From 400c32ecb906b3753176623509cf18ac28c5e09d Mon Sep 17 00:00:00 2001
From: Craig Topper <craig.topper@intel.com>
Date: Mon, 27 Nov 2017 05:52:54 +0000
Subject: [PATCH] [SelectionDAG] Teach SplitVecRes_SETCC to call GetSplitVector
 if the operands have already been split.

llvm-svn: 319010
---
 .../CodeGen/SelectionDAG/LegalizeVectorTypes.cpp   |  15 +-
 llvm/test/CodeGen/X86/vector-compare-results.ll    | 676 ++++++++++-----------
 2 files changed, 337 insertions(+), 354 deletions(-)

diff --git a/llvm/lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp b/llvm/lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
index f15a310b..cdc83cc 100644
--- a/llvm/lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
+++ b/llvm/lib/CodeGen/SelectionDAG/LegalizeVectorTypes.cpp
@@ -1269,10 +1269,19 @@ void DAGTypeLegalizer::SplitVecRes_SETCC(SDNode *N, SDValue &Lo, SDValue &Hi) {
   SDLoc DL(N);
   std::tie(LoVT, HiVT) = DAG.GetSplitDestVTs(N->getValueType(0));
 
-  // Split the input.
+  // If the input also splits, handle it directly. Otherwise split it by hand.
   SDValue LL, LH, RL, RH;
-  std::tie(LL, LH) = DAG.SplitVectorOperand(N, 0);
-  std::tie(RL, RH) = DAG.SplitVectorOperand(N, 1);
+  if (getTypeAction(N->getOperand(0).getValueType()) ==
+      TargetLowering::TypeSplitVector)
+    GetSplitVector(N->getOperand(0), LL, LH);
+  else
+    std::tie(LL, LH) = DAG.SplitVectorOperand(N, 0);
+
+  if (getTypeAction(N->getOperand(1).getValueType()) ==
+      TargetLowering::TypeSplitVector)
+    GetSplitVector(N->getOperand(1), RL, RH);
+  else
+    std::tie(RL, RH) = DAG.SplitVectorOperand(N, 1);
 
   Lo = DAG.getNode(N->getOpcode(), DL, LoVT, LL, RL, N->getOperand(2));
   Hi = DAG.getNode(N->getOpcode(), DL, HiVT, LH, RH, N->getOperand(2));
diff --git a/llvm/test/CodeGen/X86/vector-compare-results.ll b/llvm/test/CodeGen/X86/vector-compare-results.ll
index 3370bfc..3b085d4 100644
--- a/llvm/test/CodeGen/X86/vector-compare-results.ll
+++ b/llvm/test/CodeGen/X86/vector-compare-results.ll
@@ -7856,12 +7856,9 @@ define <32 x i1> @test_cmp_v32f64(<32 x double> %a0, <32 x double> %a1) nounwind
 ; SSE2-NEXT:    cmpltpd %xmm7, %xmm4
 ; SSE2-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm7
 ; SSE2-NEXT:    cmpltpd %xmm6, %xmm7
-; SSE2-NEXT:    shufps {{.*#+}} xmm4 = xmm4[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[0,1,0,2,4,5,6,7]
-; SSE2-NEXT:    shufps {{.*#+}} xmm7 = xmm7[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm6 = xmm7[0,1,0,2,4,5,6,7]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm6 = xmm6[0],xmm4[0],xmm6[1],xmm4[1]
-; SSE2-NEXT:    movsd {{.*#+}} xmm6 = xmm5[0],xmm6[1]
+; SSE2-NEXT:    shufps {{.*#+}} xmm7 = xmm7[0,2],xmm4[0,2]
+; SSE2-NEXT:    packssdw %xmm7, %xmm7
+; SSE2-NEXT:    movsd {{.*#+}} xmm7 = xmm5[0],xmm7[1]
 ; SSE2-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm4
 ; SSE2-NEXT:    cmpltpd %xmm1, %xmm4
 ; SSE2-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm1
@@ -7871,20 +7868,17 @@ define <32 x i1> @test_cmp_v32f64(<32 x double> %a0, <32 x double> %a1) nounwind
 ; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,2,3]
 ; SSE2-NEXT:    pshuflw {{.*#+}} xmm4 = xmm1[0,2,2,3,4,5,6,7]
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
+; SSE2-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm5
+; SSE2-NEXT:    cmpltpd %xmm3, %xmm5
 ; SSE2-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm0
-; SSE2-NEXT:    cmpltpd %xmm3, %xmm0
-; SSE2-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm3
-; SSE2-NEXT:    cmpltpd %xmm2, %xmm3
+; SSE2-NEXT:    cmpltpd %xmm2, %xmm0
 ; SSE2-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm1
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm2 = xmm0[0,1,0,2,4,5,6,7]
-; SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm3[0,1,0,2,4,5,6,7]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm5[0,2]
 ; SSE2-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm2
+; SSE2-NEXT:    packssdw %xmm0, %xmm0
 ; SSE2-NEXT:    movsd {{.*#+}} xmm0 = xmm4[0],xmm0[1]
 ; SSE2-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm3
-; SSE2-NEXT:    packsswb %xmm6, %xmm0
+; SSE2-NEXT:    packsswb %xmm7, %xmm0
 ; SSE2-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm4
 ; SSE2-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm4
 ; SSE2-NEXT:    shufps {{.*#+}} xmm4 = xmm4[0,2,2,3]
@@ -7895,13 +7889,10 @@ define <32 x i1> @test_cmp_v32f64(<32 x double> %a0, <32 x double> %a1) nounwind
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1]
 ; SSE2-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm4
 ; SSE2-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm4
-; SSE2-NEXT:    shufps {{.*#+}} xmm4 = xmm4[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[0,1,0,2,4,5,6,7]
 ; SSE2-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm2
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,1,0,2,4,5,6,7]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1]
+; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm4[0,2]
 ; SSE2-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm4
+; SSE2-NEXT:    packssdw %xmm2, %xmm2
 ; SSE2-NEXT:    movsd {{.*#+}} xmm2 = xmm3[0],xmm2[1]
 ; SSE2-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm3
 ; SSE2-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm3
@@ -7913,12 +7904,9 @@ define <32 x i1> @test_cmp_v32f64(<32 x double> %a0, <32 x double> %a1) nounwind
 ; SSE2-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
 ; SSE2-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm3
 ; SSE2-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm3
-; SSE2-NEXT:    shufps {{.*#+}} xmm3 = xmm3[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[0,1,0,2,4,5,6,7]
 ; SSE2-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm1
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,1,0,2,4,5,6,7]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm3[0,2]
+; SSE2-NEXT:    packssdw %xmm1, %xmm1
 ; SSE2-NEXT:    movsd {{.*#+}} xmm1 = xmm4[0],xmm1[1]
 ; SSE2-NEXT:    packsswb %xmm2, %xmm1
 ; SSE2-NEXT:    movdqa %xmm1, -{{[0-9]+}}(%rsp)
@@ -8031,82 +8019,70 @@ define <32 x i1> @test_cmp_v32f64(<32 x double> %a0, <32 x double> %a1) nounwind
 ; SSE42-NEXT:    pushq %r12
 ; SSE42-NEXT:    pushq %rbx
 ; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm8
-; SSE42-NEXT:    cmpltpd %xmm7, %xmm8
-; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm7
-; SSE42-NEXT:    cmpltpd %xmm6, %xmm7
-; SSE42-NEXT:    shufps {{.*#+}} xmm8 = xmm8[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm6 = xmm8[0,1,0,2,4,5,6,7]
-; SSE42-NEXT:    shufps {{.*#+}} xmm7 = xmm7[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm7 = xmm7[0,1,0,2,4,5,6,7]
-; SSE42-NEXT:    punpckldq {{.*#+}} xmm7 = xmm7[0],xmm6[0],xmm7[1],xmm6[1]
-; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm6
-; SSE42-NEXT:    cmpltpd %xmm5, %xmm6
+; SSE42-NEXT:    cmpltpd %xmm5, %xmm8
 ; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm5
 ; SSE42-NEXT:    cmpltpd %xmm4, %xmm5
-; SSE42-NEXT:    shufps {{.*#+}} xmm6 = xmm6[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm4 = xmm6[0,2,2,3,4,5,6,7]
+; SSE42-NEXT:    shufps {{.*#+}} xmm8 = xmm8[0,2,2,3]
+; SSE42-NEXT:    pshuflw {{.*#+}} xmm4 = xmm8[0,2,2,3,4,5,6,7]
 ; SSE42-NEXT:    shufps {{.*#+}} xmm5 = xmm5[0,2,2,3]
 ; SSE42-NEXT:    pshuflw {{.*#+}} xmm5 = xmm5[0,2,2,3,4,5,6,7]
 ; SSE42-NEXT:    punpckldq {{.*#+}} xmm5 = xmm5[0],xmm4[0],xmm5[1],xmm4[1]
-; SSE42-NEXT:    pblendw {{.*#+}} xmm5 = xmm5[0,1,2,3],xmm7[4,5,6,7]
 ; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm4
-; SSE42-NEXT:    cmpltpd %xmm3, %xmm4
-; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm3
-; SSE42-NEXT:    cmpltpd %xmm2, %xmm3
-; SSE42-NEXT:    shufps {{.*#+}} xmm4 = xmm4[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm2 = xmm4[0,1,0,2,4,5,6,7]
-; SSE42-NEXT:    shufps {{.*#+}} xmm3 = xmm3[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[0,1,0,2,4,5,6,7]
-; SSE42-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm2[0],xmm3[1],xmm2[1]
-; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm2
-; SSE42-NEXT:    cmpltpd %xmm1, %xmm2
+; SSE42-NEXT:    cmpltpd %xmm7, %xmm4
+; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm7
+; SSE42-NEXT:    cmpltpd %xmm6, %xmm7
+; SSE42-NEXT:    shufps {{.*#+}} xmm7 = xmm7[0,2],xmm4[0,2]
+; SSE42-NEXT:    packssdw %xmm7, %xmm7
+; SSE42-NEXT:    pblendw {{.*#+}} xmm7 = xmm5[0,1,2,3],xmm7[4,5,6,7]
 ; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm4
-; SSE42-NEXT:    cmpltpd %xmm0, %xmm4
+; SSE42-NEXT:    cmpltpd %xmm1, %xmm4
 ; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm1
-; SSE42-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,2,2,3,4,5,6,7]
+; SSE42-NEXT:    cmpltpd %xmm0, %xmm1
 ; SSE42-NEXT:    shufps {{.*#+}} xmm4 = xmm4[0,2,2,3]
 ; SSE42-NEXT:    pshuflw {{.*#+}} xmm0 = xmm4[0,2,2,3,4,5,6,7]
-; SSE42-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1]
+; SSE42-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,2,3]
+; SSE42-NEXT:    pshuflw {{.*#+}} xmm4 = xmm1[0,2,2,3,4,5,6,7]
+; SSE42-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm0[0],xmm4[1],xmm0[1]
+; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm5
+; SSE42-NEXT:    cmpltpd %xmm3, %xmm5
+; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm0
+; SSE42-NEXT:    cmpltpd %xmm2, %xmm0
+; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm1
+; SSE42-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm5[0,2]
 ; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm2
-; SSE42-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm3[4,5,6,7]
+; SSE42-NEXT:    packssdw %xmm0, %xmm0
+; SSE42-NEXT:    pblendw {{.*#+}} xmm0 = xmm4[0,1,2,3],xmm0[4,5,6,7]
 ; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm3
-; SSE42-NEXT:    packsswb %xmm5, %xmm0
+; SSE42-NEXT:    packsswb %xmm7, %xmm0
 ; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm4
