[X86] Teach combineTruncatedArithmetic to push truncate through subtracts where only one of the inputs is free to truncate.

[craig.topper]

Summary

Fix combineSubToSubus to handle the new DAG to avoid a regression.

There are still regressions in test14/test15/test16. Where it
looks like were trying to set up cases we could match to
umin+trunc+subus but the handling was never finished. The
regression here isn't unique to sub. Its a lost opportunity for
taking an AND with two truncated inputs and producing a larger
AND with a single truncate. The same thing could happen with
any other node we handle in combineTruncatedArithmetic since we
are moving the truncate up the DAG.

[spatel]

LGTM - mark the code and/or tests with TODO comments, so we know where the potential improvements are?

[RKSimon]

Going forward I'm still hoping that we can move most of this to DAGCombine

llvm/lib/Target/X86/X86ISelLowering.cpp

case ISD::MUL:
     if (SrcVT.getScalarType() == MVT::i64 &&
         TLI.isOperationLegal(SrcOpcode, VT) &&
         !TLI.isOperationLegal(SrcOpcode, SrcVT))
       return TruncateArithmetic(Src.getOperand(0), Src.getOperand(1));
     LLVM_FALLTHROUGH;
   case ISD::AND:
   case ISD::XOR:
   case ISD::OR:
-  case ISD::ADD: {
-    SDValue Op0 = Src.getOperand(0);
-    SDValue Op1 = Src.getOperand(1);
-    if (TLI.isOperationLegal(SrcOpcode, VT) &&
-        (Op0 == Op1 || IsFreeTruncation(Op0) || IsFreeTruncation(Op1)))
-      return TruncateArithmetic(Op0, Op1);
-    break;
-  }
+  case ISD::ADD:
   case ISD::SUB: {
-    // TODO: ISD::SUB We are conservative and require both sides to be freely
-    // truncatable to avoid interfering with combineSubToSubus.
     SDValue Op0 = Src.getOperand(0);
     SDValue Op1 = Src.getOperand(1);
     if (TLI.isOperationLegal(SrcOpcode, VT) &&
-        (Op0 == Op1 || (IsFreeTruncation(Op0) && IsFreeTruncation(Op1))))
+        (Op0 == Op1 || IsFreeTruncation(Op0) || IsFreeTruncation(Op1)))
       return TruncateArithmetic(Op0, Op1);
     break;
   }
   }
 
   return SDValue();
 }
 
if (Op0.getOpcode() == ISD::UMAX) {
     SDValue MinLHS = Op1.getOperand(0);
     SDValue MinRHS = Op1.getOperand(1);
     if (MinLHS == Op0)
       SubusRHS = MinRHS;
     else if (MinRHS == Op0)
       SubusRHS = MinLHS;
     else
       return SDValue();
+  } else if (Op1.getOpcode() == ISD::TRUNCATE &&
+             Op1.getOperand(0).getOpcode() == ISD::UMIN &&
+             (EltVT == MVT::i8 || EltVT == MVT::i16)) {
+    // Special case where the UMIN has been truncated. Try to push the truncate
+    // further up. This is similar to the i32/i64 special processing.
+    SubusLHS = Op0;
+    SDValue MinLHS = Op1.getOperand(0).getOperand(0);
+    SDValue MinRHS = Op1.getOperand(0).getOperand(1);
+    EVT TruncVT = Op1.getOperand(0).getValueType();
+    if (!(Subtarget.hasSSSE3() && (TruncVT == MVT::v8i32 ||
+                                   TruncVT == MVT::v8i64)) &&
+        !(Subtarget.useBWIRegs() && (TruncVT == MVT::v16i32)))
+      return SDValue();
+    SDValue OpToSaturate;
+    if (MinLHS.getOpcode() == ISD::ZERO_EXTEND &&
+        MinLHS.getOperand(0) == Op0)
+      OpToSaturate = MinRHS;
+    else if (MinRHS.getOpcode() == ISD::ZERO_EXTEND &&
+             MinRHS.getOperand(0) == Op0)
+      OpToSaturate = MinLHS;
+    else
+      return SDValue();
+
+    // Saturate the non-extended input and then truncate it.
+    SDLoc DL(N);
+    SDValue SaturationConst =
+        DAG.getConstant(APInt::getLowBitsSet(TruncVT.getScalarSizeInBits(),
+                                             VT.getScalarSizeInBits()),
+                        DL, TruncVT);
+    SDValue UMin = DAG.getNode(ISD::UMIN, DL, TruncVT, OpToSaturate,
+                               SaturationConst);
+    SubusRHS = DAG.getNode(ISD::TRUNCATE, DL, VT, UMin);
   } else
     return SDValue();
 
   // PSUBUS doesn't support v8i32/v8i64/v16i32, but it can be enabled with
   // special preprocessing in some cases.
   if (EltVT == MVT::i8 || EltVT == MVT::i16)
     return DAG.getNode(ISD::USUBSAT, SDLoc(N), VT, SubusLHS, SubusRHS);
 

llvm/test/CodeGen/X86/psubus.ll

vector.ph:
   %0 = insertelement <32 x i8> undef, i8 %w, i32 0
   %broadcast15 = shufflevector <32 x i8> %0, <32 x i8> undef, <32 x i32> zeroinitializer
   %1 = icmp ult <32 x i8> %x, %broadcast15
   %2 = sub <32 x i8> %x, %broadcast15
   %res = select <32 x i1> %1, <32 x i8> zeroinitializer, <32 x i8> %2
   ret <32 x i8> %res
 }
 
