SSE2-NEXT:    pand %xmm2, %xmm0
-; SSE2-NEXT:    pand {{.*}}(%rip), %xmm0
-; SSE2-NEXT:    packuswb %xmm0, %xmm0
-; SSE2-NEXT:    packuswb %xmm0, %xmm0
-; SSE2-NEXT:    movd %xmm0, (%rdi)
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: subus_v4i8:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; SSSE3-NEXT:    movdqa %xmm0, %xmm2
-; SSSE3-NEXT:    pand %xmm1, %xmm2
-; SSSE3-NEXT:    movd {{.*#+}} xmm3 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
-; SSSE3-NEXT:    pand %xmm3, %xmm1
-; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm2
-; SSSE3-NEXT:    psubd %xmm3, %xmm0
-; SSSE3-NEXT:    pand %xmm2, %xmm0
-; SSSE3-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
-; SSSE3-NEXT:    movd %xmm0, (%rdi)
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: subus_v4i8:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; SSE41-NEXT:    movdqa %xmm0, %xmm2
-; SSE41-NEXT:    pand %xmm1, %xmm2
-; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; SSE41-NEXT:    pand %xmm3, %xmm1
-; SSE41-NEXT:    pcmpgtd %xmm1, %xmm2
-; SSE41-NEXT:    psubd %xmm3, %xmm0
-; SSE41-NEXT:    pand %xmm2, %xmm0
-; SSE41-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
-; SSE41-NEXT:    movd %xmm0, (%rdi)
-; SSE41-NEXT:    retq
+; SSE-LABEL: subus_v4i8:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE-NEXT:    psubusb %xmm1, %xmm0
+; SSE-NEXT:    movd %xmm0, (%rdi)
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: subus_v4i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm2
-; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX1-NEXT:    vpand %xmm1, %xmm3, %xmm1
-; AVX1-NEXT:    vpcmpgtd %xmm1, %xmm2, %xmm1
-; AVX1-NEXT:    vpsubd %xmm3, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm0, %xmm1, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX1-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX1-NEXT:    vpsubusb %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovd %xmm0, (%rdi)
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: subus_v4i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm2
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX2-NEXT:    vpand %xmm1, %xmm3, %xmm1
-; AVX2-NEXT:    vpcmpgtd %xmm1, %xmm2, %xmm1
-; AVX2-NEXT:    vpsubd %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm0, %xmm1, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,4,8,12,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX2-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX2-NEXT:    vpsubusb %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vmovd %xmm0, (%rdi)
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: subus_v4i8:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX512-NEXT:    vmovdqa {{.*#+}} xmm1 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; AVX512-NEXT:    vpand %xmm1, %xmm0, %xmm2
-; AVX512-NEXT:    vpmovzxbd {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
-; AVX512-NEXT:    vpand %xmm1, %xmm3, %xmm1
-; AVX512-NEXT:    vpcmpnleud %xmm1, %xmm2, %k1
-; AVX512-NEXT:    vpsubd %xmm3, %xmm0, %xmm0 {%k1} {z}
+; AVX512-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX512-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX512-NEXT:    vpsubusb %xmm1, %xmm0, %xmm0
+; AVX512-NEXT:    vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
 ; AVX512-NEXT:    vpmovdb %xmm0, (%rdi)
 ; AVX512-NEXT:    retq
   %ld1 = load <4 x i8>, <4 x i8>* %p1, align 8
@@ -2419,117 +2300,51 @@
 ; SSE2-LABEL: subus_v2i8:
 ; SSE2:       # %bb.0:
 ; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,1,3]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [255,0,0,0,0,0,0,0,255,0,0,0,0,0,0,0]
-; SSE2-NEXT:    movdqa %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm1, %xmm2