 ; SSE42-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm4
 ; SSE42-NEXT:    shufps {{.*#+}} xmm4 = xmm4[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[0,1,0,2,4,5,6,7]
+; SSE42-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[0,2,2,3,4,5,6,7]
 ; SSE42-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm3
 ; SSE42-NEXT:    shufps {{.*#+}} xmm3 = xmm3[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[0,1,0,2,4,5,6,7]
+; SSE42-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[0,2,2,3,4,5,6,7]
 ; SSE42-NEXT:    punpckldq {{.*#+}} xmm3 = xmm3[0],xmm4[0],xmm3[1],xmm4[1]
 ; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm4
 ; SSE42-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm4
-; SSE42-NEXT:    shufps {{.*#+}} xmm4 = xmm4[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[0,2,2,3,4,5,6,7]
 ; SSE42-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm2
-; SSE42-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,2,2,3,4,5,6,7]
-; SSE42-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1]
+; SSE42-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm4[0,2]
 ; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm4
-; SSE42-NEXT:    pblendw {{.*#+}} xmm2 = xmm2[0,1,2,3],xmm3[4,5,6,7]
+; SSE42-NEXT:    packssdw %xmm2, %xmm2
+; SSE42-NEXT:    pblendw {{.*#+}} xmm2 = xmm3[0,1,2,3],xmm2[4,5,6,7]
 ; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm3
 ; SSE42-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm3
 ; SSE42-NEXT:    shufps {{.*#+}} xmm3 = xmm3[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[0,1,0,2,4,5,6,7]
+; SSE42-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[0,2,2,3,4,5,6,7]
 ; SSE42-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm4
 ; SSE42-NEXT:    shufps {{.*#+}} xmm4 = xmm4[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[0,1,0,2,4,5,6,7]
+; SSE42-NEXT:    pshuflw {{.*#+}} xmm4 = xmm4[0,2,2,3,4,5,6,7]
 ; SSE42-NEXT:    punpckldq {{.*#+}} xmm4 = xmm4[0],xmm3[0],xmm4[1],xmm3[1]
 ; SSE42-NEXT:    movapd {{[0-9]+}}(%rsp), %xmm3
 ; SSE42-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm3
-; SSE42-NEXT:    shufps {{.*#+}} xmm3 = xmm3[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[0,2,2,3,4,5,6,7]
 ; SSE42-NEXT:    cmpltpd {{[0-9]+}}(%rsp), %xmm1
-; SSE42-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
-; SSE42-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1]
-; SSE42-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1,2,3],xmm4[4,5,6,7]
+; SSE42-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm3[0,2]
+; SSE42-NEXT:    packssdw %xmm1, %xmm1
+; SSE42-NEXT:    pblendw {{.*#+}} xmm1 = xmm4[0,1,2,3],xmm1[4,5,6,7]
 ; SSE42-NEXT:    packsswb %xmm2, %xmm1
 ; SSE42-NEXT:    pextrb $15, %xmm1, %eax
 ; SSE42-NEXT:    andb $1, %al
@@ -8988,349 +8964,347 @@ define <32 x i1> @test_cmp_v32i64(<32 x i64> %a0, <32 x i64> %a1) nounwind {
 ; SSE2-LABEL: test_cmp_v32i64:
 ; SSE2:       # BB#0:
 ; SSE2-NEXT:    movdqa {{.*#+}} xmm8 = [2147483648,0,2147483648,0]
+; SSE2-NEXT:    pxor %xmm8, %xmm5
 ; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm9
 ; SSE2-NEXT:    pxor %xmm8, %xmm9
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm10
-; SSE2-NEXT:    pxor %xmm8, %xmm10
-; SSE2-NEXT:    movdqa %xmm10, %xmm11
-; SSE2-NEXT:    pcmpgtd %xmm9, %xmm11
-; SSE2-NEXT:    pshufd {{.*#+}} xmm12 = xmm11[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm9, %xmm10
-; SSE2-NEXT:    pshufd {{.*#+}} xmm10 = xmm10[1,1,3,3]
-; SSE2-NEXT:    pand %xmm12, %xmm10
-; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm11[1,1,3,3]
-; SSE2-NEXT:    por %xmm10, %xmm9
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm10
-; SSE2-NEXT:    pxor %xmm8, %xmm10
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm11
-; SSE2-NEXT:    pxor %xmm8, %xmm11
-; SSE2-NEXT:    movdqa %xmm11, %xmm12
-; SSE2-NEXT:    pcmpgtd %xmm10, %xmm12
-; SSE2-NEXT:    pshufd {{.*#+}} xmm13 = xmm12[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm10, %xmm11
-; SSE2-NEXT:    pshufd {{.*#+}} xmm11 = xmm11[1,1,3,3]
-; SSE2-NEXT:    pand %xmm13, %xmm11
-; SSE2-NEXT:    pshufd {{.*#+}} xmm10 = xmm12[1,1,3,3]
-; SSE2-NEXT:    por %xmm11, %xmm10
-; SSE2-NEXT:    pshufd {{.*#+}} xmm11 = xmm10[2,3,0,1]
-; SSE2-NEXT:    shufps {{.*#+}} xmm10 = xmm10[0,2],xmm9[0,2]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm10 = xmm10[0],xmm11[0],xmm10[1],xmm11[1],xmm10[2],xmm11[2],xmm10[3],xmm11[3]
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm11
-; SSE2-NEXT:    pxor %xmm8, %xmm11
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm12
-; SSE2-NEXT:    pxor %xmm8, %xmm12
-; SSE2-NEXT:    movdqa %xmm12, %xmm13
-; SSE2-NEXT:    pcmpgtd %xmm11, %xmm13
-; SSE2-NEXT:    pshufd {{.*#+}} xmm14 = xmm13[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm11, %xmm12
-; SSE2-NEXT:    pshufd {{.*#+}} xmm11 = xmm12[1,1,3,3]
-; SSE2-NEXT:    pand %xmm14, %xmm11
-; SSE2-NEXT:    pshufd {{.*#+}} xmm12 = xmm13[1,1,3,3]