+; FIXME: match this to UMIN+TRUNC+PSUBUS
 define <8 x i16> @test13(<8 x i16> %x, <8 x i32> %y) nounwind {
 ; SSE2-LABEL: test13:
 ; SSE2:       # %bb.0: # %vector.ph
-; SSE2-NEXT:    pxor %xmm4, %xmm4
-; SSE2-NEXT:    movdqa %xmm0, %xmm3
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
-; SSE2-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [2147483648,2147483648,2147483648,2147483648]
+; SSE2-NEXT:    pxor %xmm3, %xmm3
+; SSE2-NEXT:    movdqa %xmm0, %xmm4
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1],xmm4[2],xmm3[2],xmm4[3],xmm3[3]
 ; SSE2-NEXT:    movdqa %xmm0, %xmm5
-; SSE2-NEXT:    psubd %xmm2, %xmm0
-; SSE2-NEXT:    movdqa %xmm2, %xmm6
-; SSE2-NEXT:    pxor %xmm4, %xmm6
-; SSE2-NEXT:    por %xmm4, %xmm5
-; SSE2-NEXT:    pcmpgtd %xmm5, %xmm6
-; SSE2-NEXT:    movdqa %xmm1, %xmm2
-; SSE2-NEXT:    pxor %xmm4, %xmm2
-; SSE2-NEXT:    por %xmm3, %xmm4
-; SSE2-NEXT:    pcmpgtd %xmm4, %xmm2
-; SSE2-NEXT:    packssdw %xmm6, %xmm2
-; SSE2-NEXT:    psubd %xmm1, %xmm3
-; SSE2-NEXT:    pslld $16, %xmm0
-; SSE2-NEXT:    psrad $16, %xmm0
-; SSE2-NEXT:    pslld $16, %xmm3
-; SSE2-NEXT:    psrad $16, %xmm3
-; SSE2-NEXT:    packssdw %xmm0, %xmm3
-; SSE2-NEXT:    pandn %xmm3, %xmm2
-; SSE2-NEXT:    movdqa %xmm2, %xmm0
+; SSE2-NEXT:    punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm3[4],xmm5[5],xmm3[5],xmm5[6],xmm3[6],xmm5[7],xmm3[7]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm6 = [2147483648,2147483648,2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm2, %xmm7
+; SSE2-NEXT:    pxor %xmm6, %xmm7
+; SSE2-NEXT:    por %xmm6, %xmm5
+; SSE2-NEXT:    pcmpgtd %xmm5, %xmm7
+; SSE2-NEXT:    movdqa %xmm1, %xmm3
+; SSE2-NEXT:    pxor %xmm6, %xmm3
+; SSE2-NEXT:    por %xmm6, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm4, %xmm3
+; SSE2-NEXT:    packssdw %xmm7, %xmm3
+; SSE2-NEXT:    pslld $16, %xmm2
+; SSE2-NEXT:    psrad $16, %xmm2
+; SSE2-NEXT:    pslld $16, %xmm1
+; SSE2-NEXT:    psrad $16, %xmm1
+; SSE2-NEXT:    packssdw %xmm2, %xmm1
+; SSE2-NEXT:    psubw %xmm1, %xmm0
+; SSE2-NEXT:    pandn %xmm0, %xmm3
+; SSE2-NEXT:    movdqa %xmm3, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: test13:
 ; SSSE3:       # %bb.0: # %vector.ph
 ; SSSE3-NEXT:    pxor %xmm3, %xmm3
 ; SSSE3-NEXT:    movdqa %xmm0, %xmm4
 ; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1],xmm4[2],xmm3[2],xmm4[3],xmm3[3]
-; SSSE3-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648]
 ; SSSE3-NEXT:    movdqa %xmm0, %xmm5
-; SSSE3-NEXT:    psubd %xmm2, %xmm0
-; SSSE3-NEXT:    movdqa %xmm2, %xmm6
-; SSSE3-NEXT:    pxor %xmm3, %xmm6
-; SSSE3-NEXT:    por %xmm3, %xmm5
-; SSSE3-NEXT:    pcmpgtd %xmm5, %xmm6
-; SSSE3-NEXT:    movdqa %xmm1, %xmm2
-; SSSE3-NEXT:    pxor %xmm3, %xmm2
-; SSSE3-NEXT:    por %xmm4, %xmm3
-; SSSE3-NEXT:    pcmpgtd %xmm3, %xmm2
-; SSSE3-NEXT:    packssdw %xmm6, %xmm2
-; SSSE3-NEXT:    psubd %xmm1, %xmm4
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; SSSE3-NEXT:    pshufb %xmm1, %xmm0
-; SSSE3-NEXT:    pshufb %xmm1, %xmm4
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm4 = xmm4[0],xmm0[0]
-; SSSE3-NEXT:    pandn %xmm4, %xmm2
-; SSSE3-NEXT:    movdqa %xmm2, %xmm0
+; SSSE3-NEXT:    punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm3[4],xmm5[5],xmm3[5],xmm5[6],xmm3[6],xmm5[7],xmm3[7]
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm6 = [2147483648,2147483648,2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm2, %xmm7
+; SSSE3-NEXT:    pxor %xmm6, %xmm7
+; SSSE3-NEXT:    por %xmm6, %xmm5
+; SSSE3-NEXT:    pcmpgtd %xmm5, %xmm7
+; SSSE3-NEXT:    movdqa %xmm1, %xmm3
+; SSSE3-NEXT:    pxor %xmm6, %xmm3
+; SSSE3-NEXT:    por %xmm6, %xmm4
+; SSSE3-NEXT:    pcmpgtd %xmm4, %xmm3
+; SSSE3-NEXT:    packssdw %xmm7, %xmm3
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm4 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; SSSE3-NEXT:    pshufb %xmm4, %xmm2
+; SSSE3-NEXT:    pshufb %xmm4, %xmm1
+; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; SSSE3-NEXT:    psubw %xmm1, %xmm0
+; SSSE3-NEXT:    pandn %xmm0, %xmm3
+; SSSE3-NEXT:    movdqa %xmm3, %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: test13:
 ; 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 {{.*#+}} xmm4 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; SSE41-NEXT:    movdqa %xmm4, %xmm0
-; SSE41-NEXT:    pmaxud %xmm1, %xmm0
-; SSE41-NEXT:    pcmpeqd %xmm4, %xmm0
+; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm4 = xmm3[0],zero,xmm3[1],zero,xmm3[2],zero,xmm3[3],zero
+; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm5 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
+; SSE41-NEXT:    movdqa %xmm5, %xmm3
+; SSE41-NEXT:    pmaxud %xmm1, %xmm3
+; SSE41-NEXT:    pcmpeqd %xmm5, %xmm3
 ; SSE41-NEXT:    pcmpeqd %xmm5, %xmm5
-; SSE41-NEXT:    pxor %xmm5, %xmm0
-; SSE41-NEXT:    movdqa %xmm3, %xmm6
+; SSE41-NEXT:    pxor %xmm5, %xmm3
+; SSE41-NEXT:    movdqa %xmm4, %xmm6
 ; SSE41-NEXT:    pmaxud %xmm2, %xmm6
-; SSE41-NEXT:    pcmpeqd %xmm3, %xmm6
+; SSE41-NEXT:    pcmpeqd %xmm4, %xmm6
 ; SSE41-NEXT:    pxor %xmm5, %xmm6
-; SSE41-NEXT:    packssdw %xmm6, %xmm0
-; SSE41-NEXT:    psubd %xmm2, %xmm3
-; SSE41-NEXT:    psubd %xmm1, %xmm4
-; SSE41-NEXT:    pxor %xmm1, %xmm1
-; SSE41-NEXT:    pblendw {{.*#+}} xmm4 = xmm4[0],xmm1[1],xmm4[2],xmm1[3],xmm4[4],xmm1[5],xmm4[6],xmm1[7]
-; SSE41-NEXT:    pblendw {{.*#+}} xmm3 = xmm3[0],xmm1[1],xmm3[2],xmm1[3],xmm3[4],xmm1[5],xmm3[6],xmm1[7]
-; SSE41-NEXT:    packusdw %xmm3, %xmm4
-; SSE41-NEXT:    pandn %xmm4, %xmm0
+; SSE41-NEXT:    packssdw %xmm6, %xmm3
+; SSE41-NEXT:    pxor %xmm4, %xmm4
+; SSE41-NEXT:    pblendw {{.*#+}} xmm2 = xmm2[0],xmm4[1],xmm2[2],xmm4[3],xmm2[4],xmm4[5],xmm2[6],xmm4[7]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0],xmm4[1],xmm1[2],xmm4[3],xmm1[4],xmm4[5],xmm1[6],xmm4[7]
+; SSE41-NEXT:    packusdw %xmm2, %xmm1
+; SSE41-NEXT:    psubw %xmm1, %xmm0
+; SSE41-NEXT:    pandn %xmm0, %xmm3
+; SSE41-NEXT:    movdqa %xmm3, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: test13:
 ; AVX1:       # %bb.0: # %vector.ph
 ; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm2 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm3
-; AVX1-NEXT:    vpmaxud %xmm3, %xmm2, %xmm4
-; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm2, %xmm4
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm3 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm4
+; AVX1-NEXT:    vpmaxud %xmm4, %xmm2, %xmm5
+; AVX1-NEXT:    vpcmpeqd %xmm5, %xmm2, %xmm2
 ; AVX1-NEXT:    vpcmpeqd %xmm5, %xmm5, %xmm5
-; AVX1-NEXT:    vpxor %xmm5, %xmm4, %xmm4
-; AVX1-NEXT:    vpmaxud %xmm1, %xmm0, %xmm6
-; AVX1-NEXT:    vpcmpeqd %xmm6, %xmm0, %xmm6
-; AVX1-NEXT:    vpxor %xmm5, %xmm6, %xmm5
-; AVX1-NEXT:    vpackssdw %xmm4, %xmm5, %xmm4
-; AVX1-NEXT:    vpsubd %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpsubd %xmm3, %xmm2, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; AVX1-NEXT:    vpandn %xmm0, %xmm4, %xmm0
+; AVX1-NEXT:    vpxor %xmm5, %xmm2, %xmm2
+; AVX1-NEXT:    vpmaxud %xmm1, %xmm3, %xmm6
+; AVX1-NEXT:    vpcmpeqd %xmm6, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm5, %xmm3, %xmm3
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX1-NEXT:    vpshufb %xmm3, %xmm4, %xmm4
+; AVX1-NEXT:    vpshufb %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm4[0]
+; AVX1-NEXT:    vpsubw %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpandn %xmm0, %xmm2, %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:    vpmaxud %ymm1, %ymm0, %ymm2
-; AVX2-NEXT:    vpcmpeqd %ymm2, %ymm0, %ymm2
+; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm2 = 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:    vpmaxud %ymm1, %ymm2, %ymm3
+; AVX2-NEXT:    vpcmpeqd %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vpcmpeqd %ymm3, %ymm3, %ymm3
 ; AVX2-NEXT:    vpxor %ymm3, %ymm2, %ymm2
 ; AVX2-NEXT:    vextracti128 $1, %ymm2, %xmm3
 ; AVX2-NEXT:    vpackssdw %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:    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:    vpsubw %xmm1, %xmm0, %xmm0
 ; AVX2-NEXT:    vpandn %xmm0, %xmm2, %xmm0
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test13:
 ; AVX512:       # %bb.0: # %vector.ph
-; AVX512-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
-; AVX512-NEXT:    vpcmpnltud %ymm1, %ymm0, %k1
-; AVX512-NEXT:    vpsubd %ymm1, %ymm0, %ymm0
-; AVX512-NEXT:    vpmovdw %ymm0, %xmm0 {%k1} {z}
+; AVX512-NEXT:    vpmovzxwd {{.*#+}} ymm2 = 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:    vpcmpnltud %ymm1, %ymm2, %k1
+; AVX512-NEXT:    vpmovdw %ymm1, %xmm1
+; AVX512-NEXT:    vpsubw %xmm1, %xmm0, %xmm0 {%k1} {z}
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
 vector.ph:
   %lhs = zext <8 x i16> %x to <8 x i32>
   %cond = icmp ult <8 x i32> %lhs, %y
   %sub = sub <8 x i32> %lhs, %y
   %trunc = trunc <8 x i32> %sub to <8 x i16>
   %res = select <8 x i1> %cond, <8 x i16> zeroinitializer, <8 x i16> %trunc
   ret <8 x i16> %res
 }
 
+; FIXME: match this to UMIN+TRUNC+PSUBUS
 define <16 x i8> @test14(<16 x i8> %x, <16 x i32> %y) nounwind {
 ; SSE2-LABEL: test14:
 ; SSE2:       # %bb.0: # %vector.ph
-; SSE2-NEXT:    movdqa %xmm0, %xmm5
-; SSE2-NEXT:    pxor %xmm0, %xmm0
-; SSE2-NEXT:    movdqa %xmm5, %xmm6
-; SSE2-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]
-; SSE2-NEXT:    movdqa %xmm6, %xmm8
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm8 = xmm8[0],xmm0[0],xmm8[1],xmm0[1],xmm8[2],xmm0[2],xmm8[3],xmm0[3]
-; SSE2-NEXT:    punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm0[4],xmm6[5],xmm0[5],xmm6[6],xmm0[6],xmm6[7],xmm0[7]
-; SSE2-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]
-; SSE2-NEXT:    movdqa %xmm5, %xmm10
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm10 = xmm10[0],xmm0[0],xmm10[1],xmm0[1],xmm10[2],xmm0[2],xmm10[3],xmm0[3]
-; SSE2-NEXT:    punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm0[4],xmm5[5],xmm0[5],xmm5[6],xmm0[6],xmm5[7],xmm0[7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm0 = [2147483648,2147483648,2147483648,2147483648]
+; SSE2-NEXT:    pxor %xmm8, %xmm8
+; SSE2-NEXT:    movdqa %xmm0, %xmm6
 ; SSE2-NEXT:    movdqa %xmm4, %xmm9
-; SSE2-NEXT:    pxor %xmm0, %xmm9
-; SSE2-NEXT:    psubd %xmm5, %xmm4
-; SSE2-NEXT:    por %xmm0, %xmm5
-; SSE2-NEXT:    pcmpgtd %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:    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:    movdqa %xmm1, %xmm5
-; SSE2-NEXT:    pxor %xmm0, %xmm5
-; SSE2-NEXT:    por %xmm8, %xmm0
-; SSE2-NEXT:    pcmpgtd %xmm5, %xmm0
-; SSE2-NEXT:    packssdw %xmm6, %xmm0
-; SSE2-NEXT:    packsswb %xmm10, %xmm0
-; SSE2-NEXT:    psubd %xmm8, %xmm1
+; SSE2-NEXT:    movdqa %xmm3, %xmm10
+; SSE2-NEXT:    movdqa %xmm2, %xmm7
 ; 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:    movdqa %xmm1, %xmm4
 ; SSE2-NEXT:    pand %xmm5, %xmm2
 ; SSE2-NEXT:    pand %xmm5, %xmm1
 ; SSE2-NEXT:    packuswb %xmm2, %xmm1
 ; SSE2-NEXT:    packuswb %xmm3, %xmm1
+; SSE2-NEXT:    psubb %xmm0, %xmm1
+; SSE2-NEXT:    movdqa %xmm0, %xmm2
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm8[0],xmm2[1],xmm8[1],xmm2[2],xmm8[2],xmm2[3],xmm8[3],xmm2[4],xmm8[4],xmm2[5],xmm8[5],xmm2[6],xmm8[6],xmm2[7],xmm8[7]
+; SSE2-NEXT:    movdqa %xmm2, %xmm0
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm8[0],xmm0[1],xmm8[1],xmm0[2],xmm8[2],xmm0[3],xmm8[3]
+; SSE2-NEXT:    punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm8[4],xmm2[5],xmm8[5],xmm2[6],xmm8[6],xmm2[7],xmm8[7]
+; SSE2-NEXT:    punpckhbw {{.*#+}} xmm6 = xmm6[8],xmm8[8],xmm6[9],xmm8[9],xmm6[10],xmm8[10],xmm6[11],xmm8[11],xmm6[12],xmm8[12],xmm6[13],xmm8[13],xmm6[14],xmm8[14],xmm6[15],xmm8[15]
+; SSE2-NEXT:    movdqa %xmm6, %xmm3
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm8[0],xmm3[1],xmm8[1],xmm3[2],xmm8[2],xmm3[3],xmm8[3]
+; SSE2-NEXT:    punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm8[4],xmm6[5],xmm8[5],xmm6[6],xmm8[6],xmm6[7],xmm8[7]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm5 = [2147483648,2147483648,2147483648,2147483648]
+; SSE2-NEXT:    pxor %xmm5, %xmm9
+; SSE2-NEXT:    por %xmm5, %xmm6
+; SSE2-NEXT:    pcmpgtd %xmm9, %xmm6
+; SSE2-NEXT:    pxor %xmm5, %xmm10
+; SSE2-NEXT:    por %xmm5, %xmm3
+; SSE2-NEXT:    pcmpgtd %xmm10, %xmm3
+; SSE2-NEXT:    packssdw %xmm6, %xmm3
+; SSE2-NEXT:    pxor %xmm5, %xmm7
+; SSE2-NEXT:    por %xmm5, %xmm2
+; SSE2-NEXT:    pcmpgtd %xmm7, %xmm2
+; SSE2-NEXT:    pxor %xmm5, %xmm4
+; SSE2-NEXT:    por %xmm5, %xmm0
+; SSE2-NEXT:    pcmpgtd %xmm4, %xmm0
+; SSE2-NEXT:    packssdw %xmm2, %xmm0
+; SSE2-NEXT:    packsswb %xmm3, %xmm0
 ; SSE2-NEXT:    pandn %xmm1, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: test14:
 ; SSSE3:       # %bb.0: # %vector.ph
-; SSSE3-NEXT:    movdqa %xmm0, %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:    pxor %xmm8, %xmm8
+; SSSE3-NEXT:    movdqa %xmm0, %xmm6
 ; 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 %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 %xmm3, %xmm10
+; SSSE3-NEXT:    movdqa %xmm2, %xmm7
 ; 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
 ; SSSE3-NEXT:    packuswb %xmm4, %xmm3
+; SSSE3-NEXT:    movdqa %xmm1, %xmm4
 ; SSSE3-NEXT:    pand %xmm5, %xmm2
 ; SSSE3-NEXT:    pand %xmm5, %xmm1
 ; SSSE3-NEXT:    packuswb %xmm2, %xmm1
 ; SSSE3-NEXT:    packuswb %xmm3, %xmm1
+; SSSE3-NEXT:    psubb %xmm0, %xmm1
+; SSSE3-NEXT:    movdqa %xmm0, %xmm2
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm8[0],xmm2[1],xmm8[1],xmm2[2],xmm8[2],xmm2[3],xmm8[3],xmm2[4],xmm8[4],xmm2[5],xmm8[5],xmm2[6],xmm8[6],xmm2[7],xmm8[7]
+; SSSE3-NEXT:    movdqa %xmm2, %xmm0
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm8[0],xmm0[1],xmm8[1],xmm0[2],xmm8[2],xmm0[3],xmm8[3]
+; SSSE3-NEXT:    punpckhwd {{.*#+}} xmm2 = xmm2[4],xmm8[4],xmm2[5],xmm8[5],xmm2[6],xmm8[6],xmm2[7],xmm8[7]
+; SSSE3-NEXT:    punpckhbw {{.*#+}} xmm6 = xmm6[8],xmm8[8],xmm6[9],xmm8[9],xmm6[10],xmm8[10],xmm6[11],xmm8[11],xmm6[12],xmm8[12],xmm6[13],xmm8[13],xmm6[14],xmm8[14],xmm6[15],xmm8[15]
+; SSSE3-NEXT:    movdqa %xmm6, %xmm3
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm8[0],xmm3[1],xmm8[1],xmm3[2],xmm8[2],xmm3[3],xmm8[3]
+; SSSE3-NEXT:    punpckhwd {{.*#+}} xmm6 = xmm6[4],xmm8[4],xmm6[5],xmm8[5],xmm6[6],xmm8[6],xmm6[7],xmm8[7]
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm5 = [2147483648,2147483648,2147483648,2147483648]
+; SSSE3-NEXT:    pxor %xmm5, %xmm9
+; SSSE3-NEXT:    por %xmm5, %xmm6
+; SSSE3-NEXT:    pcmpgtd %xmm9, %xmm6
+; SSSE3-NEXT:    pxor %xmm5, %xmm10
+; SSSE3-NEXT:    por %xmm5, %xmm3
+; SSSE3-NEXT:    pcmpgtd %xmm10, %xmm3
+; SSSE3-NEXT:    packssdw %xmm6, %xmm3