-; SSE2-NEXT:    movd {{.*#+}} xmm3 = mem[0],zero,zero,zero
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,1,1,3]
-; SSE2-NEXT:    pand %xmm3, %xmm1
-; SSE2-NEXT:    psubq %xmm3, %xmm0
-; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,0,2147483648,0]
-; SSE2-NEXT:    por %xmm3, %xmm1
-; SSE2-NEXT:    por %xmm3, %xmm2
-; SSE2-NEXT:    movdqa %xmm2, %xmm3
-; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm1, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
-; SSE2-NEXT:    pand %xmm4, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
-; SSE2-NEXT:    por %xmm1, %xmm2
-; SSE2-NEXT:    pand %xmm0, %xmm2
-; SSE2-NEXT:    pand {{.*}}(%rip), %xmm2
-; SSE2-NEXT:    packuswb %xmm2, %xmm2
-; SSE2-NEXT:    packuswb %xmm2, %xmm2
-; SSE2-NEXT:    packuswb %xmm2, %xmm2
-; SSE2-NEXT:    movd %xmm2, %eax
+; SSE2-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE2-NEXT:    psubusb %xmm1, %xmm0
+; SSE2-NEXT:    movd %xmm0, %eax
 ; SSE2-NEXT:    movw %ax, (%rdi)
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: subus_v2i8:
 ; SSSE3:       # %bb.0:
 ; SSSE3-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [0,128,128,128,128,128,128,128,1,128,128,128,128,128,128,128]
-; SSSE3-NEXT:    pshufb %xmm1, %xmm0
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,0,0,0,0,0,0,255,0,0,0,0,0,0,0]
-; SSSE3-NEXT:    movdqa %xmm0, %xmm3
-; SSSE3-NEXT:    pand %xmm2, %xmm3
-; SSSE3-NEXT:    movd {{.*#+}} xmm4 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    pshufb %xmm1, %xmm4
-; SSSE3-NEXT:    pand %xmm4, %xmm2
-; SSSE3-NEXT:    psubq %xmm4, %xmm0
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [2147483648,0,2147483648,0]
-; SSSE3-NEXT:    por %xmm1, %xmm2
-; SSSE3-NEXT:    por %xmm1, %xmm3
-; SSSE3-NEXT:    movdqa %xmm3, %xmm1
-; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm1
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[0,0,2,2]
-; SSSE3-NEXT:    pcmpeqd %xmm2, %xmm3
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
-; SSSE3-NEXT:    pand %xmm4, %xmm2
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; SSSE3-NEXT:    por %xmm2, %xmm1
-; SSSE3-NEXT:    pand %xmm0, %xmm1
-; SSSE3-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
-; SSSE3-NEXT:    movd %xmm1, %eax
+; SSSE3-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSSE3-NEXT:    psubusb %xmm1, %xmm0
+; SSSE3-NEXT:    movd %xmm0, %eax
 ; SSSE3-NEXT:    movw %ax, (%rdi)
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: subus_v2i8:
 ; SSE41:       # %bb.0:
-; SSE41-NEXT:    pmovzxbq {{.*#+}} xmm0 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; SSE41-NEXT:    pmovzxbq {{.*#+}} xmm1 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; SSE41-NEXT:    movdqa %xmm0, %xmm2
-; SSE41-NEXT:    psubq %xmm1, %xmm2
-; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,0,2147483648,0]
-; SSE41-NEXT:    por %xmm3, %xmm1
-; SSE41-NEXT:    por %xmm3, %xmm0
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    pcmpgtd %xmm1, %xmm3
-; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm1, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
-; SSE41-NEXT:    pand %xmm4, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[1,1,3,3]
-; SSE41-NEXT:    por %xmm0, %xmm1
-; SSE41-NEXT:    pand %xmm2, %xmm1
-; SSE41-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
-; SSE41-NEXT:    pextrw $0, %xmm1, (%rdi)
+; SSE41-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE41-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE41-NEXT:    psubusb %xmm1, %xmm0
+; SSE41-NEXT:    pextrw $0, %xmm0, (%rdi)
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: subus_v2i8:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpmovzxbq {{.*#+}} xmm0 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX1-NEXT:    vpmovzxbq {{.*#+}} xmm1 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX1-NEXT:    vpsubq %xmm1, %xmm0, %xmm2
