Index: lib/Target/X86/X86ISelLowering.cpp
===================================================================
--- lib/Target/X86/X86ISelLowering.cpp
+++ lib/Target/X86/X86ISelLowering.cpp
@@ -30121,6 +30121,39 @@
       Other = LHS;
     }
 
+    // This optimization requires minimum target sse4.1, because it requires
+    // umin instruction for v*i16 or v*i32.
+    if (Other.getNode() && Other.getOpcode() == ISD::TRUNCATE &&
+        Other.getOperand(0).getOpcode() == ISD::SUB && isUnsignedIntSetCC(CC) &&
+        Subtarget.hasSSE41()) {
+      //  x >= y ? x - y : 0, where y is wider then x. We can truncate y
+      //  to x, because it is only used if it is less than x.
+      SDValue SubXY = Other.getOperand(0);
+      EVT ExtType = SubXY.getValueType();
+      SDValue SubLHS = SubXY.getOperand(0), SubRHS = SubXY.getOperand(1);
+      SDValue CondLHS = Cond.getOperand(0), CondRHS = Cond.getOperand(1);
+      ISD::CondCode NewCC = CC;
+      // Else can occur, if the sub arguments are swapped
+      //  x < y ? y - x : 0 -> y > x ? y - x : 0
+      if (CC == ISD::SETULE || CC == ISD::SETULT) {
+        std::swap(CondLHS, CondRHS);
+        NewCC = CC == ISD::SETULE ? ISD::SETUGE : ISD::SETUGT;
+      }
+      if (CondLHS.getOpcode() == ISD::ZERO_EXTEND &&
+          DAG.isEqualTo(SubRHS, CondRHS) && DAG.isEqualTo(SubLHS, CondLHS)) {
+        SDValue SaturationConst = DAG.getConstant(
+            APInt::getMaxValue(VT.getScalarSizeInBits()).getLimitedValue(),
+            SDLoc(SubRHS), ExtType);
+        SDValue UMin = DAG.getNode(ISD::UMIN, SDLoc(SubRHS), ExtType, SubRHS,
+                                   SaturationConst);
+        Other = DAG.getNode(ISD::SUB, SDLoc(SubXY), VT, CondLHS.getOperand(0),
+                            DAG.getZExtOrTrunc(UMin, SDLoc(SubRHS), VT));
+        Cond = DAG.getNode(NewCC, SDLoc(Cond), VT, Other.getOperand(0),
+                           Other.getOperand(1));
+        CC = NewCC;
+      }
+    }
+
     if (Other.getNode() && Other->getNumOperands() == 2 &&
         DAG.isEqualTo(Other->getOperand(0), Cond.getOperand(0))) {
       SDValue OpLHS = Other->getOperand(0), OpRHS = Other->getOperand(1);
Index: test/CodeGen/X86/psubus.ll
===================================================================
--- test/CodeGen/X86/psubus.ll
+++ test/CodeGen/X86/psubus.ll
@@ -604,68 +604,37 @@
 ;
 ; SSE41-LABEL: test13:
 ; SSE41:       ## BB#0: ## %vector.ph
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
-; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm4 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero
-; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm3 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; SSE41-NEXT:    movdqa {{.*#+}} xmm5 = [2147483648,2147483648,2147483648,2147483648]
-; SSE41-NEXT:    movdqa %xmm3, %xmm6
-; SSE41-NEXT:    psubd %xmm1, %xmm3
-; SSE41-NEXT:    movdqa %xmm1, %xmm0
-; SSE41-NEXT:    pxor %xmm5, %xmm0
-; SSE41-NEXT:    pxor %xmm5, %xmm6
-; SSE41-NEXT:    pcmpgtd %xmm6, %xmm0
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; SSE41-NEXT:    pshufb %xmm1, %xmm0
-; SSE41-NEXT:    movdqa %xmm2, %xmm6
-; SSE41-NEXT:    pxor %xmm5, %xmm6
-; SSE41-NEXT:    pxor %xmm4, %xmm5
-; SSE41-NEXT:    pcmpgtd %xmm5, %xmm6
-; SSE41-NEXT:    pshufb %xmm1, %xmm6
-; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm6[0]
-; SSE41-NEXT:    psubd %xmm2, %xmm4
-; SSE41-NEXT:    pshufb %xmm1, %xmm3
-; SSE41-NEXT:    pshufb %xmm1, %xmm4
-; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm4[0]
-; SSE41-NEXT:    pandn %xmm3, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [65535,65535,65535,65535]
+; SSE41-NEXT:    pminud %xmm3, %xmm2
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; SSE41-NEXT:    pshufb %xmm4, %xmm2
+; SSE41-NEXT:    pminud %xmm3, %xmm1
+; SSE41-NEXT:    pshufb %xmm4, %xmm1
+; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; SSE41-NEXT:    psubusw %xmm1, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: test13:
 ; AVX1:       ## BB#0: ## %vector.ph
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648]
-; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm4
-; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm5
