From 46d3e580ed07de0fdf021a12aafcdde12b52bef9 Mon Sep 17 00:00:00 2001
From: Chandler Carruth <chandlerc@gmail.com>
Date: Sun, 15 Feb 2015 10:12:02 +0000
Subject: [PATCH] [x86] When splitting 256-bit vectors into 128-bit vectors,
 don't extract subvectors from buildvectors. That doesn't really make any
 sense and it breaks all of the down-stream matching of buildvectors to
 cleverly lower shuffles.

With this, we now get the shift-based lowering of 256-bit vector
shuffles with AVX1 when we split them into 128-bit vectors. We also do
much better on the zero-extension patterns, although there remains quite
a bit of room for improvement here.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@229299 91177308-0d34-0410-b5e6-96231b3b80d8
---
 lib/Target/X86/X86ISelLowering.cpp         |  45 ++++++--
 test/CodeGen/X86/vec_cast2.ll              |  30 ++----
 test/CodeGen/X86/vector-shuffle-256-v16.ll |  88 +++++-----------
 test/CodeGen/X86/vector-shuffle-256-v32.ll | 115 ++++++---------------
 test/CodeGen/X86/vector-shuffle-256-v8.ll  |   4 +-
 5 files changed, 111 insertions(+), 171 deletions(-)

diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index 4e8616fb138..d07b2efee0b 100644
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -10139,14 +10139,43 @@ static SDValue splitAndLowerVectorShuffle(SDLoc DL, MVT VT, SDValue V1,
   MVT ScalarVT = VT.getScalarType();
   MVT SplitVT = MVT::getVectorVT(ScalarVT, NumElements / 2);
 
-  SDValue LoV1 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, SplitVT, V1,
-                             DAG.getIntPtrConstant(0));
-  SDValue HiV1 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, SplitVT, V1,
-                             DAG.getIntPtrConstant(SplitNumElements));
-  SDValue LoV2 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, SplitVT, V2,
-                             DAG.getIntPtrConstant(0));
-  SDValue HiV2 = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, SplitVT, V2,
-                             DAG.getIntPtrConstant(SplitNumElements));
+  // Rather than splitting build-vectors, just build two narrower build
+  // vectors. This helps shuffling with splats and zeros.
+  auto SplitVector = [&](SDValue V) {
+    while (V.getOpcode() == ISD::BITCAST)
+      V = V->getOperand(0);
+
+    MVT OrigVT = V.getSimpleValueType();
+    int OrigNumElements = OrigVT.getVectorNumElements();
+    int OrigSplitNumElements = OrigNumElements / 2;
+    MVT OrigScalarVT = OrigVT.getScalarType();
+    MVT OrigSplitVT = MVT::getVectorVT(OrigScalarVT, OrigNumElements / 2);
+
+    SDValue LoV, HiV;
+
+    auto *BV = dyn_cast<BuildVectorSDNode>(V);
+    if (!BV) {
+      LoV = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, OrigSplitVT, V,
+                        DAG.getIntPtrConstant(0));
+      HiV = DAG.getNode(ISD::EXTRACT_SUBVECTOR, DL, OrigSplitVT, V,
+                        DAG.getIntPtrConstant(OrigSplitNumElements));
+    } else {
+
+      SmallVector<SDValue, 16> LoOps, HiOps;
+      for (int i = 0; i < OrigSplitNumElements; ++i) {
+        LoOps.push_back(BV->getOperand(i));
+        HiOps.push_back(BV->getOperand(i + OrigSplitNumElements));
+      }
+      LoV = DAG.getNode(ISD::BUILD_VECTOR, DL, OrigSplitVT, LoOps);
+      HiV = DAG.getNode(ISD::BUILD_VECTOR, DL, OrigSplitVT, HiOps);
+    }
+    return std::make_pair(DAG.getNode(ISD::BITCAST, DL, SplitVT, LoV),
+                          DAG.getNode(ISD::BITCAST, DL, SplitVT, HiV));
+  };
+
+  SDValue LoV1, HiV1, LoV2, HiV2;
+  std::tie(LoV1, HiV1) = SplitVector(V1);
+  std::tie(LoV2, HiV2) = SplitVector(V2);
 
   // Now create two 4-way blends of these half-width vectors.