-; AVX1-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX1-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX1-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX1-NEXT:    vpsubusb %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vpextrw $0, %xmm0, (%rdi)
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: subus_v2i8:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpmovzxbq {{.*#+}} xmm0 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX2-NEXT:    vpmovzxbq {{.*#+}} xmm1 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX2-NEXT:    vpsubq %xmm1, %xmm0, %xmm2
-; AVX2-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,8,u,u,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX2-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX2-NEXT:    vpsubusb %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vpextrw $0, %xmm0, (%rdi)
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: subus_v2i8:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpmovzxbq {{.*#+}} xmm0 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vpmovzxbq {{.*#+}} xmm1 = mem[0],zero,zero,zero,zero,zero,zero,zero,mem[1],zero,zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vpcmpnleuq %xmm1, %xmm0, %k1
-; AVX512-NEXT:    vpsubq %xmm1, %xmm0, %xmm0 {%k1} {z}
+; AVX512-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX512-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX512-NEXT:    vpsubusb %xmm1, %xmm0, %xmm0
+; AVX512-NEXT:    vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
 ; AVX512-NEXT:    vpmovqb %xmm0, (%rdi)
 ; AVX512-NEXT:    retq
   %ld1 = load <2 x i8>, <2 x i8>* %p1, align 8
@@ -2542,87 +2357,36 @@
 }
 
 define void @subus_v4i16(<4 x i16>* %p1, <4 x i16>* %p2) {
-; SSE2-LABEL: subus_v4i16:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [65535,0,65535,0,65535,0,65535,0]
-; SSE2-NEXT:    movdqa %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm1, %xmm2
-; SSE2-NEXT:    movq {{.*#+}} xmm3 = mem[0],zero
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
-; SSE2-NEXT:    pand %xmm3, %xmm1
-; SSE2-NEXT:    pcmpgtd %xmm1, %xmm2
-; SSE2-NEXT:    psubd %xmm3, %xmm0
-; SSE2-NEXT:    pand %xmm2, %xmm0
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
-; SSE2-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,6,7]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; SSE2-NEXT:    movq %xmm0, (%rdi)
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: subus_v4i16:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [65535,0,65535,0,65535,0,65535,0]
-; SSSE3-NEXT:    movdqa %xmm0, %xmm2
-; SSSE3-NEXT:    pand %xmm1, %xmm2
-; SSSE3-NEXT:    movq {{.*#+}} xmm3 = mem[0],zero
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
-; SSSE3-NEXT:    pand %xmm3, %xmm1
-; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm2
-; SSSE3-NEXT:    psubd %xmm3, %xmm0
-; SSSE3-NEXT:    pand %xmm2, %xmm0
-; SSSE3-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; SSSE3-NEXT:    movq %xmm0, (%rdi)
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: subus_v4i16:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
-; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
-; SSE41-NEXT:    movdqa %xmm0, %xmm2
-; SSE41-NEXT:    psubd %xmm1, %xmm2
-; SSE41-NEXT:    pcmpgtd %xmm1, %xmm0
-; SSE41-NEXT:    pand %xmm2, %xmm0
-; SSE41-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; SSE41-NEXT:    movq %xmm0, (%rdi)
-; SSE41-NEXT:    retq
+; SSE-LABEL: subus_v4i16:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
+; SSE-NEXT:    movq {{.*#+}} xmm1 = mem[0],zero
+; SSE-NEXT:    psubusw %xmm1, %xmm0