-; AVX1-NEXT:    vpcmpgtd %xmm4, %xmm5, %xmm4
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm5
-; AVX1-NEXT:    vpxor %xmm3, %xmm5, %xmm6
-; AVX1-NEXT:    vpxor %xmm3, %xmm2, %xmm3
-; AVX1-NEXT:    vpcmpgtd %xmm3, %xmm6, %xmm3
-; AVX1-NEXT:    vpacksswb %xmm3, %xmm4, %xmm3
-; AVX1-NEXT:    vpsubd %xmm5, %xmm2, %xmm2
-; AVX1-NEXT:    vpsubd %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb %xmm1, %xmm2, %xmm1
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; AVX1-NEXT:    vpandn %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [65535,65535,65535,65535]
+; AVX1-NEXT:    vpminud %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX1-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpminud %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpshufb %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; AVX1-NEXT:    vpsubusw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: test13:
 ; AVX2:       ## BB#0: ## %vector.ph
-; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX2-NEXT:    vpbroadcastd {{.*}}(%rip), %ymm2
-; AVX2-NEXT:    vpxor %ymm2, %ymm1, %ymm3
-; AVX2-NEXT:    vpxor %ymm2, %ymm0, %ymm2
-; AVX2-NEXT:    vpcmpgtd %ymm2, %ymm3, %ymm2
-; AVX2-NEXT:    vextracti128 $1, %ymm2, %xmm3
-; AVX2-NEXT:    vpacksswb %xmm3, %xmm2, %xmm2
-; AVX2-NEXT:    vpsubd %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpandn %xmm0, %xmm2, %xmm0
+; AVX2-NEXT:    vpminud %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vpsubusw %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 vector.ph:
@@ -987,67 +956,37 @@
 ;
 ; SSE41-LABEL: test15:
 ; SSE41:       ## BB#0: ## %vector.ph
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
-; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero
-; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [2147483648,2147483648,2147483648,2147483648]
-; SSE41-NEXT:    movdqa %xmm0, %xmm5
-; SSE41-NEXT:    psubd %xmm1, %xmm0
-; SSE41-NEXT:    pxor %xmm4, %xmm1
-; SSE41-NEXT:    pxor %xmm4, %xmm5
-; SSE41-NEXT:    pcmpgtd %xmm1, %xmm5
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; SSE41-NEXT:    pshufb %xmm1, %xmm5
-; SSE41-NEXT:    movdqa %xmm2, %xmm6
-; SSE41-NEXT:    pxor %xmm4, %xmm6
-; SSE41-NEXT:    pxor %xmm3, %xmm4
-; SSE41-NEXT:    pcmpgtd %xmm6, %xmm4
-; SSE41-NEXT:    pshufb %xmm1, %xmm4
-; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm5 = xmm5[0],xmm4[0]
-; SSE41-NEXT:    psubd %xmm2, %xmm3
-; SSE41-NEXT:    pshufb %xmm1, %xmm0
-; SSE41-NEXT:    pshufb %xmm1, %xmm3
-; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm3[0]
-; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [65535,65535,65535,65535]
+; SSE41-NEXT:    pminud %xmm3, %xmm2
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; SSE41-NEXT:    pshufb %xmm4, %xmm2
+; SSE41-NEXT:    pminud %xmm3, %xmm1
+; SSE41-NEXT:    pshufb %xmm4, %xmm1
+; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; SSE41-NEXT:    psubusw %xmm1, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: test15:
 ; AVX1:       ## BB#0: ## %vector.ph
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648]
-; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm4
-; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm5
-; AVX1-NEXT:    vpcmpgtd %xmm4, %xmm5, %xmm4
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm5
-; AVX1-NEXT:    vpxor %xmm3, %xmm5, %xmm6
-; AVX1-NEXT:    vpxor %xmm3, %xmm2, %xmm3
-; AVX1-NEXT:    vpcmpgtd %xmm6, %xmm3, %xmm3
-; AVX1-NEXT:    vpacksswb %xmm3, %xmm4, %xmm3
-; AVX1-NEXT:    vpsubd %xmm5, %xmm2, %xmm2
-; AVX1-NEXT:    vpsubd %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb %xmm1, %xmm2, %xmm1
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; AVX1-NEXT:    vpand %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [65535,65535,65535,65535]
+; AVX1-NEXT:    vpminud %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX1-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpminud %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpshufb %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; AVX1-NEXT:    vpsubusw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: test15:
 ; AVX2:       ## BB#0: ## %vector.ph
-; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX2-NEXT:    vpbroadcastd {{.*}}(%rip), %ymm2
-; AVX2-NEXT:    vpxor %ymm2, %ymm1, %ymm3
-; AVX2-NEXT:    vpxor %ymm2, %ymm0, %ymm2
-; AVX2-NEXT:    vpcmpgtd %ymm3, %ymm2, %ymm2
-; AVX2-NEXT:    vextracti128 $1, %ymm2, %xmm3
-; AVX2-NEXT:    vpacksswb %xmm3, %xmm2, %xmm2
-; AVX2-NEXT:    vpsubd %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpand %xmm0, %xmm2, %xmm0
+; AVX2-NEXT:    vpminud %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vpsubusw %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 vector.ph:
@@ -1123,67 +1062,37 @@
 ;
 ; SSE41-LABEL: test16:
 ; SSE41:       ## BB#0: ## %vector.ph
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
-; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm3 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero
-; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [2147483648,2147483648,2147483648,2147483648]
-; SSE41-NEXT:    movdqa %xmm0, %xmm5
-; SSE41-NEXT:    psubd %xmm1, %xmm0
-; SSE41-NEXT:    pxor %xmm4, %xmm1
-; SSE41-NEXT:    pxor %xmm4, %xmm5
-; SSE41-NEXT:    pcmpgtd %xmm1, %xmm5
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; SSE41-NEXT:    pshufb %xmm1, %xmm5
-; SSE41-NEXT:    movdqa %xmm2, %xmm6
-; SSE41-NEXT:    pxor %xmm4, %xmm6
-; SSE41-NEXT:    pxor %xmm3, %xmm4
-; SSE41-NEXT:    pcmpgtd %xmm6, %xmm4
-; SSE41-NEXT:    pshufb %xmm1, %xmm4
-; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm5 = xmm5[0],xmm4[0]
-; SSE41-NEXT:    psubd %xmm2, %xmm3
-; SSE41-NEXT:    pshufb %xmm1, %xmm0
-; SSE41-NEXT:    pshufb %xmm1, %xmm3
-; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm3[0]
-; SSE41-NEXT:    pand %xmm5, %xmm0
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [65535,65535,65535,65535]
+; SSE41-NEXT:    pminud %xmm3, %xmm2
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; SSE41-NEXT:    pshufb %xmm4, %xmm2
+; SSE41-NEXT:    pminud %xmm3, %xmm1
+; SSE41-NEXT:    pshufb %xmm4, %xmm1
+; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; SSE41-NEXT:    psubusw %xmm1, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: test16:
 ; AVX1:       ## BB#0: ## %vector.ph
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm2 = xmm2[0],zero,xmm2[1],zero,xmm2[2],zero,xmm2[3],zero
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648]
-; AVX1-NEXT:    vpxor %xmm3, %xmm1, %xmm4
-; AVX1-NEXT:    vpxor %xmm3, %xmm0, %xmm5
-; AVX1-NEXT:    vpcmpgtd %xmm4, %xmm5, %xmm4
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm5
-; AVX1-NEXT:    vpxor %xmm3, %xmm5, %xmm6
-; AVX1-NEXT:    vpxor %xmm3, %xmm2, %xmm3
-; AVX1-NEXT:    vpcmpgtd %xmm6, %xmm3, %xmm3
-; AVX1-NEXT:    vpacksswb %xmm3, %xmm4, %xmm3
-; AVX1-NEXT:    vpsubd %xmm5, %xmm2, %xmm2
-; AVX1-NEXT:    vpsubd %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm1 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpshufb %xmm1, %xmm2, %xmm1
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; AVX1-NEXT:    vpand %xmm0, %xmm3, %xmm0
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [65535,65535,65535,65535]
+; AVX1-NEXT:    vpminud %xmm3, %xmm2, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX1-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
+; AVX1-NEXT:    vpminud %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpshufb %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; AVX1-NEXT:    vpsubusw %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: test16:
 ; AVX2:       ## BB#0: ## %vector.ph
-; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX2-NEXT:    vpbroadcastd {{.*}}(%rip), %ymm2
-; AVX2-NEXT:    vpxor %ymm2, %ymm1, %ymm3
-; AVX2-NEXT:    vpxor %ymm2, %ymm0, %ymm2
-; AVX2-NEXT:    vpcmpgtd %ymm3, %ymm2, %ymm2
-; AVX2-NEXT:    vextracti128 $1, %ymm2, %xmm3
-; AVX2-NEXT:    vpacksswb %xmm3, %xmm2, %xmm2
-; AVX2-NEXT:    vpsubd %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vpand %xmm0, %xmm2, %xmm0
+; AVX2-NEXT:    vpminud %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15,16,17,20,21,24,25,28,29,24,25,28,29,28,29,30,31]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
+; AVX2-NEXT:    vpsubusw %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 vector.ph: