X-Git-Url: http://plrg.eecs.uci.edu/git/?a=blobdiff_plain;f=test%2FCodeGen%2FX86%2Fvector-shuffle-combining.ll;h=a22dc6b71260048272cac16d73485abc57524c17;hb=72753f87f2b80d66cfd7ca7c7b6c0db6737d4b24;hp=87b05202d9af7cee00eec4db388e1f4bec623bc8;hpb=560bddce204e2502095821b6fde446cff7bc4840;p=oota-llvm.git

diff --git a/test/CodeGen/X86/vector-shuffle-combining.ll b/test/CodeGen/X86/vector-shuffle-combining.ll
index 87b05202d9a..a22dc6b7126 100644
--- a/test/CodeGen/X86/vector-shuffle-combining.ll
+++ b/test/CodeGen/X86/vector-shuffle-combining.ll
@@ -1,8 +1,8 @@
-; RUN: llc < %s -mcpu=x86-64 -mattr=+sse2 | FileCheck %s --check-prefix=ALL --check-prefix=SSE --check-prefix=SSE2
-; RUN: llc < %s -mcpu=x86-64 -mattr=+ssse3 | FileCheck %s --check-prefix=ALL --check-prefix=SSE --check-prefix=SSSE3
-; RUN: llc < %s -mcpu=x86-64 -mattr=+sse4.1 | FileCheck %s --check-prefix=ALL --check-prefix=SSE --check-prefix=SSE41
-; RUN: llc < %s -mcpu=x86-64 -mattr=+avx | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX1
-; RUN: llc < %s -mcpu=x86-64 -mattr=+avx2 | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2
+; RUN: llc < %s -mcpu=x86-64 -mattr=+sse2 -x86-experimental-vector-shuffle-legality | FileCheck %s --check-prefix=ALL --check-prefix=SSE --check-prefix=SSE2
+; RUN: llc < %s -mcpu=x86-64 -mattr=+ssse3 -x86-experimental-vector-shuffle-legality | FileCheck %s --check-prefix=ALL --check-prefix=SSE --check-prefix=SSSE3
+; RUN: llc < %s -mcpu=x86-64 -mattr=+sse4.1 -x86-experimental-vector-shuffle-legality | FileCheck %s --check-prefix=ALL --check-prefix=SSE --check-prefix=SSE41
+; RUN: llc < %s -mcpu=x86-64 -mattr=+avx -x86-experimental-vector-shuffle-legality | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX1
+; RUN: llc < %s -mcpu=x86-64 -mattr=+avx2 -x86-experimental-vector-shuffle-legality | FileCheck %s --check-prefix=ALL --check-prefix=AVX --check-prefix=AVX2
 ;
 ; Verify that the DAG combiner correctly folds bitwise operations across
 ; shuffles, nested shuffles with undef, pairs of nested shuffles, and other
@@ -352,17 +352,13 @@ define <4 x i32> @combine_bitwise_ops_test3b(<4 x i32> %a, <4 x i32> %b, <4 x i3
 ; SSE2-LABEL: combine_bitwise_ops_test3b:
 ; SSE2:       # BB#0:
 ; SSE2-NEXT:    xorps %xmm1, %xmm0
-; SSE2-NEXT:    xorps %xmm1, %xmm1
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSE2-NEXT:    andps {{.*}}(%rip), %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_bitwise_ops_test3b:
 ; SSSE3:       # BB#0:
 ; SSSE3-NEXT:    xorps %xmm1, %xmm0
-; SSSE3-NEXT:    xorps %xmm1, %xmm1
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSSE3-NEXT:    andps {{.*}}(%rip), %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_bitwise_ops_test3b:
@@ -396,21 +392,22 @@ define <4 x i32> @combine_bitwise_ops_test4b(<4 x i32> %a, <4 x i32> %b, <4 x i3
 ; SSE2:       # BB#0:
 ; SSE2-NEXT:    andps %xmm1, %xmm0
 ; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,1,3]
+; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
+; SSE2-NEXT:    movaps %xmm2, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_bitwise_ops_test4b:
 ; SSSE3:       # BB#0:
 ; SSSE3-NEXT:    andps %xmm1, %xmm0
 ; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,1,3]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
+; SSSE3-NEXT:    movaps %xmm2, %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_bitwise_ops_test4b:
 ; SSE41:       # BB#0:
 ; SSE41-NEXT:    pand %xmm1, %xmm0
-; SSE41-NEXT:    pblendw {{.*#+}} xmm2 = xmm2[0,1],xmm0[2,3],xmm2[4,5],xmm0[6,7]
-; SSE41-NEXT:    movdqa %xmm2, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm2[0,1],xmm0[2,3],xmm2[4,5],xmm0[6,7]
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: combine_bitwise_ops_test4b:
@@ -435,21 +432,22 @@ define <4 x i32> @combine_bitwise_ops_test5b(<4 x i32> %a, <4 x i32> %b, <4 x i3
 ; SSE2:       # BB#0:
 ; SSE2-NEXT:    orps %xmm1, %xmm0
 ; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,1,3]
+; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
+; SSE2-NEXT:    movaps %xmm2, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_bitwise_ops_test5b:
 ; SSSE3:       # BB#0:
 ; SSSE3-NEXT:    orps %xmm1, %xmm0
 ; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,1,3]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
+; SSSE3-NEXT:    movaps %xmm2, %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_bitwise_ops_test5b:
 ; SSE41:       # BB#0:
 ; SSE41-NEXT:    por %xmm1, %xmm0
-; SSE41-NEXT:    pblendw {{.*#+}} xmm2 = xmm2[0,1],xmm0[2,3],xmm2[4,5],xmm0[6,7]
-; SSE41-NEXT:    movdqa %xmm2, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm2[0,1],xmm0[2,3],xmm2[4,5],xmm0[6,7]
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: combine_bitwise_ops_test5b:
@@ -473,27 +471,20 @@ define <4 x i32> @combine_bitwise_ops_test6b(<4 x i32> %a, <4 x i32> %b, <4 x i3
 ; SSE2-LABEL: combine_bitwise_ops_test6b:
 ; SSE2:       # BB#0:
 ; SSE2-NEXT:    xorps %xmm1, %xmm0
-; SSE2-NEXT:    xorps %xmm1, %xmm1
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,1,3]
-; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    andps {{.*}}(%rip), %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_bitwise_ops_test6b:
 ; SSSE3:       # BB#0:
 ; SSSE3-NEXT:    xorps %xmm1, %xmm0
-; SSSE3-NEXT:    xorps %xmm1, %xmm1
-; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,1,3]
-; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    andps {{.*}}(%rip), %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_bitwise_ops_test6b:
 ; SSE41:       # BB#0:
 ; SSE41-NEXT:    pxor %xmm1, %xmm0
 ; SSE41-NEXT:    pxor %xmm1, %xmm1
-; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm0[2,3],xmm1[4,5],xmm0[6,7]
-; SSE41-NEXT:    movdqa %xmm1, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3],xmm1[4,5],xmm0[6,7]
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: combine_bitwise_ops_test6b:
@@ -552,18 +543,30 @@ define <4 x i32> @combine_bitwise_ops_test2c(<4 x i32> %a, <4 x i32> %b, <4 x i3
 }
 
 define <4 x i32> @combine_bitwise_ops_test3c(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
-; SSE-LABEL: combine_bitwise_ops_test3c:
-; SSE:       # BB#0:
-; SSE-NEXT:    xorps %xmm1, %xmm0
-; SSE-NEXT:    xorps %xmm1, %xmm1
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_bitwise_ops_test3c:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    xorps %xmm1, %xmm0
+; SSE2-NEXT:    xorps %xmm1, %xmm1
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[2,3]
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: combine_bitwise_ops_test3c:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    xorps %xmm1, %xmm0
+; SSSE3-NEXT:    xorps %xmm1, %xmm1
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[2,3]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_bitwise_ops_test3c:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    xorps %xmm1, %xmm0
+; SSE41-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0,2],zero,zero
+; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_bitwise_ops_test3c:
 ; AVX:       # BB#0:
 ; AVX-NEXT:    vxorps %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vxorps %xmm1, %xmm1, %xmm1
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
+; AVX-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,2],zero,zero
 ; AVX-NEXT:    retq
   %shuf1 = shufflevector <4 x i32> %a, <4 x i32> %c, <4 x i32><i32 0, i32 2, i32 5, i32 7>
   %shuf2 = shufflevector <4 x i32> %b, <4 x i32> %c, <4 x i32><i32 0, i32 2, i32 5, i32 7>
@@ -614,7 +617,7 @@ define <4 x i32> @combine_bitwise_ops_test6c(<4 x i32> %a, <4 x i32> %b, <4 x i3
 ; SSE:       # BB#0:
 ; SSE-NEXT:    xorps %xmm1, %xmm0
 ; SSE-NEXT:    xorps %xmm1, %xmm1
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[1,3]
+; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,1],xmm0[1,3]
 ; SSE-NEXT:    movaps %xmm1, %xmm0
 ; SSE-NEXT:    retq
 ;
@@ -622,7 +625,7 @@ define <4 x i32> @combine_bitwise_ops_test6c(<4 x i32> %a, <4 x i32> %b, <4 x i3
 ; AVX:       # BB#0:
 ; AVX-NEXT:    vxorps %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    vxorps %xmm1, %xmm1, %xmm1
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[0,2],xmm0[1,3]
+; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[0,1],xmm0[1,3]
 ; AVX-NEXT:    retq
   %shuf1 = shufflevector <4 x i32> %c, <4 x i32> %a, <4 x i32><i32 0, i32 2, i32 5, i32 7>
   %shuf2 = shufflevector <4 x i32> %c, <4 x i32> %b, <4 x i32><i32 0, i32 2, i32 5, i32 7>
@@ -856,16 +859,15 @@ define <4 x i32> @combine_nested_undef_test14(<4 x i32> %A, <4 x i32> %B) {
 define <4 x i32> @combine_nested_undef_test15(<4 x i32> %A, <4 x i32> %B) {
 ; SSE-LABEL: combine_nested_undef_test15:
 ; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[3,1]
-; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[2,1,0,3]
+; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[3,0]
+; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[0,1]
+; SSE-NEXT:    movaps %xmm1, %xmm0
 ; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_nested_undef_test15:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[2,0],xmm0[3,1]
-; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
+; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[3,0]
+; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[2,0],xmm0[0,1]
 ; AVX-NEXT:    retq
   %1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 0, i32 4, i32 3, i32 1>
   %2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 2, i32 1, i32 0, i32 3>
@@ -875,53 +877,55 @@ define <4 x i32> @combine_nested_undef_test15(<4 x i32> %A, <4 x i32> %B) {
 define <4 x i32> @combine_nested_undef_test16(<4 x i32> %A, <4 x i32> %B) {
 ; SSE2-LABEL: combine_nested_undef_test16:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[1,3]
 ; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_nested_undef_test16:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[1,3]
 ; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_nested_undef_test16:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5],xmm1[6,7]
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
+; SSE41-NEXT:    blendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
+; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,1,0,3]
 ; SSE41-NEXT:    retq
 ;
-; AVX1-LABEL: combine_nested_undef_test16:
-; AVX1:       # BB#0:
-; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5],xmm1[6,7]
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
-; AVX1-NEXT:    retq
-;
-; AVX2-LABEL: combine_nested_undef_test16:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
-; AVX2-NEXT:    retq
+; AVX-LABEL: combine_nested_undef_test16:
+; AVX:       # BB#0:
+; AVX-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[2,1,0,3]
+; AVX-NEXT:    retq
   %1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
   %2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 2, i32 1, i32 0, i32 3>
   ret <4 x i32> %2
 }
 
 define <4 x i32> @combine_nested_undef_test17(<4 x i32> %A, <4 x i32> %B) {
-; SSE-LABEL: combine_nested_undef_test17:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,0]
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[3,1]
-; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[2,1,0,3]
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_nested_undef_test17:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,1],xmm1[0,2]
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: combine_nested_undef_test17:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,1],xmm1[0,2]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_nested_undef_test17:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    blendps {{.*#+}} xmm0 = xmm1[0],xmm0[1,2,3]
+; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,1,0,1]
+; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_nested_undef_test17:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[0,2],xmm0[3,1]
-; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
+; AVX-NEXT:    vblendps {{.*#+}} xmm0 = xmm1[0],xmm0[1,2,3]
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[3,1,0,1]
 ; AVX-NEXT:    retq
   %1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 4, i32 1, i32 3, i32 1>
   %2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 2, i32 1, i32 0, i32 3>
@@ -944,18 +948,30 @@ define <4 x i32> @combine_nested_undef_test18(<4 x i32> %A, <4 x i32> %B) {
 }
 
 define <4 x i32> @combine_nested_undef_test19(<4 x i32> %A, <4 x i32> %B) {
-; SSE-LABEL: combine_nested_undef_test19:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[0,0]
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,2]
-; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,0,0,0]
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_nested_undef_test19:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[0,0]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: combine_nested_undef_test19:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[0,0]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_nested_undef_test19:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    blendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
+; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0,0,0]
+; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_nested_undef_test19:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[0,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,2]
-; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,0,0,0]
+; AVX-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[1,0,0,0]
 ; AVX-NEXT:    retq
   %1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 0, i32 4, i32 5, i32 6>
   %2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 2, i32 0, i32 0, i32 0>
@@ -963,16 +979,30 @@ define <4 x i32> @combine_nested_undef_test19(<4 x i32> %A, <4 x i32> %B) {
 }
 
 define <4 x i32> @combine_nested_undef_test20(<4 x i32> %A, <4 x i32> %B) {
-; SSE-LABEL: combine_nested_undef_test20:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,2],xmm1[0,0]
-; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_nested_undef_test20:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[2,3]
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,3,1]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: combine_nested_undef_test20:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[2,3]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,3,1]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_nested_undef_test20:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    blendpd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
+; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,3,0]
+; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_nested_undef_test20:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[3,2],xmm1[0,0]
-; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,1,0,3]
+; AVX-NEXT:    vblendpd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,3,0]
 ; AVX-NEXT:    retq
   %1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 3, i32 2, i32 4, i32 4>
   %2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 2, i32 1, i32 0, i32 3>
@@ -980,18 +1010,30 @@ define <4 x i32> @combine_nested_undef_test20(<4 x i32> %A, <4 x i32> %B) {
 }
 
 define <4 x i32> @combine_nested_undef_test21(<4 x i32> %A, <4 x i32> %B) {
-; SSE-LABEL: combine_nested_undef_test21:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,0]
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[3,1]
-; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,1,0,3]
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_nested_undef_test21:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,1]
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,1,3]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: combine_nested_undef_test21:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,1]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,1,3]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_nested_undef_test21:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    blendps {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2,3]
+; SSE41-NEXT:    movddup {{.*#+}} xmm0 = xmm0[0,0]
+; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_nested_undef_test21:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[1,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[0,2],xmm0[3,1]
-; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
+; AVX-NEXT:    vblendps {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2,3]
+; AVX-NEXT:    vmovddup {{.*#+}} xmm0 = xmm0[0,0]
 ; AVX-NEXT:    retq
   %1 = shufflevector <4 x i32> %A, <4 x i32> %B, <4 x i32> <i32 4, i32 1, i32 3, i32 1>
   %2 = shufflevector <4 x i32> %1, <4 x i32> undef, <4 x i32> <i32 0, i32 1, i32 0, i32 3>
@@ -1118,24 +1160,10 @@ define <4 x i32> @combine_nested_undef_test28(<4 x i32> %A, <4 x i32> %B) {
 }
 
 define <4 x float> @combine_test1(<4 x float> %a, <4 x float> %b) {
-; SSE2-LABEL: combine_test1:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movaps %xmm1, %xmm2
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm1[1,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,1,3]
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: combine_test1:
-; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movaps %xmm1, %xmm0
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: combine_test1:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps %xmm1, %xmm0
-; SSE41-NEXT:    retq
+; SSE-LABEL: combine_test1:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps %xmm1, %xmm0
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_test1:
 ; AVX:       # BB#0:
@@ -1149,24 +1177,19 @@ define <4 x float> @combine_test1(<4 x float> %a, <4 x float> %b) {
 define <4 x float> @combine_test2(<4 x float> %a, <4 x float> %b) {
 ; SSE2-LABEL: combine_test2:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[3,0]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2]
+; SSE2-NEXT:    movss {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_test2:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[3,0]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2]
+; SSSE3-NEXT:    movss {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_test2:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    blendps {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
-; SSE41-NEXT:    movaps %xmm1, %xmm0
+; SSE41-NEXT:    blendps {{.*#+}} xmm0 = xmm0[0],xmm1[1,2,3]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_test2:
@@ -1212,28 +1235,19 @@ define <4 x float> @combine_test4(<4 x float> %a, <4 x float> %b) {
 define <4 x float> @combine_test5(<4 x float> %a, <4 x float> %b) {
 ; SSE2-LABEL: combine_test5:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movaps %xmm1, %xmm2
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[3,0],xmm2[2,0]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,1],xmm1[2,0]
-; SSE2-NEXT:    movaps %xmm2, %xmm0
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm1[0,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[2,3]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_test5:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movaps %xmm1, %xmm2
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[3,0],xmm2[2,0]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,1],xmm1[2,0]
-; SSSE3-NEXT:    movaps %xmm2, %xmm0
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm1[0,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[2,3]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_test5:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    blendps {{.*#+}} xmm1 = xmm1[0],xmm0[1],xmm1[2,3]
-; SSE41-NEXT:    movaps %xmm1, %xmm0
+; SSE41-NEXT:    blendps {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2,3]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_test5:
@@ -1246,24 +1260,10 @@ define <4 x float> @combine_test5(<4 x float> %a, <4 x float> %b) {
 }
 
 define <4 x i32> @combine_test6(<4 x i32> %a, <4 x i32> %b) {
-; SSE2-LABEL: combine_test6:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movaps %xmm1, %xmm2
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm1[1,3]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,1,3]
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: combine_test6:
-; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movaps %xmm1, %xmm0
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: combine_test6:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps %xmm1, %xmm0
-; SSE41-NEXT:    retq
+; SSE-LABEL: combine_test6:
+; SSE:       # BB#0:
+; SSE-NEXT:    movaps %xmm1, %xmm0
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_test6:
 ; AVX:       # BB#0:
@@ -1277,24 +1277,19 @@ define <4 x i32> @combine_test6(<4 x i32> %a, <4 x i32> %b) {
 define <4 x i32> @combine_test7(<4 x i32> %a, <4 x i32> %b) {
 ; SSE2-LABEL: combine_test7:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[3,0]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2]
+; SSE2-NEXT:    movss {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_test7:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[3,0]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2]
+; SSSE3-NEXT:    movss {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_test7:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3,4,5,6,7]
-; SSE41-NEXT:    movdqa %xmm1, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3,4,5,6,7]
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: combine_test7:
@@ -1345,28 +1340,19 @@ define <4 x i32> @combine_test9(<4 x i32> %a, <4 x i32> %b) {
 define <4 x i32> @combine_test10(<4 x i32> %a, <4 x i32> %b) {
 ; SSE2-LABEL: combine_test10:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movaps %xmm1, %xmm2
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[3,0],xmm2[2,0]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,1],xmm1[2,0]
-; SSE2-NEXT:    movaps %xmm2, %xmm0