   auto HalfBlend = [&](ArrayRef<int> HalfMask) {
diff --git a/test/CodeGen/X86/vec_cast2.ll b/test/CodeGen/X86/vec_cast2.ll
index ab391889b71..d37a93dddfb 100644
--- a/test/CodeGen/X86/vec_cast2.ll
+++ b/test/CodeGen/X86/vec_cast2.ll
@@ -5,7 +5,7 @@ define <8 x float> @foo1_8(<8 x i8> %src) {
 ; CHECK-LABEL: foo1_8:
 ; CHECK:       ## BB#0:
 ; CHECK-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm0[4,4,5,5,6,6,7,7]
-; CHECK-NEXT:    vpmovzxwd %xmm0, %xmm0
+; CHECK-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
 ; CHECK-NEXT:    vpslld $24, %xmm0, %xmm0
 ; CHECK-NEXT:    vpsrad $24, %xmm0, %xmm0
 ; CHECK-NEXT:    vpslld $24, %xmm1, %xmm1
@@ -16,7 +16,7 @@ define <8 x float> @foo1_8(<8 x i8> %src) {
 ;
 ; CHECK-WIDE-LABEL: foo1_8:
 ; CHECK-WIDE:       ## BB#0:
-; CHECK-WIDE-NEXT:    vpmovzxbd %xmm0, %xmm1
+; CHECK-WIDE-NEXT:    vpmovzxbd {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
 ; CHECK-WIDE-NEXT:    vpslld $24, %xmm1, %xmm1
 ; CHECK-WIDE-NEXT:    vpsrad $24, %xmm1, %xmm1
 ; CHECK-WIDE-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
@@ -40,7 +40,7 @@ define <4 x float> @foo1_4(<4 x i8> %src) {
 ;
 ; CHECK-WIDE-LABEL: foo1_4:
 ; CHECK-WIDE:       ## BB#0:
-; CHECK-WIDE-NEXT:    vpmovzxbd %xmm0, %xmm0
+; CHECK-WIDE-NEXT:    vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
 ; CHECK-WIDE-NEXT:    vpslld $24, %xmm0, %xmm0
 ; CHECK-WIDE-NEXT:    vpsrad $24, %xmm0, %xmm0
 ; CHECK-WIDE-NEXT:    vcvtdq2ps %xmm0, %xmm0
@@ -52,7 +52,7 @@ define <4 x float> @foo1_4(<4 x i8> %src) {
 define <8 x float> @foo2_8(<8 x i8> %src) {
 ; CHECK-LABEL: foo2_8:
 ; CHECK:       ## BB#0:
-; CHECK-NEXT:    vpmovzxwd %xmm0, %xmm1
+; CHECK-NEXT:    vpmovzxwd {{.*#+}} xmm1 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
 ; CHECK-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm0[4,4,5,5,6,6,7,7]
 ; CHECK-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; CHECK-NEXT:    vandps LCPI2_0, %ymm0, %ymm0
@@ -61,20 +61,12 @@ define <8 x float> @foo2_8(<8 x i8> %src) {
 ;
 ; CHECK-WIDE-LABEL: foo2_8:
 ; CHECK-WIDE:       ## BB#0:
-; CHECK-WIDE-NEXT:    vxorps %xmm1, %xmm1, %xmm1
-; CHECK-WIDE-NEXT:    vextractf128 $1, %ymm1, %xmm2
-; CHECK-WIDE-NEXT:    vmovdqa {{.*#+}} xmm3 = <1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u>
-; CHECK-WIDE-NEXT:    vpshufb %xmm3, %xmm2, %xmm4
-; CHECK-WIDE-NEXT:    vmovdqa {{.*#+}} xmm5 = <2,6,10,14,u,u,u,u,u,u,u,u,u,u,u,u>
-; CHECK-WIDE-NEXT:    vpshufb %xmm5, %xmm2, %xmm2
-; CHECK-WIDE-NEXT:    vpshufd {{.*#+}} xmm6 = xmm0[1,1,2,3]
-; CHECK-WIDE-NEXT:    vpunpcklbw {{.*#+}} xmm2 = xmm6[0],xmm2[0],xmm6[1],xmm2[1],xmm6[2],xmm2[2],xmm6[3],xmm2[3],xmm6[4],xmm2[4],xmm6[5],xmm2[5],xmm6[6],xmm2[6],xmm6[7],xmm2[7]
-; CHECK-WIDE-NEXT:    vpunpcklbw {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1],xmm2[2],xmm4[2],xmm2[3],xmm4[3],xmm2[4],xmm4[4],xmm2[5],xmm4[5],xmm2[6],xmm4[6],xmm2[7],xmm4[7]