+; SSE-NEXT:    movq %xmm0, (%rdi)
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: subus_v4i16:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
-; AVX1-NEXT:    vpsubd %xmm1, %xmm0, %xmm2
-; AVX1-NEXT:    vpcmpgtd %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX1-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX1-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX1-NEXT:    vpsubusw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovq %xmm0, (%rdi)
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: subus_v4i16:
 ; AVX2:       # %bb.0:
-; AVX2-NEXT:    vpmovzxwd {{.*#+}} xmm0 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
-; AVX2-NEXT:    vpmovzxwd {{.*#+}} xmm1 = mem[0],zero,mem[1],zero,mem[2],zero,mem[3],zero
-; AVX2-NEXT:    vpsubd %xmm1, %xmm0, %xmm2
-; AVX2-NEXT:    vpcmpgtd %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX2-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX2-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX2-NEXT:    vpsubusw %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vmovq %xmm0, (%rdi)
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: subus_v4i16:
 ; AVX512:       # %bb.0:
 ; AVX512-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
-; AVX512-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512-NEXT:    vpunpcklwd {{.*#+}} xmm2 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; AVX512-NEXT:    vmovq {{.*#+}} xmm1 = mem[0],zero
+; AVX512-NEXT:    vpsubusw %xmm1, %xmm0, %xmm0
 ; AVX512-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX512-NEXT:    vmovq {{.*#+}} xmm3 = mem[0],zero
-; AVX512-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3]
-; AVX512-NEXT:    vpmovzxwd {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero
-; AVX512-NEXT:    vpcmpnleud %xmm1, %xmm2, %k1
-; AVX512-NEXT:    vpsubd %xmm3, %xmm0, %xmm0 {%k1} {z}
 ; AVX512-NEXT:    vpmovdw %xmm0, (%rdi)
 ; AVX512-NEXT:    retq
   %ld1 = load <4 x i16>, <4 x i16>* %p1, align 8
@@ -2635,133 +2399,36 @@
 }
 
 define void @subus_v2i16(<2 x i16>* %p1, <2 x i16>* %p2) {
-; SSE2-LABEL: subus_v2i16:
-; SSE2:       # %bb.0:
-; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
-; SSE2-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,5,6,7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm1 = [65535,0,0,0,65535,0,0,0]
-; SSE2-NEXT:    movdqa %xmm0, %xmm2
-; SSE2-NEXT:    pand %xmm1, %xmm2
-; SSE2-NEXT:    movd {{.*#+}} xmm3 = mem[0],zero,zero,zero
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,1,0,3]
-; SSE2-NEXT:    pshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,5,5,6,7]
-; SSE2-NEXT:    pand %xmm3, %xmm1
-; SSE2-NEXT:    psubq %xmm3, %xmm0
-; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,0,2147483648,0]
-; SSE2-NEXT:    por %xmm3, %xmm1
-; SSE2-NEXT:    por %xmm3, %xmm2
-; SSE2-NEXT:    movdqa %xmm2, %xmm3
-; SSE2-NEXT:    pcmpgtd %xmm1, %xmm3
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm1, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
-; SSE2-NEXT:    pand %xmm4, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
-; SSE2-NEXT:    por %xmm1, %xmm2
-; SSE2-NEXT:    pand %xmm0, %xmm2
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
-; SSE2-NEXT:    movd %xmm0, (%rdi)
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: subus_v2i16:
-; SSSE3:       # %bb.0:
-; SSSE3-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
-; SSSE3-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,5,5,6,7]
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [65535,0,0,0,65535,0,0,0]
-; SSSE3-NEXT:    movdqa %xmm0, %xmm2
-; SSSE3-NEXT:    pand %xmm1, %xmm2
-; SSSE3-NEXT:    movd {{.*#+}} xmm3 = mem[0],zero,zero,zero