+; SSSE3-NEXT:    pxor %xmm5, %xmm7
+; SSSE3-NEXT:    por %xmm5, %xmm2
+; SSSE3-NEXT:    pcmpgtd %xmm7, %xmm2
+; SSSE3-NEXT:    pxor %xmm5, %xmm4
+; SSSE3-NEXT:    por %xmm5, %xmm0
+; SSSE3-NEXT:    pcmpgtd %xmm4, %xmm0
+; SSSE3-NEXT:    packssdw %xmm2, %xmm0
+; SSSE3-NEXT:    packsswb %xmm3, %xmm0
 ; SSSE3-NEXT:    pandn %xmm1, %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: test14:
 ; SSE41:       # %bb.0: # %vector.ph
 ; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm0[1,1,2,3]
-; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm11 = xmm5[0],zero,zero,zero,xmm5[1],zero,zero,zero,xmm5[2],zero,zero,zero,xmm5[3],zero,zero,zero
-; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm8 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
-; SSE41-NEXT:    pshufd {{.*#+}} xmm6 = xmm0[2,3,0,1]
-; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm9 = xmm6[0],zero,zero,zero,xmm6[1],zero,zero,zero,xmm6[2],zero,zero,zero,xmm6[3],zero,zero,zero
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,2,3]
-; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm10 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
-; SSE41-NEXT:    movdqa %xmm4, %xmm0
-; SSE41-NEXT:    pmaxud %xmm10, %xmm0
-; SSE41-NEXT:    pcmpeqd %xmm4, %xmm0
-; SSE41-NEXT:    pcmpeqd %xmm6, %xmm6
-; SSE41-NEXT:    pxor %xmm6, %xmm0
-; SSE41-NEXT:    movdqa %xmm3, %xmm7
-; SSE41-NEXT:    pmaxud %xmm9, %xmm7
+; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm8 = xmm5[0],zero,zero,zero,xmm5[1],zero,zero,zero,xmm5[2],zero,zero,zero,xmm5[3],zero,zero,zero
+; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm5 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; SSE41-NEXT:    pshufd {{.*#+}} xmm7 = xmm0[2,3,0,1]
+; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm7 = xmm7[0],zero,zero,zero,xmm7[1],zero,zero,zero,xmm7[2],zero,zero,zero,xmm7[3],zero,zero,zero
+; SSE41-NEXT:    pshufd {{.*#+}} xmm6 = xmm0[3,1,2,3]
+; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm6 = xmm6[0],zero,zero,zero,xmm6[1],zero,zero,zero,xmm6[2],zero,zero,zero,xmm6[3],zero,zero,zero
+; SSE41-NEXT:    pmaxud %xmm4, %xmm6
+; SSE41-NEXT:    pcmpeqd %xmm4, %xmm6
+; SSE41-NEXT:    pcmpeqd %xmm9, %xmm9
+; SSE41-NEXT:    pxor %xmm9, %xmm6
+; SSE41-NEXT:    pmaxud %xmm3, %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 %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
-; SSE41-NEXT:    psubd %xmm10, %xmm4
-; SSE41-NEXT:    movdqa {{.*#+}} xmm5 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
-; SSE41-NEXT:    pand %xmm5, %xmm4
-; SSE41-NEXT:    pand %xmm5, %xmm3
+; SSE41-NEXT:    pxor %xmm9, %xmm7
+; SSE41-NEXT:    packssdw %xmm6, %xmm7
+; SSE41-NEXT:    pmaxud %xmm1, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm1, %xmm5
+; SSE41-NEXT:    pxor %xmm9, %xmm5
+; SSE41-NEXT:    pmaxud %xmm2, %xmm8
+; SSE41-NEXT:    pcmpeqd %xmm2, %xmm8
+; SSE41-NEXT:    pxor %xmm9, %xmm8
+; SSE41-NEXT:    packssdw %xmm8, %xmm5
+; SSE41-NEXT:    packsswb %xmm7, %xmm5
+; SSE41-NEXT:    movdqa {{.*#+}} xmm6 = [255,0,0,0,255,0,0,0,255,0,0,0,255,0,0,0]
+; SSE41-NEXT:    pand %xmm6, %xmm4
+; SSE41-NEXT:    pand %xmm6, %xmm3
 ; SSE41-NEXT:    packusdw %xmm4, %xmm3
-; SSE41-NEXT:    pand %xmm5, %xmm1
-; SSE41-NEXT:    pand %xmm5, %xmm2
+; SSE41-NEXT:    pand %xmm6, %xmm2
+; SSE41-NEXT:    pand %xmm6, %xmm1
 ; SSE41-NEXT:    packusdw %xmm2, %xmm1
 ; SSE41-NEXT:    packuswb %xmm3, %xmm1
-; SSE41-NEXT:    pandn %xmm1, %xmm0
+; SSE41-NEXT:    psubb %xmm0, %xmm1
+; SSE41-NEXT:    pandn %xmm1, %xmm5
+; SSE41-NEXT:    movdqa %xmm5, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: test14:
 ; AVX1:       # %bb.0: # %vector.ph
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,2,3]
-; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm8 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
-; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm9 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm4 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm5 = xmm0[2,3,0,1]
-; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm11 = xmm5[0],zero,zero,zero,xmm5[1],zero,zero,zero,xmm5[2],zero,zero,zero,xmm5[3],zero,zero,zero
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,3,0,1]
-; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
-; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm6
-; AVX1-NEXT:    vpmaxud %xmm0, %xmm6, %xmm7
-; AVX1-NEXT:    vpcmpeqd %xmm7, %xmm6, %xmm7
-; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm3, %xmm3
-; AVX1-NEXT:    vpxor %xmm3, %xmm7, %xmm7
-; AVX1-NEXT:    vpmaxud %xmm11, %xmm2, %xmm4
-; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm2, %xmm4
-; AVX1-NEXT:    vpxor %xmm3, %xmm4, %xmm4
-; AVX1-NEXT:    vpackssdw %xmm7, %xmm4, %xmm10
-; AVX1-NEXT:    vpmaxud %xmm9, %xmm1, %xmm7
-; AVX1-NEXT:    vpcmpeqd %xmm7, %xmm1, %xmm7
-; AVX1-NEXT:    vpxor %xmm3, %xmm7, %xmm7
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm5 = xmm5[0],zero,zero,zero,xmm5[1],zero,zero,zero,xmm5[2],zero,zero,zero,xmm5[3],zero,zero,zero
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm6 = xmm0[3,3,0,1]
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm6 = xmm6[0],zero,zero,zero,xmm6[1],zero,zero,zero,xmm6[2],zero,zero,zero,xmm6[3],zero,zero,zero
+; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm7
+; AVX1-NEXT:    vpmaxud %xmm6, %xmm7, %xmm6
+; AVX1-NEXT:    vpcmpeqd %xmm6, %xmm7, %xmm6
+; AVX1-NEXT:    vpcmpeqd %xmm7, %xmm7, %xmm7
+; AVX1-NEXT:    vpxor %xmm7, %xmm6, %xmm6
+; AVX1-NEXT:    vpmaxud %xmm5, %xmm2, %xmm5
+; AVX1-NEXT:    vpcmpeqd %xmm5, %xmm2, %xmm5
+; AVX1-NEXT:    vpxor %xmm7, %xmm5, %xmm5
+; AVX1-NEXT:    vpackssdw %xmm6, %xmm5, %xmm5