-; CHECK-WIDE-NEXT:    vpshufb %xmm3, %xmm1, %xmm3
-; CHECK-WIDE-NEXT:    vpshufb %xmm5, %xmm1, %xmm1
-; CHECK-WIDE-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3],xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
-; CHECK-WIDE-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
-; CHECK-WIDE-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; CHECK-WIDE-NEXT:    vpmovzxbd {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; CHECK-WIDE-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
+; CHECK-WIDE-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; CHECK-WIDE-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
+; CHECK-WIDE-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
+; CHECK-WIDE-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; CHECK-WIDE-NEXT:    vcvtdq2ps %ymm0, %ymm0
 ; CHECK-WIDE-NEXT:    retl
   %res = uitofp <8 x i8> %src to <8 x float>
@@ -90,7 +82,7 @@ define <4 x float> @foo2_4(<4 x i8> %src) {
 ;
 ; CHECK-WIDE-LABEL: foo2_4:
 ; CHECK-WIDE:       ## BB#0:
-; CHECK-WIDE-NEXT:    vpmovzxbd %xmm0, %xmm0
+; CHECK-WIDE-NEXT:    vpmovzxbd {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
 ; CHECK-WIDE-NEXT:    vcvtdq2ps %xmm0, %xmm0
 ; CHECK-WIDE-NEXT:    retl
   %res = uitofp <4 x i8> %src to <4 x float>
diff --git a/test/CodeGen/X86/vector-shuffle-256-v16.ll b/test/CodeGen/X86/vector-shuffle-256-v16.ll
index 96327257d0b..f53466ff19a 100644
--- a/test/CodeGen/X86/vector-shuffle-256-v16.ll
+++ b/test/CodeGen/X86/vector-shuffle-256-v16.ll
@@ -1370,15 +1370,10 @@ define <16 x i16> @shuffle_v16i16_00_16_01_17_02_18_03_19_04_20_05_21_06_22_07_2
 define <16 x i16> @shuffle_v16i16_zz_00_zz_02_zz_04_zz_06_zz_08_zz_10_zz_12_zz_14(<16 x i16> %a) {
 ; AVX1-LABEL: shuffle_v16i16_zz_00_zz_02_zz_04_zz_06_zz_08_zz_10_zz_12_zz_14:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %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:    vpxor %xmm3, %xmm3, %xmm3
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm3 = xmm3[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3]
-; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vpslld $16, %xmm0, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpslld $16, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_zz_00_zz_02_zz_04_zz_06_zz_08_zz_10_zz_12_zz_14:
@@ -1392,20 +1387,10 @@ define <16 x i16> @shuffle_v16i16_zz_00_zz_02_zz_04_zz_06_zz_08_zz_10_zz_12_zz_1
 define <16 x i16> @shuffle_v16i16_zz_zz_zz_00_zz_zz_zz_04_zz_zz_zz_08_zz_zz_zz_12(<16 x i16> %a) {
 ; AVX1-LABEL: shuffle_v16i16_zz_zz_zz_00_zz_zz_zz_04_zz_zz_zz_08_zz_zz_zz_12:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[0,2,2,3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[0,2,2,3,4,5,6,7]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm2 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[0,1,0,3,4,5,6,7]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm3 = xmm1[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm2 = xmm3[0],xmm2[0],xmm3[1],xmm2[1],xmm3[2],xmm2[2],xmm3[3],xmm2[3]