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,1,0,3]
-; SSSE3-NEXT:    pshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,5,5,6,7]
-; SSSE3-NEXT:    pand %xmm3, %xmm1
-; SSSE3-NEXT:    psubq %xmm3, %xmm0
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,0,2147483648,0]
-; SSSE3-NEXT:    por %xmm3, %xmm1
-; SSSE3-NEXT:    por %xmm3, %xmm2
-; SSSE3-NEXT:    movdqa %xmm2, %xmm3
-; SSSE3-NEXT:    pcmpgtd %xmm1, %xmm3
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
-; SSSE3-NEXT:    pcmpeqd %xmm1, %xmm2
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm2[1,1,3,3]
-; SSSE3-NEXT:    pand %xmm4, %xmm1
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
-; SSSE3-NEXT:    por %xmm1, %xmm2
-; SSSE3-NEXT:    pand %xmm0, %xmm2
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,2,3]
-; SSSE3-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
-; SSSE3-NEXT:    movd %xmm0, (%rdi)
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: subus_v2i16:
-; SSE41:       # %bb.0:
-; SSE41-NEXT:    pmovzxwq {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; SSE41-NEXT:    pmovzxwq {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; SSE41-NEXT:    movdqa %xmm0, %xmm2
-; SSE41-NEXT:    psubq %xmm1, %xmm2
-; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,0,2147483648,0]
-; SSE41-NEXT:    por %xmm3, %xmm1
-; SSE41-NEXT:    por %xmm3, %xmm0
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    pcmpgtd %xmm1, %xmm3
-; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[0,0,2,2]
-; SSE41-NEXT:    pcmpeqd %xmm1, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
-; SSE41-NEXT:    pand %xmm4, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[1,1,3,3]
-; SSE41-NEXT:    por %xmm0, %xmm1
-; SSE41-NEXT:    pand %xmm2, %xmm1
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
-; SSE41-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
-; SSE41-NEXT:    movd %xmm0, (%rdi)
-; SSE41-NEXT:    retq
+; SSE-LABEL: subus_v2i16:
+; SSE:       # %bb.0:
+; SSE-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; SSE-NEXT:    psubusw %xmm1, %xmm0
+; SSE-NEXT:    movd %xmm0, (%rdi)
+; SSE-NEXT:    retq
 ;
 ; AVX1-LABEL: subus_v2i16:
 ; AVX1:       # %bb.0:
-; AVX1-NEXT:    vpmovzxwq {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX1-NEXT:    vpmovzxwq {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX1-NEXT:    vpsubq %xmm1, %xmm0, %xmm2
-; AVX1-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; AVX1-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX1-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX1-NEXT:    vpsubusw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovd %xmm0, (%rdi)
 ; AVX1-NEXT:    retq
 ;
-; AVX2-SLOW-LABEL: subus_v2i16:
-; AVX2-SLOW:       # %bb.0:
-; AVX2-SLOW-NEXT:    vpmovzxwq {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX2-SLOW-NEXT:    vpmovzxwq {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX2-SLOW-NEXT:    vpsubq %xmm1, %xmm0, %xmm2
-; AVX2-SLOW-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm0
-; AVX2-SLOW-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX2-SLOW-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; AVX2-SLOW-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
-; AVX2-SLOW-NEXT:    vmovd %xmm0, (%rdi)
-; AVX2-SLOW-NEXT:    retq
-;
-; AVX2-FAST-LABEL: subus_v2i16:
-; AVX2-FAST:       # %bb.0:
-; AVX2-FAST-NEXT:    vpmovzxwq {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX2-FAST-NEXT:    vpmovzxwq {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX2-FAST-NEXT:    vpsubq %xmm1, %xmm0, %xmm2
-; AVX2-FAST-NEXT:    vpcmpgtq %xmm1, %xmm0, %xmm0
-; AVX2-FAST-NEXT:    vpand %xmm2, %xmm0, %xmm0
-; AVX2-FAST-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,10,11,8,9,10,11,12,13,14,15]