+; AVX1-NEXT:    vpmaxud %xmm4, %xmm1, %xmm4
+; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm1, %xmm4
+; AVX1-NEXT:    vpxor %xmm7, %xmm4, %xmm4
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm6
+; AVX1-NEXT:    vpmaxud %xmm3, %xmm6, %xmm3
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm6, %xmm3
+; AVX1-NEXT:    vpxor %xmm7, %xmm3, %xmm3
+; AVX1-NEXT:    vpackssdw %xmm3, %xmm4, %xmm3
+; AVX1-NEXT:    vpacksswb %xmm5, %xmm3, %xmm3
+; AVX1-NEXT:    vmovaps {{.*#+}} ymm4 = [255,255,255,255,255,255,255,255]
+; AVX1-NEXT:    vandps %ymm4, %ymm2, %ymm2
+; AVX1-NEXT:    vextractf128 $1, %ymm2, %xmm5
+; AVX1-NEXT:    vpackusdw %xmm5, %xmm2, %xmm2
+; AVX1-NEXT:    vandps %ymm4, %ymm1, %ymm1
 ; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm4
-; AVX1-NEXT:    vpmaxud %xmm8, %xmm4, %xmm5
-; AVX1-NEXT:    vpcmpeqd %xmm5, %xmm4, %xmm5
-; AVX1-NEXT:    vpxor %xmm3, %xmm5, %xmm3
-; AVX1-NEXT:    vpackssdw %xmm3, %xmm7, %xmm3
-; AVX1-NEXT:    vpacksswb %xmm10, %xmm3, %xmm3
-; AVX1-NEXT:    vpsubd %xmm8, %xmm4, %xmm4
-; AVX1-NEXT:    vpsubd %xmm9, %xmm1, %xmm1
-; AVX1-NEXT:    vpsubd %xmm11, %xmm2, %xmm2
-; AVX1-NEXT:    vpsubd %xmm0, %xmm6, %xmm0
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm5 = [255,255,255,255]
-; AVX1-NEXT:    vpand %xmm5, %xmm0, %xmm0
-; AVX1-NEXT:    vpand %xmm5, %xmm2, %xmm2
-; AVX1-NEXT:    vpackusdw %xmm0, %xmm2, %xmm0
-; AVX1-NEXT:    vpand %xmm5, %xmm1, %xmm1
-; AVX1-NEXT:    vpand %xmm5, %xmm4, %xmm2
-; AVX1-NEXT:    vpackusdw %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpackuswb %xmm0, %xmm1, %xmm0
+; AVX1-NEXT:    vpackusdw %xmm4, %xmm1, %xmm1
+; AVX1-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
+; AVX1-NEXT:    vpsubb %xmm0, %xmm1, %xmm0
 ; AVX1-NEXT:    vpandn %xmm0, %xmm3, %xmm0
 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: test14:
 ; AVX2:       # %bb.0: # %vector.ph
 ; AVX2-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
 ; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm3 = xmm3[0],zero,zero,zero,xmm3[1],zero,zero,zero,xmm3[2],zero,zero,zero,xmm3[3],zero,zero,zero,xmm3[4],zero,zero,zero,xmm3[5],zero,zero,zero,xmm3[6],zero,zero,zero,xmm3[7],zero,zero,zero
-; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
-; AVX2-NEXT:    vpmaxud %ymm0, %ymm1, %ymm4
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} ymm4 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero
+; AVX2-NEXT:    vpmaxud %ymm4, %ymm1, %ymm4
 ; AVX2-NEXT:    vpcmpeqd %ymm4, %ymm1, %ymm4
 ; AVX2-NEXT:    vpcmpeqd %ymm5, %ymm5, %ymm5
 ; AVX2-NEXT:    vpxor %ymm5, %ymm4, %ymm4
 ; AVX2-NEXT:    vextracti128 $1, %ymm4, %xmm6
 ; AVX2-NEXT:    vpackssdw %xmm6, %xmm4, %xmm4
-; AVX2-NEXT:    vpmaxud %ymm3, %ymm2, %ymm6
-; AVX2-NEXT:    vpcmpeqd %ymm6, %ymm2, %ymm6
-; AVX2-NEXT:    vpxor %ymm5, %ymm6, %ymm5
-; AVX2-NEXT:    vextracti128 $1, %ymm5, %xmm6
-; AVX2-NEXT:    vpackssdw %xmm6, %xmm5, %xmm5
-; AVX2-NEXT:    vpacksswb %xmm5, %xmm4, %xmm4
-; AVX2-NEXT:    vpsubd %ymm3, %ymm2, %ymm2
-; AVX2-NEXT:    vpsubd %ymm0, %ymm1, %ymm0
-; AVX2-NEXT:    vmovdqa {{.*#+}} 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:    vpshufb %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,2,2,3]
-; AVX2-NEXT:    vmovdqa {{.*#+}} xmm3 = [255,255,255,255,255,255,255,255]
-; AVX2-NEXT:    vpand %xmm3, %xmm0, %xmm0
-; AVX2-NEXT:    vpshufb %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vpmaxud %ymm3, %ymm2, %ymm3
+; AVX2-NEXT:    vpcmpeqd %ymm3, %ymm2, %ymm3
+; AVX2-NEXT:    vpxor %ymm5, %ymm3, %ymm3
+; AVX2-NEXT:    vextracti128 $1, %ymm3, %xmm5
+; AVX2-NEXT:    vpackssdw %xmm5, %xmm3, %xmm3
+; AVX2-NEXT:    vpacksswb %xmm3, %xmm4, %xmm3
+; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = [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:    vpshufb %ymm4, %ymm2, %ymm2
+; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[0,2,2,3]
+; AVX2-NEXT:    vmovdqa {{.*#+}} xmm5 = [255,255,255,255,255,255,255,255]
+; AVX2-NEXT:    vpand %xmm5, %xmm2, %xmm2
+; AVX2-NEXT:    vpshufb %ymm4, %ymm1, %ymm1
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[0,2,2,3]
-; AVX2-NEXT:    vpand %xmm3, %xmm1, %xmm1
-; AVX2-NEXT:    vpackuswb %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vpandn %xmm0, %xmm4, %xmm0
+; AVX2-NEXT:    vpand %xmm5, %xmm1, %xmm1
+; AVX2-NEXT:    vpackuswb %xmm2, %xmm1, %xmm1
+; AVX2-NEXT:    vpsubb %xmm0, %xmm1, %xmm0
+; AVX2-NEXT:    vpandn %xmm0, %xmm3, %xmm0
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test14:
 ; AVX512:       # %bb.0: # %vector.ph
-; AVX512-NEXT:    vpmovzxbd {{.*#+}} zmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
-; AVX512-NEXT:    vpcmpnltud %zmm0, %zmm1, %k1
-; AVX512-NEXT:    vpsubd %zmm0, %zmm1, %zmm0
-; AVX512-NEXT:    vpmovdb %zmm0, %xmm0 {%k1} {z}
+; AVX512-NEXT:    vpmovzxbd {{.*#+}} zmm2 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[4],zero,zero,zero,xmm0[5],zero,zero,zero,xmm0[6],zero,zero,zero,xmm0[7],zero,zero,zero,xmm0[8],zero,zero,zero,xmm0[9],zero,zero,zero,xmm0[10],zero,zero,zero,xmm0[11],zero,zero,zero,xmm0[12],zero,zero,zero,xmm0[13],zero,zero,zero,xmm0[14],zero,zero,zero,xmm0[15],zero,zero,zero
+; AVX512-NEXT:    vpcmpnltud %zmm2, %zmm1, %k1
+; AVX512-NEXT:    vpmovdb %zmm1, %xmm1
+; AVX512-NEXT:    vpsubb %xmm0, %xmm1, %xmm0 {%k1} {z}
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
 vector.ph:
   %rhs = zext <16 x i8> %x to <16 x i32>
   %cond = icmp ult <16 x i32> %y, %rhs
   %sub = sub <16 x i32> %y, %rhs
   %truncsub = trunc <16 x i32> %sub to <16 x i8>
   %res = select <16 x i1> %cond, <16 x i8> zeroinitializer, <16 x i8> %truncsub
   ret <16 x i8> %res
 }
 