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,2,3,4,5,6,7]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,1,0,3,4,5,6,7]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3]
-; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vpsllq $48, %xmm0, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpsllq $48, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_zz_zz_zz_00_zz_zz_zz_04_zz_zz_zz_08_zz_zz_zz_12:
@@ -1419,15 +1404,10 @@ define <16 x i16> @shuffle_v16i16_zz_zz_zz_00_zz_zz_zz_04_zz_zz_zz_08_zz_zz_zz_1
 define <16 x i16> @shuffle_v16i16_01_zz_03_zz_05_zz_07_zz_09_zz_11_zz_13_zz_15_zz(<16 x i16> %a) {
 ; AVX1-LABEL: shuffle_v16i16_01_zz_03_zz_05_zz_07_zz_09_zz_11_zz_13_zz_15_zz:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [2,3,6,7,10,11,14,15,14,15,10,11,12,13,14,15]
-; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm3 = xmm3[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3]
-; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vpsrld $16, %xmm0, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpsrld $16, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_01_zz_03_zz_05_zz_07_zz_09_zz_11_zz_13_zz_15_zz:
@@ -1441,22 +1421,10 @@ define <16 x i16> @shuffle_v16i16_01_zz_03_zz_05_zz_07_zz_09_zz_11_zz_13_zz_15_z
 define <16 x i16> @shuffle_v16i16_02_03_zz_zz_06_07_zz_zz_10_11_zz_zz_14_15_zz_zz(<16 x i16> %a) {
 ; AVX1-LABEL: shuffle_v16i16_02_03_zz_zz_06_07_zz_zz_10_11_zz_zz_14_15_zz_zz:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm2[2,3,0,1]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3]
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm4 = [8,9,12,13,2,3,2,3,8,9,12,13,12,13,14,15]
-; AVX1-NEXT:    vpshufb %xmm4, %xmm3, %xmm3
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; AVX1-NEXT:    vpshufb %xmm4, %xmm2, %xmm2
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm3[0]
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[2,3,0,1]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm3 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3]
-; AVX1-NEXT:    vpshufb %xmm4, %xmm3, %xmm3
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX1-NEXT:    vpshufb %xmm4, %xmm0, %xmm0
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm3[0]
-; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vpsrlq $32, %xmm0, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpsrlq $32, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_02_03_zz_zz_06_07_zz_zz_10_11_zz_zz_14_15_zz_zz:
@@ -1470,14 +1438,12 @@ define <16 x i16> @shuffle_v16i16_02_03_zz_zz_06_07_zz_zz_10_11_zz_zz_14_15_zz_z
 define <16 x i16> @shuffle_v16i16_16_zz_zz_zz_17_zz_zz_zz_18_zz_zz_zz_19_zz_zz_zz(<16 x i16> %a) {