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm1[0,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[2,3]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_test10:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movaps %xmm1, %xmm2
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[3,0],xmm2[2,0]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,1],xmm1[2,0]
-; SSSE3-NEXT:    movaps %xmm2, %xmm0
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm1[0,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[2,3]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_test10:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm0[2,3],xmm1[4,5,6,7]
-; SSE41-NEXT:    movdqa %xmm1, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3],xmm1[4,5,6,7]
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: combine_test10:
@@ -1395,22 +1381,19 @@ define <4 x float> @combine_test11(<4 x float> %a, <4 x float> %b) {
 define <4 x float> @combine_test12(<4 x float> %a, <4 x float> %b) {
 ; SSE2-LABEL: combine_test12:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movss %xmm0, %xmm1
-; SSE2-NEXT:    movss %xmm0, %xmm1
+; SSE2-NEXT:    movss {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
 ; SSE2-NEXT:    movaps %xmm1, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_test12:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movss %xmm0, %xmm1
-; SSSE3-NEXT:    movss %xmm0, %xmm1
+; SSSE3-NEXT:    movss {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
 ; SSSE3-NEXT:    movaps %xmm1, %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_test12:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    blendps {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
-; SSE41-NEXT:    movaps %xmm1, %xmm0
+; SSE41-NEXT:    blendps {{.*#+}} xmm0 = xmm0[0],xmm1[1,2,3]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_test12:
@@ -1455,26 +1438,19 @@ define <4 x float> @combine_test14(<4 x float> %a, <4 x float> %b) {
 define <4 x float> @combine_test15(<4 x float> %a, <4 x float> %b) {
 ; SSE2-LABEL: combine_test15:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movaps %xmm0, %xmm2
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,0],xmm1[0,0]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[2,0],xmm1[2,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm2[0,0]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm2[2,3]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm1[0,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[2,3]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_test15:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movaps %xmm0, %xmm2
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,0],xmm1[0,0]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[2,0],xmm1[2,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm2[0,0]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm2[2,3]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm1[0,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[2,3]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_test15:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    blendps {{.*#+}} xmm1 = xmm1[0],xmm0[1],xmm1[2,3]
-; SSE41-NEXT:    movaps %xmm1, %xmm0
+; SSE41-NEXT:    blendps {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2,3]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_test15:
@@ -1498,22 +1474,19 @@ define <4 x i32> @combine_test16(<4 x i32> %a, <4 x i32> %b) {
 define <4 x i32> @combine_test17(<4 x i32> %a, <4 x i32> %b) {
 ; SSE2-LABEL: combine_test17:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movss %xmm0, %xmm1
-; SSE2-NEXT:    movss %xmm0, %xmm1
+; SSE2-NEXT:    movss {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
 ; SSE2-NEXT:    movaps %xmm1, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_test17:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movss %xmm0, %xmm1
-; SSSE3-NEXT:    movss %xmm0, %xmm1
+; SSSE3-NEXT:    movss {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
 ; SSSE3-NEXT:    movaps %xmm1, %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_test17:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3,4,5,6,7]
-; SSE41-NEXT:    movdqa %xmm1, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3,4,5,6,7]
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: combine_test17:
@@ -1563,26 +1536,19 @@ define <4 x i32> @combine_test19(<4 x i32> %a, <4 x i32> %b) {