+; FIXME: match this to UMIN+TRUNC+PSUBUS
 define <8 x i16> @test15(<8 x i16> %x, <8 x i32> %y) nounwind {
 ; SSE2-LABEL: test15:
 ; SSE2:       # %bb.0: # %vector.ph
-; SSE2-NEXT:    movdqa %xmm0, %xmm3
-; SSE2-NEXT:    pxor %xmm4, %xmm4
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3]
-; SSE2-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm4[4],xmm3[5],xmm4[5],xmm3[6],xmm4[6],xmm3[7],xmm4[7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [2147483648,2147483648,2147483648,2147483648]
-; SSE2-NEXT:    movdqa %xmm3, %xmm5
-; SSE2-NEXT:    psubd %xmm2, %xmm3
-; SSE2-NEXT:    pxor %xmm4, %xmm2
-; SSE2-NEXT:    por %xmm4, %xmm5
-; SSE2-NEXT:    pcmpgtd %xmm2, %xmm5
-; SSE2-NEXT:    movdqa %xmm1, %xmm2
-; SSE2-NEXT:    pxor %xmm4, %xmm2
-; SSE2-NEXT:    por %xmm0, %xmm4
-; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pxor %xmm3, %xmm3
+; SSE2-NEXT:    movdqa %xmm0, %xmm4
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1],xmm4[2],xmm3[2],xmm4[3],xmm3[3]
+; SSE2-NEXT:    movdqa %xmm0, %xmm5
+; SSE2-NEXT:    punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm3[4],xmm5[5],xmm3[5],xmm5[6],xmm3[6],xmm5[7],xmm3[7]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm2, %xmm6
+; SSE2-NEXT:    pxor %xmm3, %xmm6
+; SSE2-NEXT:    por %xmm3, %xmm5
+; SSE2-NEXT:    pcmpgtd %xmm6, %xmm5
+; SSE2-NEXT:    movdqa %xmm1, %xmm6
+; SSE2-NEXT:    pxor %xmm3, %xmm6
+; SSE2-NEXT:    por %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm6, %xmm4
 ; SSE2-NEXT:    packssdw %xmm5, %xmm4
-; SSE2-NEXT:    psubd %xmm1, %xmm0
-; SSE2-NEXT:    pslld $16, %xmm3
-; SSE2-NEXT:    psrad $16, %xmm3
-; SSE2-NEXT:    pslld $16, %xmm0
-; SSE2-NEXT:    psrad $16, %xmm0
-; SSE2-NEXT:    packssdw %xmm3, %xmm0
+; SSE2-NEXT:    pslld $16, %xmm2
+; SSE2-NEXT:    psrad $16, %xmm2
+; SSE2-NEXT:    pslld $16, %xmm1
+; SSE2-NEXT:    psrad $16, %xmm1
+; SSE2-NEXT:    packssdw %xmm2, %xmm1
+; SSE2-NEXT:    psubw %xmm1, %xmm0
 ; SSE2-NEXT:    pand %xmm4, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: test15:
 ; SSSE3:       # %bb.0: # %vector.ph
-; SSSE3-NEXT:    pxor %xmm4, %xmm4
-; SSSE3-NEXT:    movdqa %xmm0, %xmm3
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
-; SSSE3-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm4 = [2147483648,2147483648,2147483648,2147483648]
+; SSSE3-NEXT:    pxor %xmm3, %xmm3
+; SSSE3-NEXT:    movdqa %xmm0, %xmm4
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1],xmm4[2],xmm3[2],xmm4[3],xmm3[3]
 ; SSSE3-NEXT:    movdqa %xmm0, %xmm5
-; SSSE3-NEXT:    psubd %xmm2, %xmm0
-; SSSE3-NEXT:    pxor %xmm4, %xmm2
-; SSSE3-NEXT:    por %xmm4, %xmm5
-; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm5
-; SSSE3-NEXT:    movdqa %xmm1, %xmm2
-; SSSE3-NEXT:    pxor %xmm4, %xmm2
+; SSSE3-NEXT:    punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm3[4],xmm5[5],xmm3[5],xmm5[6],xmm3[6],xmm5[7],xmm3[7]
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm2, %xmm6
+; SSSE3-NEXT:    pxor %xmm3, %xmm6
+; SSSE3-NEXT:    por %xmm3, %xmm5
+; SSSE3-NEXT:    pcmpgtd %xmm6, %xmm5
+; SSSE3-NEXT:    movdqa %xmm1, %xmm6
+; SSSE3-NEXT:    pxor %xmm3, %xmm6
 ; SSSE3-NEXT:    por %xmm3, %xmm4
-; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSSE3-NEXT:    pcmpgtd %xmm6, %xmm4
 ; SSSE3-NEXT:    packssdw %xmm5, %xmm4
-; SSSE3-NEXT:    psubd %xmm1, %xmm3
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; SSSE3-NEXT:    pshufb %xmm1, %xmm0
-; SSSE3-NEXT:    pshufb %xmm1, %xmm3
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm0[0]
-; SSSE3-NEXT:    pand %xmm4, %xmm3
-; SSSE3-NEXT:    movdqa %xmm3, %xmm0
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; SSSE3-NEXT:    pshufb %xmm3, %xmm2
+; SSSE3-NEXT:    pshufb %xmm3, %xmm1
+; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; SSSE3-NEXT:    psubw %xmm1, %xmm0
+; SSSE3-NEXT:    pand %xmm4, %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; 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 %xmm0, %xmm4
-; SSE41-NEXT:    pminud %xmm1, %xmm4
-; SSE41-NEXT:    pcmpeqd %xmm0, %xmm4
-; SSE41-NEXT:    pcmpeqd %xmm5, %xmm5
-; SSE41-NEXT:    pxor %xmm5, %xmm4
+; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm4 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
+; SSE41-NEXT:    movdqa %xmm4, %xmm5
+; SSE41-NEXT:    pminud %xmm1, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm4, %xmm5
+; SSE41-NEXT:    pcmpeqd %xmm4, %xmm4
+; SSE41-NEXT:    pxor %xmm4, %xmm5
 ; SSE41-NEXT:    movdqa %xmm3, %xmm6
 ; SSE41-NEXT:    pminud %xmm2, %xmm6
 ; SSE41-NEXT:    pcmpeqd %xmm3, %xmm6
-; SSE41-NEXT:    pxor %xmm5, %xmm6
-; SSE41-NEXT:    packssdw %xmm6, %xmm4
-; SSE41-NEXT:    psubd %xmm2, %xmm3
-; SSE41-NEXT:    psubd %xmm1, %xmm0
-; SSE41-NEXT:    pxor %xmm1, %xmm1
-; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3],xmm0[4],xmm1[5],xmm0[6],xmm1[7]
-; SSE41-NEXT:    pblendw {{.*#+}} xmm3 = xmm3[0],xmm1[1],xmm3[2],xmm1[3],xmm3[4],xmm1[5],xmm3[6],xmm1[7]
-; SSE41-NEXT:    packusdw %xmm3, %xmm0
-; SSE41-NEXT:    pand %xmm4, %xmm0
+; SSE41-NEXT:    pxor %xmm4, %xmm6
+; SSE41-NEXT:    packssdw %xmm6, %xmm5
+; SSE41-NEXT:    pxor %xmm3, %xmm3
+; SSE41-NEXT:    pblendw {{.*#+}} xmm2 = xmm2[0],xmm3[1],xmm2[2],xmm3[3],xmm2[4],xmm3[5],xmm2[6],xmm3[7]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0],xmm3[1],xmm1[2],xmm3[3],xmm1[4],xmm3[5],xmm1[6],xmm3[7]
+; SSE41-NEXT:    packusdw %xmm2, %xmm1
+; SSE41-NEXT:    psubw %xmm1, %xmm0
+; SSE41-NEXT:    pand %xmm5, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: test15:
 ; AVX1:       # %bb.0: # %vector.ph
 ; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm2 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm3
-; AVX1-NEXT:    vpminud %xmm3, %xmm2, %xmm4
-; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm2, %xmm4
-; AVX1-NEXT:    vpminud %xmm1, %xmm0, %xmm5
-; AVX1-NEXT:    vpcmpeqd %xmm5, %xmm0, %xmm5
-; AVX1-NEXT:    vpsubd %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpandn %xmm0, %xmm5, %xmm0
-; AVX1-NEXT:    vpsubd %xmm3, %xmm2, %xmm1
-; AVX1-NEXT:    vpandn %xmm1, %xmm4, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm3 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm4
+; AVX1-NEXT:    vpminud %xmm4, %xmm2, %xmm5
+; AVX1-NEXT:    vpcmpeqd %xmm5, %xmm2, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm5, %xmm5, %xmm5
+; AVX1-NEXT:    vpxor %xmm5, %xmm2, %xmm2
+; AVX1-NEXT:    vpminud %xmm1, %xmm3, %xmm6
+; AVX1-NEXT:    vpcmpeqd %xmm6, %xmm3, %xmm3
+; AVX1-NEXT:    vpxor %xmm5, %xmm3, %xmm3
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX1-NEXT:    vpshufb %xmm3, %xmm4, %xmm4
+; AVX1-NEXT:    vpshufb %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm4[0]
+; AVX1-NEXT:    vpsubw %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpand %xmm0, %xmm2, %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:    vpminud %ymm1, %ymm0, %ymm2
-; AVX2-NEXT:    vpcmpeqd %ymm2, %ymm0, %ymm2
-; AVX2-NEXT:    vpsubd %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpandn %ymm0, %ymm2, %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:    # kill: def $xmm0 killed $xmm0 killed $ymm0
+; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm2 = 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:    vpminud %ymm1, %ymm2, %ymm3
+; AVX2-NEXT:    vpcmpeqd %ymm3, %ymm2, %ymm2
+; AVX2-NEXT:    vpcmpeqd %ymm3, %ymm3, %ymm3
+; AVX2-NEXT:    vpxor %ymm3, %ymm2, %ymm2
+; AVX2-NEXT:    vextracti128 $1, %ymm2, %xmm3
+; AVX2-NEXT:    vpackssdw %xmm3, %xmm2, %xmm2
+; 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:    vpsubw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpand %xmm0, %xmm2, %xmm0
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test15:
 ; AVX512:       # %bb.0: # %vector.ph
-; AVX512-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
-; AVX512-NEXT:    vpcmpnleud %ymm1, %ymm0, %k1
-; AVX512-NEXT:    vpsubd %ymm1, %ymm0, %ymm0
-; AVX512-NEXT:    vpmovdw %ymm0, %xmm0 {%k1} {z}
+; AVX512-NEXT:    vpmovzxwd {{.*#+}} ymm2 = 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:    vpcmpnleud %ymm1, %ymm2, %k1
+; AVX512-NEXT:    vpmovdw %ymm1, %xmm1
+; AVX512-NEXT:    vpsubw %xmm1, %xmm0, %xmm0 {%k1} {z}
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
 vector.ph:
   %lhs = zext <8 x i16> %x to <8 x i32>
   %cond = icmp ugt <8 x i32> %lhs, %y
   %sub = sub <8 x i32> %lhs, %y
   %truncsub = trunc <8 x i32> %sub to <8 x i16>
   %res = select <8 x i1> %cond, <8 x i16> %truncsub, <8 x i16> zeroinitializer
   ret <8 x i16> %res
 }
 