 ; AVX1-LABEL: shuffle_v16i16_16_zz_zz_zz_17_zz_zz_zz_18_zz_zz_zz_19_zz_zz_zz:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm2 = xmm0[8,9,2,3,12,13,2,3,8,9,12,13,12,13,14,15]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm2 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,1,2,1,4,5,6,7]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vpmovzxwq {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero
+; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
+; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[8,9,2,3,12,13,2,3,8,9,12,13,12,13,14,15]
+; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_16_zz_zz_zz_17_zz_zz_zz_18_zz_zz_zz_19_zz_zz_zz:
@@ -1491,11 +1457,11 @@ define <16 x i16> @shuffle_v16i16_16_zz_zz_zz_17_zz_zz_zz_18_zz_zz_zz_19_zz_zz_z
 define <16 x i16> @shuffle_v16i16_16_zz_17_zz_18_zz_19_zz_20_zz_21_zz_22_zz_22_zz(<16 x i16> %a) {
 ; AVX1-LABEL: shuffle_v16i16_16_zz_17_zz_18_zz_19_zz_20_zz_21_zz_22_zz_22_zz:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpmovzxwq {{.*#+}} xmm1 = xmm1[0],zero,zero,zero,xmm1[1],zero,zero,zero
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3]
+; AVX1-NEXT:    vpmovzxwd {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
diff --git a/test/CodeGen/X86/vector-shuffle-256-v32.ll b/test/CodeGen/X86/vector-shuffle-256-v32.ll
index b8e2c9c6832..c0c5f1b21de 100644
--- a/test/CodeGen/X86/vector-shuffle-256-v32.ll
+++ b/test/CodeGen/X86/vector-shuffle-256-v32.ll
@@ -1662,16 +1662,10 @@ define <32 x i8> @shuffle_v32i8_00_32_01_33_02_34_03_35_04_36_05_37_06_38_07_39_
 define <32 x i8> @shuffle_v32i8_zz_00_zz_02_zz_04_zz_06_zz_08_zz_10_zz_12_zz_14_zz_16_zz_18_zz_20_zz_22_zz_24_zz_26_zz_28_zz_30(<32 x i8> %a) {
 ; AVX1-LABEL: shuffle_v32i8_zz_00_zz_02_zz_04_zz_06_zz_08_zz_10_zz_12_zz_14_zz_16_zz_18_zz_20_zz_22_zz_24_zz_26_zz_28_zz_30:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = <0,2,4,6,8,10,12,14,u,u,u,u,u,u,u,u>
-; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm3 = xmm3[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm3 = xmm3[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm1 = xmm3[0],xmm1[0],xmm3[1],xmm1[1],xmm3[2],xmm1[2],xmm3[3],xmm1[3],xmm3[4],xmm1[4],xmm3[5],xmm1[5],xmm3[6],xmm1[6],xmm3[7],xmm1[7]
-; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm3[0],xmm0[0],xmm3[1],xmm0[1],xmm3[2],xmm0[2],xmm3[3],xmm0[3],xmm3[4],xmm0[4],xmm3[5],xmm0[5],xmm3[6],xmm0[6],xmm3[7],xmm0[7]
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vpsllw $8, %xmm0, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpsllw $8, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v32i8_zz_00_zz_02_zz_04_zz_06_zz_08_zz_10_zz_12_zz_14_zz_16_zz_18_zz_20_zz_22_zz_24_zz_26_zz_28_zz_30:
@@ -1685,15 +1679,10 @@ define <32 x i8> @shuffle_v32i8_zz_00_zz_02_zz_04_zz_06_zz_08_zz_10_zz_12_zz_14_
 define <32 x i8> @shuffle_v32i8_zz_zz_00_01_zz_zz_04_05_zz_zz_08_09_zz_zz_12_13_zz_zz_16_17_zz_zz_20_21_zz_zz_24_25_zz_zz_28_29(<32 x i8> %a) {