-; SSE2-NEXT:    por %xmm11, %xmm12
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm11
-; SSE2-NEXT:    pxor %xmm8, %xmm11
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm13
-; SSE2-NEXT:    pxor %xmm8, %xmm13
-; SSE2-NEXT:    movdqa %xmm13, %xmm14
-; SSE2-NEXT:    pcmpgtd %xmm11, %xmm14
-; SSE2-NEXT:    pshufd {{.*#+}} xmm15 = xmm14[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm11, %xmm13
-; SSE2-NEXT:    pshufd {{.*#+}} xmm11 = xmm13[1,1,3,3]
-; SSE2-NEXT:    pand %xmm15, %xmm11
-; SSE2-NEXT:    pshufd {{.*#+}} xmm13 = xmm14[1,1,3,3]
-; SSE2-NEXT:    por %xmm11, %xmm13
-; SSE2-NEXT:    shufps {{.*#+}} xmm13 = xmm13[0,2],xmm12[0,2]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm11 = xmm13[0,2,2,3,4,5,6,7]
-; SSE2-NEXT:    pshufhw {{.*#+}} xmm11 = xmm11[0,1,2,3,4,6,6,7]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm11 = xmm11[0,1,0,2]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm9[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm9 = xmm9[0,1,0,2,4,5,6,7]
-; SSE2-NEXT:    movsd {{.*#+}} xmm11 = xmm9[0],xmm11[1]
-; SSE2-NEXT:    movss {{.*#+}} xmm11 = xmm10[0],xmm11[1,2,3]
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm9
-; SSE2-NEXT:    pxor %xmm8, %xmm9
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm10
-; SSE2-NEXT:    pxor %xmm8, %xmm10
-; SSE2-NEXT:    movdqa %xmm10, %xmm12
-; SSE2-NEXT:    pcmpgtd %xmm9, %xmm12
-; SSE2-NEXT:    pshufd {{.*#+}} xmm13 = xmm12[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm9, %xmm10
-; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm10[1,1,3,3]
-; SSE2-NEXT:    pand %xmm13, %xmm9
-; SSE2-NEXT:    pshufd {{.*#+}} xmm10 = xmm12[1,1,3,3]
-; SSE2-NEXT:    por %xmm9, %xmm10
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm9
-; SSE2-NEXT:    pxor %xmm8, %xmm9
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm12
-; SSE2-NEXT:    pxor %xmm8, %xmm12
-; SSE2-NEXT:    movdqa %xmm12, %xmm13
-; SSE2-NEXT:    pcmpgtd %xmm9, %xmm13
-; SSE2-NEXT:    pshufd {{.*#+}} xmm14 = xmm13[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm9, %xmm12
-; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm12[1,1,3,3]
-; SSE2-NEXT:    pand %xmm14, %xmm9
-; SSE2-NEXT:    pshufd {{.*#+}} xmm12 = xmm13[1,1,3,3]
-; SSE2-NEXT:    por %xmm9, %xmm12
-; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm12[2,3,0,1]
-; SSE2-NEXT:    shufps {{.*#+}} xmm12 = xmm12[0,2],xmm10[0,2]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm12 = xmm12[0],xmm9[0],xmm12[1],xmm9[1],xmm12[2],xmm9[2],xmm12[3],xmm9[3]
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm9
-; SSE2-NEXT:    pxor %xmm8, %xmm9
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm13
-; SSE2-NEXT:    pxor %xmm8, %xmm13
-; SSE2-NEXT:    movdqa %xmm13, %xmm14
-; SSE2-NEXT:    pcmpgtd %xmm9, %xmm14
-; SSE2-NEXT:    pshufd {{.*#+}} xmm15 = xmm14[0,0,2,2]
-; SSE2-NEXT:    pcmpeqd %xmm9, %xmm13
-; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm13[1,1,3,3]
-; SSE2-NEXT:    pand %xmm15, %xmm9
-; SSE2-NEXT:    pshufd {{.*#+}} xmm13 = xmm14[1,1,3,3]
-; SSE2-NEXT:    por %xmm9, %xmm13
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm9
-; SSE2-NEXT:    pxor %xmm8, %xmm9
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm14
-; SSE2-NEXT:    pxor %xmm8, %xmm14
-; SSE2-NEXT:    movdqa %xmm14, %xmm15
-; SSE2-NEXT:    pcmpgtd %xmm9, %xmm15
-; SSE2-NEXT:    pcmpeqd %xmm9, %xmm14
-; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm15[0,0,2,2]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm14 = xmm14[1,1,3,3]
-; SSE2-NEXT:    pand %xmm9, %xmm14
-; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm15[1,1,3,3]
-; SSE2-NEXT:    por %xmm14, %xmm9
-; SSE2-NEXT:    shufps {{.*#+}} xmm9 = xmm9[0,2],xmm13[0,2]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm9 = xmm9[0,2,2,3,4,5,6,7]
-; SSE2-NEXT:    pshufhw {{.*#+}} xmm9 = xmm9[0,1,2,3,4,6,6,7]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm9[0,1,0,2]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm10 = xmm10[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm10 = xmm10[0,1,0,2,4,5,6,7]
-; SSE2-NEXT:    movsd {{.*#+}} xmm9 = xmm10[0],xmm9[1]
-; SSE2-NEXT:    movaps {{.*#+}} xmm10 = [255,255,255,255,255,255,255,255]
-; SSE2-NEXT:    andps %xmm10, %xmm11
-; SSE2-NEXT:    movss {{.*#+}} xmm9 = xmm12[0],xmm9[1,2,3]
-; SSE2-NEXT:    andps %xmm10, %xmm9
-; SSE2-NEXT:    packuswb %xmm11, %xmm9
-; SSE2-NEXT:    pxor %xmm8, %xmm5
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm11
-; SSE2-NEXT:    pxor %xmm8, %xmm11
-; SSE2-NEXT:    movdqa %xmm5, %xmm12
-; SSE2-NEXT:    pcmpgtd %xmm11, %xmm12
-; SSE2-NEXT:    pcmpeqd %xmm5, %xmm11
-; SSE2-NEXT:    pshufd {{.*#+}} xmm13 = xmm12[0,0,2,2]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm11[1,1,3,3]
-; SSE2-NEXT:    pand %xmm13, %xmm5
-; SSE2-NEXT:    pshufd {{.*#+}} xmm11 = xmm12[1,1,3,3]
-; SSE2-NEXT:    por %xmm5, %xmm11
+; SSE2-NEXT:    movdqa %xmm5, %xmm10
+; SSE2-NEXT:    pcmpgtd %xmm9, %xmm10