-; AVX2-FAST-NEXT:    vmovd %xmm0, (%rdi)
-; AVX2-FAST-NEXT:    retq
+; AVX2-LABEL: subus_v2i16:
+; AVX2:       # %bb.0:
+; AVX2-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX2-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX2-NEXT:    vpsubusw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vmovd %xmm0, (%rdi)
+; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: subus_v2i16:
 ; AVX512:       # %bb.0:
-; AVX512-NEXT:    vpmovzxwq {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX512-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX512-NEXT:    vpblendw {{.*#+}} xmm2 = xmm0[0],xmm1[1,2,3],xmm0[4],xmm1[5,6,7]
-; AVX512-NEXT:    vpmovzxwq {{.*#+}} xmm3 = mem[0],zero,zero,zero,mem[1],zero,zero,zero
-; AVX512-NEXT:    vpblendw {{.*#+}} xmm1 = xmm3[0],xmm1[1,2,3],xmm3[4],xmm1[5,6,7]
-; AVX512-NEXT:    vpcmpnleuq %xmm1, %xmm2, %k1
-; AVX512-NEXT:    vpsubq %xmm3, %xmm0, %xmm0 {%k1} {z}
+; AVX512-NEXT:    vmovd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; AVX512-NEXT:    vmovd {{.*#+}} xmm1 = mem[0],zero,zero,zero
+; AVX512-NEXT:    vpsubusw %xmm1, %xmm0, %xmm0
+; AVX512-NEXT:    vpmovzxwq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
 ; AVX512-NEXT:    vpmovqw %xmm0, (%rdi)
 ; AVX512-NEXT:    retq
   %ld1 = load <2 x i16>, <2 x i16>* %p1, align 8
Index: llvm/trunk/test/CodeGen/X86/sse2-intrinsics-canonical.ll
===================================================================
--- llvm/trunk/test/CodeGen/X86/sse2-intrinsics-canonical.ll
+++ llvm/trunk/test/CodeGen/X86/sse2-intrinsics-canonical.ll
@@ -93,43 +93,40 @@
 define <8 x i8> @test_x86_sse2_paddus_b_64(<8 x i8> %a0, <8 x i8> %a1) {
 ; SSE-LABEL: test_x86_sse2_paddus_b_64:
 ; SSE:       ## %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
+; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [255,255,255,255,255,255,255,255]
 ; SSE-NEXT:    ## encoding: [0x66,0x0f,0x6f,0x15,A,A,A,A]
 ; SSE-NEXT:    ## fixup A - offset: 4, value: LCPI4_0, kind: FK_Data_4
-; SSE-NEXT:    paddw %xmm0, %xmm1 ## encoding: [0x66,0x0f,0xfd,0xc8]
+; SSE-NEXT:    pand %xmm2, %xmm1 ## encoding: [0x66,0x0f,0xdb,0xca]
+; SSE-NEXT:    packuswb %xmm1, %xmm1 ## encoding: [0x66,0x0f,0x67,0xc9]
 ; SSE-NEXT:    pand %xmm2, %xmm0 ## encoding: [0x66,0x0f,0xdb,0xc2]
-; SSE-NEXT:    pand %xmm1, %xmm2 ## encoding: [0x66,0x0f,0xdb,0xd1]
-; SSE-NEXT:    pcmpgtw %xmm2, %xmm0 ## encoding: [0x66,0x0f,0x65,0xc2]
-; SSE-NEXT:    movdqa %xmm0, %xmm2 ## encoding: [0x66,0x0f,0x6f,0xd0]
-; SSE-NEXT:    pandn %xmm1, %xmm2 ## encoding: [0x66,0x0f,0xdf,0xd1]
-; SSE-NEXT:    pand LCPI4_0, %xmm0 ## encoding: [0x66,0x0f,0xdb,0x05,A,A,A,A]
-; SSE-NEXT:    ## fixup A - offset: 4, value: LCPI4_0, kind: FK_Data_4
-; SSE-NEXT:    por %xmm2, %xmm0 ## encoding: [0x66,0x0f,0xeb,0xc2]
+; SSE-NEXT:    packuswb %xmm0, %xmm0 ## encoding: [0x66,0x0f,0x67,0xc0]
+; SSE-NEXT:    paddusb %xmm1, %xmm0 ## encoding: [0x66,0x0f,0xdc,0xc1]
+; SSE-NEXT:    punpcklbw %xmm0, %xmm0 ## encoding: [0x66,0x0f,0x60,0xc0]
+; SSE-NEXT:    ## xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; SSE-NEXT:    retl ## encoding: [0xc3]
 ;
 ; AVX2-LABEL: test_x86_sse2_paddus_b_64:
 ; AVX2:       ## %bb.0:
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    ## encoding: [0xc5,0xf9,0x6f,0x15,A,A,A,A]
 ; AVX2-NEXT:    ## fixup A - offset: 4, value: LCPI4_0, kind: FK_Data_4
-; AVX2-NEXT:    vpand %xmm2, %xmm0, %xmm3 ## encoding: [0xc5,0xf9,0xdb,0xda]
-; AVX2-NEXT:    vpaddw %xmm1, %xmm0, %xmm0 ## encoding: [0xc5,0xf9,0xfd,0xc1]
-; AVX2-NEXT:    vpand %xmm2, %xmm0, %xmm1 ## encoding: [0xc5,0xf9,0xdb,0xca]
-; AVX2-NEXT:    vpcmpgtw %xmm1, %xmm3, %xmm1 ## encoding: [0xc5,0xe1,0x65,0xc9]
-; AVX2-NEXT:    vpblendvb %xmm1, %xmm2, %xmm0, %xmm0 ## encoding: [0xc4,0xe3,0x79,0x4c,0xc2,0x10]