 ; AVX1-LABEL: shuffle_v32i8_zz_zz_00_01_zz_zz_04_05_zz_zz_08_09_zz_zz_12_13_zz_zz_16_17_zz_zz_20_21_zz_zz_24_25_zz_zz_28_29:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [128,128,0,1,128,128,4,5,128,128,8,9,128,128,12,13]
-; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm3 = xmm3[0,0],zero,zero,xmm3[0,0],zero,zero,xmm3[0,0],zero,zero,xmm3[0,0],zero,zero
-; AVX1-NEXT:    vpor %xmm1, %xmm3, %xmm1
-; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vpslld $16, %xmm0, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpslld $16, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v32i8_zz_zz_00_01_zz_zz_04_05_zz_zz_08_09_zz_zz_12_13_zz_zz_16_17_zz_zz_20_21_zz_zz_24_25_zz_zz_28_29:
@@ -1707,15 +1696,10 @@ define <32 x i8> @shuffle_v32i8_zz_zz_00_01_zz_zz_04_05_zz_zz_08_09_zz_zz_12_13_
 define <32 x i8> @shuffle_v32i8_zz_zz_zz_zz_zz_zz_00_01_zz_zz_zz_zz_zz_zz_08_09_zz_zz_zz_zz_zz_zz_16_17_zz_zz_zz_zz_zz_zz_24_25(<32 x i8> %a) {
 ; AVX1-LABEL: shuffle_v32i8_zz_zz_zz_zz_zz_zz_00_01_zz_zz_zz_zz_zz_zz_08_09_zz_zz_zz_zz_zz_zz_16_17_zz_zz_zz_zz_zz_zz_24_25:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [128,128,128,128,128,128,0,1,128,128,128,128,128,128,8,9]
-; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm3 = xmm3[0,0,0,0,0,0],zero,zero,xmm3[0,0,0,0,0,0],zero,zero
-; AVX1-NEXT:    vpor %xmm1, %xmm3, %xmm1
-; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vpsllq $48, %xmm0, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpsllq $48, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v32i8_zz_zz_zz_zz_zz_zz_00_01_zz_zz_zz_zz_zz_zz_08_09_zz_zz_zz_zz_zz_zz_16_17_zz_zz_zz_zz_zz_zz_24_25:
@@ -1729,16 +1713,10 @@ define <32 x i8> @shuffle_v32i8_zz_zz_zz_zz_zz_zz_00_01_zz_zz_zz_zz_zz_zz_08_09_
 define <32 x i8> @shuffle_v32i8_01_zz_03_zz_05_zz_07_zz_09_zz_11_zz_13_zz_15_zz_17_zz_19_zz_21_zz_23_zz_25_zz_27_zz_29_zz_31_zz(<32 x i8> %a) {
 ; AVX1-LABEL: shuffle_v32i8_01_zz_03_zz_05_zz_07_zz_09_zz_11_zz_13_zz_15_zz_17_zz_19_zz_21_zz_23_zz_25_zz_27_zz_29_zz_31_zz:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = <1,3,5,7,9,11,13,15,u,u,u,u,u,u,u,u>
-; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm3 = xmm3[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm3 = xmm3[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3],xmm1[4],xmm3[4],xmm1[5],xmm3[5],xmm1[6],xmm3[6],xmm1[7],xmm3[7]
-; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vpsrlw $8, %xmm0, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpsrlw $8, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v32i8_01_zz_03_zz_05_zz_07_zz_09_zz_11_zz_13_zz_15_zz_17_zz_19_zz_21_zz_23_zz_25_zz_27_zz_29_zz_31_zz:
@@ -1752,15 +1730,10 @@ define <32 x i8> @shuffle_v32i8_01_zz_03_zz_05_zz_07_zz_09_zz_11_zz_13_zz_15_zz_
 define <32 x i8> @shuffle_v32i8_02_03_zz_zz_06_07_zz_zz_10_11_zz_zz_14_15_zz_zz_18_19_zz_zz_22_23_zz_zz_26_27_zz_zz_30_31_zz_zz(<32 x i8> %a) {
 ; AVX1-LABEL: shuffle_v32i8_02_03_zz_zz_06_07_zz_zz_10_11_zz_zz_14_15_zz_zz_18_19_zz_zz_22_23_zz_zz_26_27_zz_zz_30_31_zz_zz:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [2,3,128,128,6,7,128,128,10,11,128,128,14,15,128,128]
-; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm3 = zero,zero,xmm3[0,0],zero,zero,xmm3[0,0],zero,zero,xmm3[0,0],zero,zero,xmm3[0,0]
-; AVX1-NEXT:    vpor %xmm3, %xmm1, %xmm1
-; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm3, %xmm0, %xmm0
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vpsrld $16, %xmm0, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpsrld $16, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v32i8_02_03_zz_zz_06_07_zz_zz_10_11_zz_zz_14_15_zz_zz_18_19_zz_zz_22_23_zz_zz_26_27_zz_zz_30_31_zz_zz:
@@ -1774,19 +1747,10 @@ define <32 x i8> @shuffle_v32i8_02_03_zz_zz_06_07_zz_zz_10_11_zz_zz_14_15_zz_zz_
 define <32 x i8> @shuffle_v32i8_07_zz_zz_zz_zz_zz_zz_zz_15_zz_zz_zz_zz_z_zz_zz_23_zz_zz_zz_zz_zz_zz_zz_31_zz_zz_zz_zz_zz_zz_zz(<32 x i8> %a) {
 ; AVX1-LABEL: shuffle_v32i8_07_zz_zz_zz_zz_zz_zz_zz_15_zz_zz_zz_zz_z_zz_zz_23_zz_zz_zz_zz_zz_zz_zz_31_zz_zz_zz_zz_zz_zz_zz:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = <7,128,128,128,15,128,128,128,u,u,u,u,u,u,u,u>
-; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX1-NEXT:    vpxor %xmm3, %xmm3, %xmm3
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm4 = zero,xmm3[0,0,0],zero,xmm3[0,0,0,u,u,u,u,u,u,u,u]
-; AVX1-NEXT:    vpor %xmm1, %xmm4, %xmm1
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm3 = xmm3[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm3 = xmm3[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3],xmm1[4],xmm3[4],xmm1[5],xmm3[5],xmm1[6],xmm3[6],xmm1[7],xmm3[7]
-; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX1-NEXT:    vpor %xmm0, %xmm4, %xmm0
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vpsrlq $56, %xmm0, %xmm1
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vpsrlq $56, %xmm0, %xmm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v32i8_07_zz_zz_zz_zz_zz_zz_zz_15_zz_zz_zz_zz_z_zz_zz_23_zz_zz_zz_zz_zz_zz_zz_31_zz_zz_zz_zz_zz_zz_zz:
@@ -1802,14 +1766,8 @@ define <32 x i8> @shuffle_v32i8_32_zz_zz_zz_zz_zz_zz_zz_33_zz_zz_zz_zz_zz_zz_zz_
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero,xmm0[u,u,u,u,u,u,u,u]
 ; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm3 = zero,xmm2[0,0,0],zero,xmm2[0,0,0,u,u,u,u,u,u,u,u]
-; AVX1-NEXT:    vpor %xmm1, %xmm3, %xmm1
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[0,0,0,0,4,5,6,7]
 ; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[u,u,u,u,u,u,u,u]
-; AVX1-NEXT:    vpor %xmm0, %xmm3, %xmm0
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
+; AVX1-NEXT:    vpmovzxbq {{.*#+}} xmm0 = xmm0[0],zero,zero,zero,zero,zero,zero,zero,xmm0[1],zero,zero,zero,zero,zero,zero,zero
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
@@ -1825,16 +1783,12 @@ define <32 x i8> @shuffle_v32i8_32_zz_zz_zz_zz_zz_zz_zz_33_zz_zz_zz_zz_zz_zz_zz_