+; SSE2-NEXT:    pshufd {{.*#+}} xmm11 = xmm10[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm5, %xmm9
+; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm9[1,1,3,3]
+; SSE2-NEXT:    pand %xmm11, %xmm9
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm10[1,1,3,3]
+; SSE2-NEXT:    por %xmm9, %xmm5
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[0,2,2,3]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm9 = xmm5[0,2,2,3,4,5,6,7]
 ; SSE2-NEXT:    pxor %xmm8, %xmm4
 ; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm5
 ; SSE2-NEXT:    pxor %xmm8, %xmm5
-; SSE2-NEXT:    movdqa %xmm4, %xmm12
-; SSE2-NEXT:    pcmpgtd %xmm5, %xmm12
+; SSE2-NEXT:    movdqa %xmm4, %xmm10
+; SSE2-NEXT:    pcmpgtd %xmm5, %xmm10
+; SSE2-NEXT:    pshufd {{.*#+}} xmm11 = xmm10[0,0,2,2]
 ; SSE2-NEXT:    pcmpeqd %xmm4, %xmm5
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm12[0,0,2,2]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3]
-; SSE2-NEXT:    pand %xmm4, %xmm5
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm11[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm11 = xmm4[0,2,2,3,4,5,6,7]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm12[1,1,3,3]
-; SSE2-NEXT:    por %xmm5, %xmm4
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm12 = xmm4[0,2,2,3,4,5,6,7]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm12 = xmm12[0],xmm11[0],xmm12[1],xmm11[1]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm5[1,1,3,3]
+; SSE2-NEXT:    pand %xmm11, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm10[1,1,3,3]
+; SSE2-NEXT:    por %xmm4, %xmm5
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm5[0,2,2,3]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm10 = xmm4[0,2,2,3,4,5,6,7]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm10 = xmm10[0],xmm9[0],xmm10[1],xmm9[1]
 ; SSE2-NEXT:    pxor %xmm8, %xmm7
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm4
-; SSE2-NEXT:    pxor %xmm8, %xmm4
-; SSE2-NEXT:    movdqa %xmm7, %xmm5
-; SSE2-NEXT:    pcmpgtd %xmm4, %xmm5
-; SSE2-NEXT:    pcmpeqd %xmm7, %xmm4
-; SSE2-NEXT:    pshufd {{.*#+}} xmm7 = xmm5[0,0,2,2]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3]
-; SSE2-NEXT:    pand %xmm7, %xmm4
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm5
+; SSE2-NEXT:    pxor %xmm8, %xmm5
+; SSE2-NEXT:    movdqa %xmm7, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm5, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm7, %xmm5
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3]
-; SSE2-NEXT:    por %xmm4, %xmm5
+; SSE2-NEXT:    pand %xmm9, %xmm5
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm5, %xmm4
 ; SSE2-NEXT:    pxor %xmm8, %xmm6
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm4
-; SSE2-NEXT:    pxor %xmm8, %xmm4
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm5
+; SSE2-NEXT:    pxor %xmm8, %xmm5
 ; SSE2-NEXT:    movdqa %xmm6, %xmm7
-; SSE2-NEXT:    pcmpgtd %xmm4, %xmm7
-; SSE2-NEXT:    pcmpeqd %xmm6, %xmm4
-; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm7[0,0,2,2]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[1,1,3,3]
-; SSE2-NEXT:    pand %xmm6, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm5, %xmm7
+; SSE2-NEXT:    pshufd {{.*#+}} xmm9 = xmm7[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm6, %xmm5
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3]
+; SSE2-NEXT:    pand %xmm9, %xmm5
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm7[1,1,3,3]
-; SSE2-NEXT:    por %xmm4, %xmm6
-; SSE2-NEXT:    shufps {{.*#+}} xmm6 = xmm6[0,2],xmm5[0,2]
+; SSE2-NEXT:    por %xmm5, %xmm6
+; SSE2-NEXT:    shufps {{.*#+}} xmm6 = xmm6[0,2],xmm4[0,2]
 ; SSE2-NEXT:    pshuflw {{.*#+}} xmm4 = xmm6[0,2,2,3,4,5,6,7]
 ; SSE2-NEXT:    pshufhw {{.*#+}} xmm4 = xmm4[0,1,2,3,4,6,6,7]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm4[0,1,0,2]
-; SSE2-NEXT:    movsd {{.*#+}} xmm4 = xmm12[0],xmm4[1]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,1,0,2]
+; SSE2-NEXT:    movsd {{.*#+}} xmm5 = xmm10[0],xmm5[1]
+; SSE2-NEXT:    movapd {{.*#+}} xmm9 = [255,255,255,255,255,255,255,255]
+; SSE2-NEXT:    andpd %xmm9, %xmm5
 ; SSE2-NEXT:    pxor %xmm8, %xmm1
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm5
-; SSE2-NEXT:    pxor %xmm8, %xmm5
-; SSE2-NEXT:    movdqa %xmm1, %xmm6
-; SSE2-NEXT:    pcmpgtd %xmm5, %xmm6
-; SSE2-NEXT:    pcmpeqd %xmm1, %xmm5
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm6[0,0,2,2]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3]
-; SSE2-NEXT:    pand %xmm1, %xmm5
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm6
+; SSE2-NEXT:    pxor %xmm8, %xmm6
+; SSE2-NEXT:    movdqa %xmm1, %xmm7
+; SSE2-NEXT:    pcmpgtd %xmm6, %xmm7
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm7[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm6
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm6[1,1,3,3]
-; SSE2-NEXT:    por %xmm5, %xmm1