+; AVX2-NEXT:    vpshufb %xmm2, %xmm1, %xmm1 ## encoding: [0xc4,0xe2,0x71,0x00,0xca]
+; AVX2-NEXT:    vpshufb %xmm2, %xmm0, %xmm0 ## encoding: [0xc4,0xe2,0x79,0x00,0xc2]
+; AVX2-NEXT:    vpaddusb %xmm1, %xmm0, %xmm0 ## encoding: [0xc5,0xf9,0xdc,0xc1]
+; AVX2-NEXT:    vpmovzxbw %xmm0, %xmm0 ## encoding: [0xc4,0xe2,0x79,0x30,0xc0]
+; AVX2-NEXT:    ## xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX2-NEXT:    retl ## encoding: [0xc3]
 ;
 ; SKX-LABEL: test_x86_sse2_paddus_b_64:
 ; SKX:       ## %bb.0:
-; SKX-NEXT:    vmovdqa LCPI4_0, %xmm2 ## EVEX TO VEX Compression xmm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
+; SKX-NEXT:    vmovdqa LCPI4_0, %xmm2 ## EVEX TO VEX Compression xmm2 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
 ; SKX-NEXT:    ## encoding: [0xc5,0xf9,0x6f,0x15,A,A,A,A]
 ; SKX-NEXT:    ## fixup A - offset: 4, value: LCPI4_0, kind: FK_Data_4
-; SKX-NEXT:    vpand %xmm2, %xmm0, %xmm3 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0xdb,0xda]
-; SKX-NEXT:    vpaddw %xmm1, %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0xfd,0xc1]
-; SKX-NEXT:    vpand %xmm2, %xmm0, %xmm1 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0xdb,0xca]
-; SKX-NEXT:    vpcmpnleuw %xmm1, %xmm3, %k1 ## encoding: [0x62,0xf3,0xe5,0x08,0x3e,0xc9,0x06]
-; SKX-NEXT:    vmovdqu16 LCPI4_0, %xmm0 {%k1} ## encoding: [0x62,0xf1,0xff,0x09,0x6f,0x05,A,A,A,A]
-; SKX-NEXT:    ## fixup A - offset: 6, value: LCPI4_0, kind: FK_Data_4
+; SKX-NEXT:    vpshufb %xmm2, %xmm1, %xmm1 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x71,0x00,0xca]
+; SKX-NEXT:    vpshufb %xmm2, %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x00,0xc2]
+; SKX-NEXT:    vpaddusb %xmm1, %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0xdc,0xc1]
+; SKX-NEXT:    vpmovzxbw %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x30,0xc0]
+; SKX-NEXT:    ## xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; SKX-NEXT:    retl ## encoding: [0xc3]
   %1 = add <8 x i8> %a0, %a1
   %2 = icmp ugt <8 x i8> %a0, %1
@@ -140,46 +137,45 @@
 define <4 x i16> @test_x86_sse2_paddus_w_64(<4 x i16> %a0, <4 x i16> %a1) {
 ; SSE-LABEL: test_x86_sse2_paddus_w_64:
 ; SSE:       ## %bb.0:
-; SSE-NEXT:    movdqa {{.*#+}} xmm2 = [65535,0,65535,0,65535,0,65535,0]
-; SSE-NEXT:    ## encoding: [0x66,0x0f,0x6f,0x15,A,A,A,A]
-; SSE-NEXT:    ## fixup A - offset: 4, value: LCPI5_0, kind: FK_Data_4
-; SSE-NEXT:    paddd %xmm0, %xmm1 ## encoding: [0x66,0x0f,0xfe,0xc8]
-; SSE-NEXT:    pand %xmm2, %xmm0 ## encoding: [0x66,0x0f,0xdb,0xc2]
-; SSE-NEXT:    pand %xmm1, %xmm2 ## encoding: [0x66,0x0f,0xdb,0xd1]
-; SSE-NEXT:    pcmpgtd %xmm2, %xmm0 ## encoding: [0x66,0x0f,0x66,0xc2]
-; SSE-NEXT:    movdqa %xmm0, %xmm2 ## encoding: [0x66,0x0f,0x6f,0xd0]
-; SSE-NEXT:    pandn %xmm1, %xmm2 ## encoding: [0x66,0x0f,0xdf,0xd1]
-; SSE-NEXT:    pand LCPI5_0, %xmm0 ## encoding: [0x66,0x0f,0xdb,0x05,A,A,A,A]
-; SSE-NEXT:    ## fixup A - offset: 4, value: LCPI5_0, kind: FK_Data_4
-; SSE-NEXT:    por %xmm2, %xmm0 ## encoding: [0x66,0x0f,0xeb,0xc2]
+; SSE-NEXT:    pshuflw $232, %xmm1, %xmm1 ## encoding: [0xf2,0x0f,0x70,0xc9,0xe8]
+; SSE-NEXT:    ## xmm1 = xmm1[0,2,2,3,4,5,6,7]
+; SSE-NEXT:    pshufhw $232, %xmm1, %xmm1 ## encoding: [0xf3,0x0f,0x70,0xc9,0xe8]
+; SSE-NEXT:    ## xmm1 = xmm1[0,1,2,3,4,6,6,7]
+; SSE-NEXT:    pshufd $232, %xmm1, %xmm1 ## encoding: [0x66,0x0f,0x70,0xc9,0xe8]
+; SSE-NEXT:    ## xmm1 = xmm1[0,2,2,3]
+; SSE-NEXT:    pshuflw $232, %xmm0, %xmm0 ## encoding: [0xf2,0x0f,0x70,0xc0,0xe8]
+; SSE-NEXT:    ## xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSE-NEXT:    pshufhw $232, %xmm0, %xmm0 ## encoding: [0xf3,0x0f,0x70,0xc0,0xe8]
+; SSE-NEXT:    ## xmm0 = xmm0[0,1,2,3,4,6,6,7]