 define <32 x i8> @shuffle_v32i8_32_zz_zz_zz_33_zz_zz_zz_34_zz_zz_zz_35_zz_zz_zz_36_zz_zz_zz_37_zz_zz_zz_38_zz_zz_zz_39_zz_zz_zz(<32 x i8> %a) {
 ; AVX1-LABEL: shuffle_v32i8_32_zz_zz_zz_33_zz_zz_zz_34_zz_zz_zz_35_zz_zz_zz_36_zz_zz_zz_37_zz_zz_zz_38_zz_zz_zz_39_zz_zz_zz:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[1,1,2,3]
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = xmm0[0],zero,zero,zero,xmm0[1],zero,zero,zero,xmm0[2],zero,zero,zero,xmm0[3],zero,zero,zero
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
 ; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm3 = xmm2[0,0,1,1,2,2,3,3]
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm1 = xmm1[0],xmm3[0],xmm1[1],xmm3[1],xmm1[2],xmm3[2],xmm1[3],xmm3[3],xmm1[4],xmm3[4],xmm1[5],xmm3[5],xmm1[6],xmm3[6],xmm1[7],xmm3[7]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3],xmm0[4],xmm3[4],xmm0[5],xmm3[5],xmm0[6],xmm3[6],xmm0[7],xmm3[7]
 ; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v32i8_32_zz_zz_zz_33_zz_zz_zz_34_zz_zz_zz_35_zz_zz_zz_36_zz_zz_zz_37_zz_zz_zz_38_zz_zz_zz_39_zz_zz_zz:
@@ -1848,12 +1802,11 @@ define <32 x i8> @shuffle_v32i8_32_zz_zz_zz_33_zz_zz_zz_34_zz_zz_zz_35_zz_zz_zz_
 define <32 x i8> @shuffle_v32i8_32_zz_33_zz_34_zz_35_zz_36_zz_37_zz_38_zz_39_zz_40_zz_41_zz_42_zz_43_zz_44_zz_45_zz_46_zz_47_zz(<32 x i8> %a) {
 ; AVX1-LABEL: shuffle_v32i8_32_zz_33_zz_34_zz_35_zz_36_zz_37_zz_38_zz_39_zz_40_zz_41_zz_42_zz_43_zz_44_zz_45_zz_46_zz_47_zz:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm0[2,3,0,1]
-; AVX1-NEXT:    vpxor %xmm2, %xmm2, %xmm2
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm2 = xmm2[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3],xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
-; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpsllq $56, %xmm1, %xmm1
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm1 = xmm2[0],xmm1[0],xmm2[1],xmm1[1],xmm2[2],xmm1[2],xmm2[3],xmm1[3],xmm2[4],xmm1[4],xmm2[5],xmm1[5],xmm2[6],xmm1[6],xmm2[7],xmm1[7]
+; AVX1-NEXT:    vpmovzxbw {{.*#+}} xmm0 = xmm0[0],zero,xmm0[1],zero,xmm0[2],zero,xmm0[3],zero,xmm0[4],zero,xmm0[5],zero,xmm0[6],zero,xmm0[7],zero
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
diff --git a/test/CodeGen/X86/vector-shuffle-256-v8.ll b/test/CodeGen/X86/vector-shuffle-256-v8.ll
index 7c237651d52..8d13ae45400 100644
--- a/test/CodeGen/X86/vector-shuffle-256-v8.ll
+++ b/test/CodeGen/X86/vector-shuffle-256-v8.ll
@@ -954,8 +954,8 @@ define <8 x i32> @shuffle_v8i32_01014545(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_00112233(<8 x i32> %a, <8 x i32> %b) {
 ; AVX1-LABEL: shuffle_v8i32_00112233:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vunpcklps {{.*#+}} xmm1 = xmm0[0,0,1,1]
-; AVX1-NEXT:    vunpckhps {{.*#+}} xmm0 = xmm0[2,2,3,3]
+; AVX1-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[0,0,1,1]
+; AVX1-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[2,2,3,3]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;