+; SSE2-NEXT:    pand %xmm4, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm7[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm4[0,2,2,3]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm4 = xmm1[0,2,2,3,4,5,6,7]
 ; SSE2-NEXT:    pxor %xmm8, %xmm0
-; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm5
-; SSE2-NEXT:    pxor %xmm8, %xmm5
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm1
+; SSE2-NEXT:    pxor %xmm8, %xmm1
 ; SSE2-NEXT:    movdqa %xmm0, %xmm6
-; SSE2-NEXT:    pcmpgtd %xmm5, %xmm6
-; SSE2-NEXT:    pcmpeqd %xmm0, %xmm5
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm6[0,0,2,2]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3]
-; SSE2-NEXT:    pand %xmm0, %xmm5
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm6
+; SSE2-NEXT:    pshufd {{.*#+}} xmm7 = xmm6[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm0, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
+; SSE2-NEXT:    pand %xmm7, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm6[1,1,3,3]
+; SSE2-NEXT:    por %xmm0, %xmm1
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,2,2,3]
 ; SSE2-NEXT:    pshuflw {{.*#+}} xmm1 = xmm0[0,2,2,3,4,5,6,7]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm6[1,1,3,3]
-; SSE2-NEXT:    por %xmm5, %xmm0
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
-; SSE2-NEXT:    punpckldq {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm4[0],xmm1[1],xmm4[1]
 ; SSE2-NEXT:    pxor %xmm8, %xmm3
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm0
+; SSE2-NEXT:    pxor %xmm8, %xmm0
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm3, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pand %xmm6, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm0, %xmm3
+; SSE2-NEXT:    pxor %xmm8, %xmm2
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm0
+; SSE2-NEXT:    pxor %xmm8, %xmm0
+; SSE2-NEXT:    movdqa %xmm2, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm0, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; SSE2-NEXT:    pand %xmm6, %xmm0
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm0, %xmm2
+; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm3[0,2]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm2[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[0,1,0,2]
+; SSE2-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
+; SSE2-NEXT:    andpd %xmm9, %xmm0
+; SSE2-NEXT:    packuswb %xmm5, %xmm0
 ; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm1
 ; SSE2-NEXT:    pxor %xmm8, %xmm1
-; SSE2-NEXT:    movdqa %xmm3, %xmm5
-; SSE2-NEXT:    pcmpgtd %xmm1, %xmm5
-; SSE2-NEXT:    pcmpeqd %xmm3, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm5[0,0,2,2]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; SSE2-NEXT:    pand %xmm3, %xmm1
-; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm5[1,1,3,3]
-; SSE2-NEXT:    por %xmm1, %xmm3
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm2
 ; SSE2-NEXT:    pxor %xmm8, %xmm2
-; SSE2-NEXT:    pxor {{[0-9]+}}(%rsp), %xmm8
-; SSE2-NEXT:    movdqa %xmm2, %xmm1
-; SSE2-NEXT:    pcmpgtd %xmm8, %xmm1
-; SSE2-NEXT:    pcmpeqd %xmm2, %xmm8
-; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[0,0,2,2]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm8[1,1,3,3]
-; SSE2-NEXT:    pand %xmm2, %xmm5
-; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; SSE2-NEXT:    por %xmm5, %xmm1
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm3[0,2]
+; 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:    pshufd {{.*#+}} xmm1 = xmm2[0,2,2,3]
 ; SSE2-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm2
+; SSE2-NEXT:    pxor %xmm8, %xmm2
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm3
+; SSE2-NEXT:    pxor %xmm8, %xmm3
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[0,2,2,3]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,2,2,3,4,5,6,7]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1]
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm1
+; SSE2-NEXT:    pxor %xmm8, %xmm1
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm3
+; SSE2-NEXT:    pxor %xmm8, %xmm3
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm3[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 {{[0-9]+}}(%rsp), %xmm1
+; SSE2-NEXT:    pxor %xmm8, %xmm1
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm4
+; SSE2-NEXT:    pxor %xmm8, %xmm4
+; SSE2-NEXT:    movdqa %xmm4, %xmm5
+; SSE2-NEXT:    pcmpgtd %xmm1, %xmm5
+; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm1, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm4[1,1,3,3]
+; SSE2-NEXT:    pand %xmm6, %xmm1
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm5[1,1,3,3]
+; SSE2-NEXT:    por %xmm1, %xmm4
+; SSE2-NEXT:    shufps {{.*#+}} xmm4 = xmm4[0,2],xmm3[0,2]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm1 = xmm4[0,2,2,3,4,5,6,7]
 ; SSE2-NEXT:    pshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,6,6,7]
 ; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,0,2]
-; SSE2-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