+; FIXME: match this to UMIN+TRUNC+PSUBUS
 define <8 x i16> @test16(<8 x i16> %x, <8 x i32> %y) nounwind {
 ; SSE2-LABEL: test16:
 ; SSE2:       # %bb.0: # %vector.ph
-; SSE2-NEXT:    movdqa %xmm0, %xmm3
-; SSE2-NEXT:    pxor %xmm4, %xmm4
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm4[0],xmm0[1],xmm4[1],xmm0[2],xmm4[2],xmm0[3],xmm4[3]
-; SSE2-NEXT:    punpckhwd {{.*#+}} xmm3 = xmm3[4],xmm4[4],xmm3[5],xmm4[5],xmm3[6],xmm4[6],xmm3[7],xmm4[7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [2147483648,2147483648,2147483648,2147483648]
-; SSE2-NEXT:    movdqa %xmm3, %xmm5
-; SSE2-NEXT:    psubd %xmm2, %xmm3
-; SSE2-NEXT:    pxor %xmm4, %xmm2
-; SSE2-NEXT:    por %xmm4, %xmm5
-; SSE2-NEXT:    pcmpgtd %xmm2, %xmm5
-; SSE2-NEXT:    movdqa %xmm1, %xmm2
-; SSE2-NEXT:    pxor %xmm4, %xmm2
-; SSE2-NEXT:    por %xmm0, %xmm4
-; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pxor %xmm3, %xmm3
+; SSE2-NEXT:    movdqa %xmm0, %xmm4
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1],xmm4[2],xmm3[2],xmm4[3],xmm3[3]
+; SSE2-NEXT:    movdqa %xmm0, %xmm5
+; SSE2-NEXT:    punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm3[4],xmm5[5],xmm3[5],xmm5[6],xmm3[6],xmm5[7],xmm3[7]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm2, %xmm6
+; SSE2-NEXT:    pxor %xmm3, %xmm6
+; SSE2-NEXT:    por %xmm3, %xmm5
+; SSE2-NEXT:    pcmpgtd %xmm6, %xmm5
+; SSE2-NEXT:    movdqa %xmm1, %xmm6
+; SSE2-NEXT:    pxor %xmm3, %xmm6
+; SSE2-NEXT:    por %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm6, %xmm4
 ; SSE2-NEXT:    packssdw %xmm5, %xmm4
-; SSE2-NEXT:    psubd %xmm1, %xmm0
-; SSE2-NEXT:    pslld $16, %xmm3
-; SSE2-NEXT:    psrad $16, %xmm3
-; SSE2-NEXT:    pslld $16, %xmm0
-; SSE2-NEXT:    psrad $16, %xmm0
-; SSE2-NEXT:    packssdw %xmm3, %xmm0
+; SSE2-NEXT:    pslld $16, %xmm2
+; SSE2-NEXT:    psrad $16, %xmm2
+; SSE2-NEXT:    pslld $16, %xmm1
+; SSE2-NEXT:    psrad $16, %xmm1
+; SSE2-NEXT:    packssdw %xmm2, %xmm1
+; SSE2-NEXT:    psubw %xmm1, %xmm0
 ; SSE2-NEXT:    pand %xmm4, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: test16:
 ; SSSE3:       # %bb.0: # %vector.ph
-; SSSE3-NEXT:    pxor %xmm4, %xmm4
-; SSSE3-NEXT:    movdqa %xmm0, %xmm3
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
-; SSSE3-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm4 = [2147483648,2147483648,2147483648,2147483648]
+; SSSE3-NEXT:    pxor %xmm3, %xmm3
+; SSSE3-NEXT:    movdqa %xmm0, %xmm4
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1],xmm4[2],xmm3[2],xmm4[3],xmm3[3]
 ; SSSE3-NEXT:    movdqa %xmm0, %xmm5
-; SSSE3-NEXT:    psubd %xmm2, %xmm0
-; SSSE3-NEXT:    pxor %xmm4, %xmm2
-; SSSE3-NEXT:    por %xmm4, %xmm5
-; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm5
-; SSSE3-NEXT:    movdqa %xmm1, %xmm2
-; SSSE3-NEXT:    pxor %xmm4, %xmm2
+; SSSE3-NEXT:    punpckhwd {{.*#+}} xmm5 = xmm5[4],xmm3[4],xmm5[5],xmm3[5],xmm5[6],xmm3[6],xmm5[7],xmm3[7]
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm2, %xmm6
+; SSSE3-NEXT:    pxor %xmm3, %xmm6
+; SSSE3-NEXT:    por %xmm3, %xmm5
+; SSSE3-NEXT:    pcmpgtd %xmm6, %xmm5
+; SSSE3-NEXT:    movdqa %xmm1, %xmm6
+; SSSE3-NEXT:    pxor %xmm3, %xmm6
 ; SSSE3-NEXT:    por %xmm3, %xmm4
-; SSSE3-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSSE3-NEXT:    pcmpgtd %xmm6, %xmm4
 ; SSSE3-NEXT:    packssdw %xmm5, %xmm4
-; SSSE3-NEXT:    psubd %xmm1, %xmm3
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm1 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; SSSE3-NEXT:    pshufb %xmm1, %xmm0
-; SSSE3-NEXT:    pshufb %xmm1, %xmm3
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm3 = xmm3[0],xmm0[0]
-; SSSE3-NEXT:    pand %xmm4, %xmm3
-; SSSE3-NEXT:    movdqa %xmm3, %xmm0
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; SSSE3-NEXT:    pshufb %xmm3, %xmm2
+; SSSE3-NEXT:    pshufb %xmm3, %xmm1
+; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; SSSE3-NEXT:    psubw %xmm1, %xmm0
+; SSSE3-NEXT:    pand %xmm4, %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; 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 %xmm0, %xmm4
-; SSE41-NEXT:    pminud %xmm1, %xmm4
-; SSE41-NEXT:    pcmpeqd %xmm0, %xmm4
+; SSE41-NEXT:    pmovzxwd {{.*#+}} xmm4 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
+; SSE41-NEXT:    pmaxud %xmm1, %xmm4
+; SSE41-NEXT:    pcmpeqd %xmm1, %xmm4
 ; SSE41-NEXT:    pcmpeqd %xmm5, %xmm5
 ; SSE41-NEXT:    pxor %xmm5, %xmm4
-; SSE41-NEXT:    movdqa %xmm3, %xmm6
-; SSE41-NEXT:    pminud %xmm2, %xmm6
-; SSE41-NEXT:    pcmpeqd %xmm3, %xmm6
-; SSE41-NEXT:    pxor %xmm5, %xmm6
-; SSE41-NEXT:    packssdw %xmm6, %xmm4
-; SSE41-NEXT:    psubd %xmm2, %xmm3
-; SSE41-NEXT:    psubd %xmm1, %xmm0
-; SSE41-NEXT:    pxor %xmm1, %xmm1
-; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3],xmm0[4],xmm1[5],xmm0[6],xmm1[7]
-; SSE41-NEXT:    pblendw {{.*#+}} xmm3 = xmm3[0],xmm1[1],xmm3[2],xmm1[3],xmm3[4],xmm1[5],xmm3[6],xmm1[7]
-; SSE41-NEXT:    packusdw %xmm3, %xmm0
+; SSE41-NEXT:    pmaxud %xmm2, %xmm3
+; SSE41-NEXT:    pcmpeqd %xmm2, %xmm3
+; SSE41-NEXT:    pxor %xmm5, %xmm3
+; SSE41-NEXT:    packssdw %xmm3, %xmm4
+; SSE41-NEXT:    pxor %xmm3, %xmm3
+; SSE41-NEXT:    pblendw {{.*#+}} xmm2 = xmm2[0],xmm3[1],xmm2[2],xmm3[3],xmm2[4],xmm3[5],xmm2[6],xmm3[7]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0],xmm3[1],xmm1[2],xmm3[3],xmm1[4],xmm3[5],xmm1[6],xmm3[7]
+; SSE41-NEXT:    packusdw %xmm2, %xmm1
+; SSE41-NEXT:    psubw %xmm1, %xmm0
 ; SSE41-NEXT:    pand %xmm4, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: test16:
 ; AVX1:       # %bb.0: # %vector.ph
 ; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm2 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
-; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm3
-; AVX1-NEXT:    vpminud %xmm3, %xmm2, %xmm4
-; AVX1-NEXT:    vpcmpeqd %xmm4, %xmm2, %xmm4
-; AVX1-NEXT:    vpminud %xmm1, %xmm0, %xmm5
-; AVX1-NEXT:    vpcmpeqd %xmm5, %xmm0, %xmm5
-; AVX1-NEXT:    vpsubd %xmm1, %xmm0, %xmm0
-; AVX1-NEXT:    vpandn %xmm0, %xmm5, %xmm0
-; AVX1-NEXT:    vpsubd %xmm3, %xmm2, %xmm1
-; AVX1-NEXT:    vpandn %xmm1, %xmm4, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm3 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm4
+; AVX1-NEXT:    vpmaxud %xmm2, %xmm4, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm2, %xmm4, %xmm2
+; AVX1-NEXT:    vpcmpeqd %xmm5, %xmm5, %xmm5
+; AVX1-NEXT:    vpxor %xmm5, %xmm2, %xmm2
+; AVX1-NEXT:    vpmaxud %xmm3, %xmm1, %xmm3
+; AVX1-NEXT:    vpcmpeqd %xmm3, %xmm1, %xmm3
+; AVX1-NEXT:    vpxor %xmm5, %xmm3, %xmm3
+; AVX1-NEXT:    vpackssdw %xmm2, %xmm3, %xmm2
+; AVX1-NEXT:    vmovdqa {{.*#+}} xmm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX1-NEXT:    vpshufb %xmm3, %xmm4, %xmm4
+; AVX1-NEXT:    vpshufb %xmm3, %xmm1, %xmm1
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm4[0]
+; AVX1-NEXT:    vpsubw %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpand %xmm0, %xmm2, %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:    vpminud %ymm1, %ymm0, %ymm2
-; AVX2-NEXT:    vpcmpeqd %ymm2, %ymm0, %ymm2
-; AVX2-NEXT:    vpsubd %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vpandn %ymm0, %ymm2, %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:    # kill: def $xmm0 killed $xmm0 killed $ymm0
+; AVX2-NEXT:    vpmovzxwd {{.*#+}} ymm2 = 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:    vpmaxud %ymm2, %ymm1, %ymm2
+; AVX2-NEXT:    vpcmpeqd %ymm2, %ymm1, %ymm2
+; AVX2-NEXT:    vpcmpeqd %ymm3, %ymm3, %ymm3
+; AVX2-NEXT:    vpxor %ymm3, %ymm2, %ymm2
+; AVX2-NEXT:    vextracti128 $1, %ymm2, %xmm3
+; AVX2-NEXT:    vpackssdw %xmm3, %xmm2, %xmm2
+; 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:    vpsubw %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpand %xmm0, %xmm2, %xmm0
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: test16:
 ; AVX512:       # %bb.0: # %vector.ph
-; AVX512-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
-; AVX512-NEXT:    vpcmpnleud %ymm1, %ymm0, %k1
-; AVX512-NEXT:    vpsubd %ymm1, %ymm0, %ymm0
-; AVX512-NEXT:    vpmovdw %ymm0, %xmm0 {%k1} {z}
+; AVX512-NEXT:    vpmovzxwd {{.*#+}} ymm2 = 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:    vpcmpltud %ymm2, %ymm1, %k1
+; AVX512-NEXT:    vpmovdw %ymm1, %xmm1
+; AVX512-NEXT:    vpsubw %xmm1, %xmm0, %xmm0 {%k1} {z}
 ; AVX512-NEXT:    vzeroupper
 ; AVX512-NEXT:    retq
 vector.ph:
   %lhs = zext <8 x i16> %x to <8 x i32>
   %cond = icmp ult <8 x i32> %y, %lhs
   %sub = sub <8 x i32> %lhs, %y
   %truncsub = trunc <8 x i32> %sub to <8 x i16>
   %res = select <8 x i1> %cond, <8 x i16> %truncsub, <8 x i16> zeroinitializer
vector.ph:
   %sub = sub <8 x i32> %max, %y
   %res = trunc <8 x i32> %sub to <8 x i16>
   ret <8 x i16> %res
 }
 