+; SSE-NEXT:    pshufd $232, %xmm0, %xmm0 ## encoding: [0x66,0x0f,0x70,0xc0,0xe8]
+; SSE-NEXT:    ## xmm0 = xmm0[0,2,2,3]
+; SSE-NEXT:    paddusw %xmm1, %xmm0 ## encoding: [0x66,0x0f,0xdd,0xc1]
+; SSE-NEXT:    punpcklwd %xmm0, %xmm0 ## encoding: [0x66,0x0f,0x61,0xc0]
+; SSE-NEXT:    ## xmm0 = xmm0[0,0,1,1,2,2,3,3]
 ; SSE-NEXT:    retl ## encoding: [0xc3]
 ;
 ; AVX2-LABEL: test_x86_sse2_paddus_w_64:
 ; AVX2:       ## %bb.0:
-; AVX2-NEXT:    vpxor %xmm2, %xmm2, %xmm2 ## encoding: [0xc5,0xe9,0xef,0xd2]
-; AVX2-NEXT:    vpblendw $170, %xmm2, %xmm0, %xmm3 ## encoding: [0xc4,0xe3,0x79,0x0e,0xda,0xaa]
-; AVX2-NEXT:    ## xmm3 = xmm0[0],xmm2[1],xmm0[2],xmm2[3],xmm0[4],xmm2[5],xmm0[6],xmm2[7]
-; AVX2-NEXT:    vpaddd %xmm1, %xmm0, %xmm0 ## encoding: [0xc5,0xf9,0xfe,0xc1]
-; AVX2-NEXT:    vpblendw $170, %xmm2, %xmm0, %xmm1 ## encoding: [0xc4,0xe3,0x79,0x0e,0xca,0xaa]
-; AVX2-NEXT:    ## xmm1 = xmm0[0],xmm2[1],xmm0[2],xmm2[3],xmm0[4],xmm2[5],xmm0[6],xmm2[7]
-; AVX2-NEXT:    vpcmpgtd %xmm1, %xmm3, %xmm1 ## encoding: [0xc5,0xe1,0x66,0xc9]
-; AVX2-NEXT:    vbroadcastss {{.*#+}} xmm2 = [65535,65535,65535,65535]
-; AVX2-NEXT:    ## encoding: [0xc4,0xe2,0x79,0x18,0x15,A,A,A,A]
-; AVX2-NEXT:    ## fixup A - offset: 5, value: LCPI5_0, kind: FK_Data_4
-; AVX2-NEXT:    vblendvps %xmm1, %xmm2, %xmm0, %xmm0 ## encoding: [0xc4,0xe3,0x79,0x4a,0xc2,0x10]
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX2-NEXT:    ## encoding: [0xc5,0xf9,0x6f,0x15,A,A,A,A]
+; AVX2-NEXT:    ## fixup A - offset: 4, value: LCPI5_0, kind: FK_Data_4
+; AVX2-NEXT:    vpshufb %xmm2, %xmm1, %xmm1 ## encoding: [0xc4,0xe2,0x71,0x00,0xca]
+; AVX2-NEXT:    vpshufb %xmm2, %xmm0, %xmm0 ## encoding: [0xc4,0xe2,0x79,0x00,0xc2]
+; AVX2-NEXT:    vpaddusw %xmm1, %xmm0, %xmm0 ## encoding: [0xc5,0xf9,0xdd,0xc1]
+; AVX2-NEXT:    vpmovzxwd %xmm0, %xmm0 ## encoding: [0xc4,0xe2,0x79,0x33,0xc0]
+; AVX2-NEXT:    ## xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
 ; AVX2-NEXT:    retl ## encoding: [0xc3]
 ;
 ; SKX-LABEL: test_x86_sse2_paddus_w_64:
 ; SKX:       ## %bb.0:
-; SKX-NEXT:    vpxor %xmm2, %xmm2, %xmm2 ## EVEX TO VEX Compression encoding: [0xc5,0xe9,0xef,0xd2]
-; SKX-NEXT:    vpblendw $170, %xmm2, %xmm0, %xmm3 ## encoding: [0xc4,0xe3,0x79,0x0e,0xda,0xaa]
-; SKX-NEXT:    ## xmm3 = xmm0[0],xmm2[1],xmm0[2],xmm2[3],xmm0[4],xmm2[5],xmm0[6],xmm2[7]
-; SKX-NEXT:    vpaddd %xmm1, %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0xfe,0xc1]
-; SKX-NEXT:    vpblendw $170, %xmm2, %xmm0, %xmm1 ## encoding: [0xc4,0xe3,0x79,0x0e,0xca,0xaa]
-; SKX-NEXT:    ## xmm1 = xmm0[0],xmm2[1],xmm0[2],xmm2[3],xmm0[4],xmm2[5],xmm0[6],xmm2[7]
-; SKX-NEXT:    vpcmpnleud %xmm1, %xmm3, %k1 ## encoding: [0x62,0xf3,0x65,0x08,0x1e,0xc9,0x06]
-; SKX-NEXT:    vpbroadcastd LCPI5_0, %xmm0 {%k1} ## encoding: [0x62,0xf2,0x7d,0x09,0x58,0x05,A,A,A,A]
-; SKX-NEXT:    ## fixup A - offset: 6, value: LCPI5_0, kind: FK_Data_4
+; SKX-NEXT:    vmovdqa LCPI5_0, %xmm2 ## EVEX TO VEX Compression xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; SKX-NEXT:    ## encoding: [0xc5,0xf9,0x6f,0x15,A,A,A,A]
+; SKX-NEXT:    ## fixup A - offset: 4, value: LCPI5_0, kind: FK_Data_4
+; SKX-NEXT:    vpshufb %xmm2, %xmm1, %xmm1 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x71,0x00,0xca]
+; SKX-NEXT:    vpshufb %xmm2, %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x00,0xc2]
+; SKX-NEXT:    vpaddusw %xmm1, %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0xdd,0xc1]
+; SKX-NEXT:    vpmovzxwd %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x33,0xc0]
+; SKX-NEXT:    ## xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
 ; SKX-NEXT:    retl ## encoding: [0xc3]
   %1 = add <4 x i16> %a0, %a1
   %2 = icmp ugt <4 x i16> %a0, %1