-; SSE2-NEXT:    andpd %xmm10, %xmm4
-; SSE2-NEXT:    andpd %xmm10, %xmm1
-; SSE2-NEXT:    packuswb %xmm4, %xmm1
-; SSE2-NEXT:    movdqa %xmm1, -{{[0-9]+}}(%rsp)
+; SSE2-NEXT:    movsd {{.*#+}} xmm1 = xmm2[0],xmm1[1]
+; SSE2-NEXT:    andpd %xmm9, %xmm1
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm2
+; SSE2-NEXT:    pxor %xmm8, %xmm2
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm3
+; SSE2-NEXT:    pxor %xmm8, %xmm3
+; SSE2-NEXT:    movdqa %xmm3, %xmm4
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm4[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[1,1,3,3]
+; SSE2-NEXT:    pand %xmm5, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm3[0,2,2,3]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm3 = xmm2[0,2,2,3,4,5,6,7]
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm2
+; SSE2-NEXT:    pxor %xmm8, %xmm2
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm4
+; SSE2-NEXT:    pxor %xmm8, %xmm4
+; SSE2-NEXT:    movdqa %xmm4, %xmm5
+; SSE2-NEXT:    pcmpgtd %xmm2, %xmm5
+; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm4[1,1,3,3]
+; SSE2-NEXT:    pand %xmm6, %xmm2
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm5[1,1,3,3]
+; SSE2-NEXT:    por %xmm2, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm2 = xmm4[0,2,2,3]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm2 = xmm2[0,2,2,3,4,5,6,7]
+; SSE2-NEXT:    punpckldq {{.*#+}} xmm2 = xmm2[0],xmm3[0],xmm2[1],xmm3[1]
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm3
+; SSE2-NEXT:    pxor %xmm8, %xmm3
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm4
+; SSE2-NEXT:    pxor %xmm8, %xmm4
+; SSE2-NEXT:    movdqa %xmm4, %xmm5
+; SSE2-NEXT:    pcmpgtd %xmm3, %xmm5
+; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm3, %xmm4
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm4[1,1,3,3]
+; SSE2-NEXT:    pand %xmm6, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm4 = xmm5[1,1,3,3]
+; SSE2-NEXT:    por %xmm3, %xmm4
+; SSE2-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm3
+; SSE2-NEXT:    pxor %xmm8, %xmm3
+; SSE2-NEXT:    pxor {{[0-9]+}}(%rsp), %xmm8
+; SSE2-NEXT:    movdqa %xmm8, %xmm5
+; SSE2-NEXT:    pcmpgtd %xmm3, %xmm5
+; SSE2-NEXT:    pshufd {{.*#+}} xmm6 = xmm5[0,0,2,2]
+; SSE2-NEXT:    pcmpeqd %xmm3, %xmm8
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm8[1,1,3,3]
+; SSE2-NEXT:    pand %xmm6, %xmm3
+; SSE2-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3]
+; SSE2-NEXT:    por %xmm3, %xmm5
+; SSE2-NEXT:    shufps {{.*#+}} xmm5 = xmm5[0,2],xmm4[0,2]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm3 = xmm5[0,2,2,3,4,5,6,7]
+; SSE2-NEXT:    pshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,4,6,6,7]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[0,1,0,2]
+; SSE2-NEXT:    movsd {{.*#+}} xmm3 = xmm2[0],xmm3[1]
+; SSE2-NEXT:    andpd %xmm9, %xmm3
+; SSE2-NEXT:    packuswb %xmm1, %xmm3
+; SSE2-NEXT:    movdqa %xmm3, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
+; SSE2-NEXT:    movb %al, 2(%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
 ; SSE2-NEXT:    andb $1, %cl
-; SSE2-NEXT:    movb %cl, (%rdi)
+; SSE2-NEXT:    movb %cl, 2(%rdi)
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, (%rdi)
-; SSE2-NEXT:    movdqa %xmm9, -{{[0-9]+}}(%rsp)
+; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movdqa %xmm0, -{{[0-9]+}}(%rsp)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %al
 ; SSE2-NEXT:    movb -{{[0-9]+}}(%rsp), %cl
 ; SSE2-NEXT:    andb $1, %cl
-; SSE2-NEXT:    movb %cl, 2(%rdi)
+; SSE2-NEXT:    movb %cl, (%rdi)
 ; SSE2-NEXT:    andb $1, %al
-; SSE2-NEXT:    movb %al, 2(%rdi)
+; SSE2-NEXT:    movb %al, (%rdi)
 ; SSE2-NEXT:    movq %rdi, %rax
 ; SSE2-NEXT:    retq
 ;
 ; SSE42-LABEL: test_cmp_v32i64:
 ; SSE42:       # BB#0:
+; SSE42-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm8
 ; SSE42-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm10
-; SSE42-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm9
 ; SSE42-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm11
-; SSE42-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm8
-; SSE42-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm14
+; SSE42-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm12
+; SSE42-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm9
 ; SSE42-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm13
+; SSE42-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm14
 ; SSE42-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm15
-; SSE42-NEXT:    movdqa {{[0-9]+}}(%rsp), %xmm12
 ; SSE42-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm5
 ; SSE42-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[0,2,2,3]
 ; SSE42-NEXT:    pshuflw {{.*#+}} xmm5 = xmm5[0,2,2,3,4,5,6,7]
@@ -9356,77 +9330,77 @@ define <32 x i1> @test_cmp_v32i64(<32 x i64> %a0, <32 x i64> %a1) nounwind {
 ; SSE42-NEXT:    packssdw %xmm2, %xmm2
 ; SSE42-NEXT:    pblendw {{.*#+}} xmm2 = xmm0[0,1,2,3],xmm2[4,5,6,7]