 define <8 x i16> @psubus_i16_i32_min(<8 x i16> %x, <8 x i32> %y) nounwind {
 ; SSE2-LABEL: psubus_i16_i32_min:
 ; SSE2:       # %bb.0: # %vector.ph
-; SSE2-NEXT:    pxor %xmm4, %xmm4
-; SSE2-NEXT:    movdqa %xmm0, %xmm3
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1],xmm3[2],xmm4[2],xmm3[3],xmm4[3]
-; SSE2-NEXT:    punpckhwd {{.*#+}} xmm0 = xmm0[4],xmm4[4],xmm0[5],xmm4[5],xmm0[6],xmm4[6],xmm0[7],xmm4[7]
-; SSE2-NEXT:    movdqa {{.*#+}} xmm4 = [2147483648,2147483648,2147483648,2147483648]
-; SSE2-NEXT:    movdqa %xmm2, %xmm5
-; SSE2-NEXT:    pxor %xmm4, %xmm5
-; SSE2-NEXT:    movdqa %xmm0, %xmm6
+; SSE2-NEXT:    pxor %xmm3, %xmm3
+; SSE2-NEXT:    movdqa %xmm0, %xmm4
+; SSE2-NEXT:    punpckhwd {{.*#+}} xmm4 = xmm4[4],xmm3[4],xmm4[5],xmm3[5],xmm4[6],xmm3[6],xmm4[7],xmm3[7]
+; SSE2-NEXT:    movdqa %xmm0, %xmm5
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm5 = xmm5[0],xmm3[0],xmm5[1],xmm3[1],xmm5[2],xmm3[2],xmm5[3],xmm3[3]
+; SSE2-NEXT:    movdqa {{.*#+}} xmm6 = [2147483648,2147483648,2147483648,2147483648]
+; SSE2-NEXT:    movdqa %xmm1, %xmm3
+; SSE2-NEXT:    pxor %xmm6, %xmm3
+; SSE2-NEXT:    movdqa %xmm5, %xmm7
+; SSE2-NEXT:    por %xmm6, %xmm7
+; SSE2-NEXT:    pcmpgtd %xmm7, %xmm3
+; SSE2-NEXT:    pand %xmm3, %xmm5
+; SSE2-NEXT:    pandn %xmm1, %xmm3
+; SSE2-NEXT:    por %xmm5, %xmm3
+; SSE2-NEXT:    movdqa %xmm2, %xmm1
+; SSE2-NEXT:    pxor %xmm6, %xmm1
 ; SSE2-NEXT:    por %xmm4, %xmm6
-; SSE2-NEXT:    pcmpgtd %xmm6, %xmm5
-; SSE2-NEXT:    movdqa %xmm0, %xmm6
-; SSE2-NEXT:    pand %xmm5, %xmm6
-; SSE2-NEXT:    pandn %xmm2, %xmm5
-; SSE2-NEXT:    por %xmm6, %xmm5
-; SSE2-NEXT:    movdqa %xmm1, %xmm2
-; SSE2-NEXT:    pxor %xmm4, %xmm2
-; SSE2-NEXT:    por %xmm3, %xmm4
-; SSE2-NEXT:    pcmpgtd %xmm4, %xmm2
-; SSE2-NEXT:    movdqa %xmm3, %xmm4
-; SSE2-NEXT:    pand %xmm2, %xmm4
-; SSE2-NEXT:    pandn %xmm1, %xmm2
-; SSE2-NEXT:    por %xmm4, %xmm2
-; SSE2-NEXT:    psubd %xmm2, %xmm3
-; SSE2-NEXT:    psubd %xmm5, %xmm0
-; SSE2-NEXT:    pslld $16, %xmm0
-; SSE2-NEXT:    psrad $16, %xmm0
+; SSE2-NEXT:    pcmpgtd %xmm6, %xmm1
+; SSE2-NEXT:    pand %xmm1, %xmm4
+; SSE2-NEXT:    pandn %xmm2, %xmm1
+; SSE2-NEXT:    por %xmm4, %xmm1
+; SSE2-NEXT:    pslld $16, %xmm1
+; SSE2-NEXT:    psrad $16, %xmm1
 ; SSE2-NEXT:    pslld $16, %xmm3
 ; SSE2-NEXT:    psrad $16, %xmm3
-; SSE2-NEXT:    packssdw %xmm0, %xmm3
-; SSE2-NEXT:    movdqa %xmm3, %xmm0
+; SSE2-NEXT:    packssdw %xmm1, %xmm3
+; SSE2-NEXT:    psubw %xmm3, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: psubus_i16_i32_min:
 ; SSSE3:       # %bb.0: # %vector.ph
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm4 = [2147483648,2147483648,2147483648,2147483648]
-; SSSE3-NEXT:    movdqa %xmm2, %xmm5
-; SSSE3-NEXT:    pxor %xmm4, %xmm5
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm6 = [2147549183,2147549183,2147549183,2147549183]
-; SSSE3-NEXT:    movdqa %xmm6, %xmm7
-; SSSE3-NEXT:    pcmpgtd %xmm5, %xmm7
-; SSSE3-NEXT:    movdqa {{.*#+}} xmm5 = [65535,65535,65535,65535]
-; SSSE3-NEXT:    pand %xmm7, %xmm2
-; SSSE3-NEXT:    pandn %xmm5, %xmm7
-; SSSE3-NEXT:    por %xmm2, %xmm7
-; SSSE3-NEXT:    pshufb %xmm3, %xmm7
-; SSSE3-NEXT:    pxor %xmm1, %xmm4
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm3 = [2147483648,2147483648,2147483648,2147483648]
+; SSSE3-NEXT:    movdqa %xmm2, %xmm4
+; SSSE3-NEXT:    pxor %xmm3, %xmm4
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm5 = [2147549183,2147549183,2147549183,2147549183]
+; SSSE3-NEXT:    movdqa %xmm5, %xmm6
 ; SSSE3-NEXT:    pcmpgtd %xmm4, %xmm6
-; SSSE3-NEXT:    pand %xmm6, %xmm1
-; SSSE3-NEXT:    pandn %xmm5, %xmm6
-; SSSE3-NEXT:    por %xmm1, %xmm6
-; SSSE3-NEXT:    pshufb %xmm3, %xmm6
-; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm6 = xmm6[0],xmm7[0]
-; SSSE3-NEXT:    psubusw %xmm6, %xmm0
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm4 = [65535,65535,65535,65535]
+; SSSE3-NEXT:    pand %xmm6, %xmm2
+; SSSE3-NEXT:    pandn %xmm4, %xmm6
+; SSSE3-NEXT:    por %xmm2, %xmm6
+; SSSE3-NEXT:    movdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; SSSE3-NEXT:    pshufb %xmm2, %xmm6
+; SSSE3-NEXT:    pxor %xmm1, %xmm3
+; SSSE3-NEXT:    pcmpgtd %xmm3, %xmm5
+; SSSE3-NEXT:    pand %xmm5, %xmm1
+; SSSE3-NEXT:    pandn %xmm4, %xmm5
+; SSSE3-NEXT:    por %xmm1, %xmm5
+; SSSE3-NEXT:    pshufb %xmm2, %xmm5
+; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm5 = xmm5[0],xmm6[0]
+; SSSE3-NEXT:    psubusw %xmm5, %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: psubus_i16_i32_min:
 ; SSE41:       # %bb.0: # %vector.ph
 ; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [65535,65535,65535,65535]
 ; SSE41-NEXT:    pminud %xmm3, %xmm2
 ; SSE41-NEXT:    pminud %xmm3, %xmm1
 ; SSE41-NEXT:    packusdw %xmm2, %xmm1