 ; SSE42-NEXT:    packsswb %xmm6, %xmm2
-; SSE42-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm12
-; SSE42-NEXT:    movdqa {{.*#+}} xmm0 = [0,1,8,9,2,3,10,11,8,9,12,13,10,11,14,15]
-; SSE42-NEXT:    pshufb %xmm0, %xmm12
 ; SSE42-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm15
-; SSE42-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm13
-; SSE42-NEXT:    shufps {{.*#+}} xmm13 = xmm13[0,2],xmm15[0,2]
-; SSE42-NEXT:    movdqa {{.*#+}} xmm1 = [0,1,4,5,4,5,6,7,0,1,4,5,8,9,12,13]
-; SSE42-NEXT:    pshufb %xmm1, %xmm13
+; SSE42-NEXT:    pshufd {{.*#+}} xmm0 = xmm15[0,2,2,3]
+; SSE42-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
 ; SSE42-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm14
-; SSE42-NEXT:    pshufd {{.*#+}} xmm3 = xmm14[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm3 = xmm3[0,1,0,2,4,5,6,7]
-; SSE42-NEXT:    pblendw {{.*#+}} xmm3 = xmm3[0,1,2,3],xmm13[4,5,6,7]
-; SSE42-NEXT:    pblendw {{.*#+}} xmm3 = xmm12[0,1],xmm3[2,3,4,5,6,7]
-; SSE42-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm8
-; SSE42-NEXT:    pshufb %xmm0, %xmm8
-; SSE42-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm11
+; SSE42-NEXT:    pshufd {{.*#+}} xmm1 = xmm14[0,2,2,3]
+; SSE42-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
+; SSE42-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
+; SSE42-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm13
 ; SSE42-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm9
-; SSE42-NEXT:    shufps {{.*#+}} xmm9 = xmm9[0,2],xmm11[0,2]
-; SSE42-NEXT:    pshufb %xmm1, %xmm9
+; SSE42-NEXT:    shufps {{.*#+}} xmm9 = xmm9[0,2],xmm13[0,2]
+; SSE42-NEXT:    packssdw %xmm9, %xmm9
+; SSE42-NEXT:    pblendw {{.*#+}} xmm9 = xmm1[0,1,2,3],xmm9[4,5,6,7]
+; SSE42-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm12
+; SSE42-NEXT:    pshufd {{.*#+}} xmm0 = xmm12[0,2,2,3]
+; SSE42-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
+; SSE42-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm11
+; SSE42-NEXT:    pshufd {{.*#+}} xmm1 = xmm11[0,2,2,3]
+; SSE42-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,2,2,3,4,5,6,7]
+; SSE42-NEXT:    punpckldq {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1]
 ; SSE42-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm10
-; SSE42-NEXT:    pshufd {{.*#+}} xmm0 = xmm10[0,2,2,3]
-; SSE42-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,1,0,2,4,5,6,7]
-; SSE42-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm9[4,5,6,7]
-; SSE42-NEXT:    pblendw {{.*#+}} xmm0 = xmm8[0,1],xmm0[2,3,4,5,6,7]
-; SSE42-NEXT:    packsswb %xmm3, %xmm0
-; SSE42-NEXT:    pextrb $15, %xmm0, %eax
+; SSE42-NEXT:    pcmpgtq {{[0-9]+}}(%rsp), %xmm8
+; SSE42-NEXT:    shufps {{.*#+}} xmm8 = xmm8[0,2],xmm10[0,2]
+; SSE42-NEXT:    packssdw %xmm8, %xmm8
+; SSE42-NEXT:    pblendw {{.*#+}} xmm8 = xmm1[0,1,2,3],xmm8[4,5,6,7]
+; SSE42-NEXT:    packsswb %xmm9, %xmm8
+; SSE42-NEXT:    pextrb $15, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
-; SSE42-NEXT:    pextrb $14, %xmm0, %eax
+; SSE42-NEXT:    pextrb $14, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
-; SSE42-NEXT:    pextrb $13, %xmm0, %eax
+; SSE42-NEXT:    pextrb $13, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
-; SSE42-NEXT:    pextrb $12, %xmm0, %eax
+; SSE42-NEXT:    pextrb $12, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
-; SSE42-NEXT:    pextrb $11, %xmm0, %eax
+; SSE42-NEXT:    pextrb $11, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
-; SSE42-NEXT:    pextrb $10, %xmm0, %eax
+; SSE42-NEXT:    pextrb $10, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
-; SSE42-NEXT:    pextrb $9, %xmm0, %eax
+; SSE42-NEXT:    pextrb $9, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
-; SSE42-NEXT:    pextrb $8, %xmm0, %eax
+; SSE42-NEXT:    pextrb $8, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
-; SSE42-NEXT:    pextrb $7, %xmm0, %eax
+; SSE42-NEXT:    pextrb $7, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
-; SSE42-NEXT:    pextrb $6, %xmm0, %eax
+; SSE42-NEXT:    pextrb $6, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
-; SSE42-NEXT:    pextrb $5, %xmm0, %eax
+; SSE42-NEXT:    pextrb $5, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
-; SSE42-NEXT:    pextrb $4, %xmm0, %eax
+; SSE42-NEXT:    pextrb $4, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
-; SSE42-NEXT:    pextrb $3, %xmm0, %eax
+; SSE42-NEXT:    pextrb $3, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
-; SSE42-NEXT:    pextrb $2, %xmm0, %eax
+; SSE42-NEXT:    pextrb $2, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
-; SSE42-NEXT:    pextrb $1, %xmm0, %eax
+; SSE42-NEXT:    pextrb $1, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
-; SSE42-NEXT:    pextrb $0, %xmm0, %eax
+; SSE42-NEXT:    pextrb $0, %xmm8, %eax
 ; SSE42-NEXT:    andb $1, %al
 ; SSE42-NEXT:    movb %al, 2(%rdi)
 ; SSE42-NEXT:    pextrb $15, %xmm2, %eax
-- 
2.7.4