 define <4 x i32> @combine_test20(<4 x i32> %a, <4 x i32> %b) {
 ; SSE2-LABEL: combine_test20:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movaps %xmm0, %xmm2
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,0],xmm1[0,0]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[2,0],xmm1[2,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm2[0,0]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm2[2,3]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm1[0,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[2,3]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_test20:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movaps %xmm0, %xmm2
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,0],xmm1[0,0]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[2,0],xmm1[2,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm2[0,0]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm2[2,3]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm1[0,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[2,3]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_test20:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm0[2,3],xmm1[4,5,6,7]
-; SSE41-NEXT:    movdqa %xmm1, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3],xmm1[4,5,6,7]
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: combine_test20:
@@ -1599,45 +1565,70 @@ define <4 x i32> @combine_test20(<4 x i32> %a, <4 x i32> %b) {
   ret <4 x i32> %2
 }
 
+define <4 x i32> @combine_test21(<8 x i32> %a, <4 x i32>* %ptr) {
+; SSE-LABEL: combine_test21:
+; SSE:       # BB#0:
+; SSE-NEXT:    movdqa %xmm0, %xmm2
+; SSE-NEXT:    punpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm1[0]
+; SSE-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
+; SSE-NEXT:    movdqa %xmm2, (%rdi)
+; SSE-NEXT:    retq
+;
+; AVX1-LABEL: combine_test21:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm0[0],xmm1[0]
+; AVX1-NEXT:    vpunpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
+; AVX1-NEXT:    vmovdqa %xmm2, (%rdi)
+; AVX1-NEXT:    vzeroupper
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_test21:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vextracti128 $1, %ymm0, %xmm1
+; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm0[0],xmm1[0]
+; AVX2-NEXT:    vpunpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
+; AVX2-NEXT:    vmovdqa %xmm2, (%rdi)
+; AVX2-NEXT:    vzeroupper
+; AVX2-NEXT:    retq
+  %1 = shufflevector <8 x i32> %a, <8 x i32> %a, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
+  %2 = shufflevector <8 x i32> %a, <8 x i32> %a, <4 x i32> <i32 2, i32 3, i32 6, i32 7>
+  store <4 x i32> %1, <4 x i32>* %ptr, align 16
+  ret <4 x i32> %2
+}
+
+define <8 x float> @combine_test22(<2 x float>* %a, <2 x float>* %b) {
+; SSE-LABEL: combine_test22:
+; SSE:       # BB#0:
+; SSE-NEXT:    movq {{.*#+}} xmm0 = mem[0],zero
+; SSE-NEXT:    movhpd (%rsi), %xmm0
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: combine_test22:
+; AVX:       # BB#0:
+; AVX-NEXT:    vmovq {{.*#+}} xmm0 = mem[0],zero
+; AVX-NEXT:    vmovhpd (%rsi), %xmm0, %xmm0
+; AVX-NEXT:    retq
+; Current AVX2 lowering of this is still awful, not adding a test case.
+  %1 = load <2 x float>* %a, align 8
+  %2 = load <2 x float>* %b, align 8
+  %3 = shufflevector <2 x float> %1, <2 x float> %2, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 undef, i32 undef, i32 undef, i32 undef>
+  ret <8 x float> %3
+}
 
 ; Check some negative cases.
 ; FIXME: Do any of these really make sense? Are they redundant with the above tests?
 
 define <4 x float> @combine_test1b(<4 x float> %a, <4 x float> %b) {
-; SSE2-LABEL: combine_test1b:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movaps %xmm1, %xmm2
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm2[0,0]
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm2[2,0]
-; SSE2-NEXT:    movaps %xmm1, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: combine_test1b:
-; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movaps %xmm1, %xmm2
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm2[0,0]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm2[2,0]
-; SSSE3-NEXT:    movaps %xmm1, %xmm0
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: combine_test1b:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps %xmm1, %xmm2
-; SSE41-NEXT:    blendps {{.*#+}} xmm2 = xmm2[0],xmm0[1],xmm2[2],xmm0[3]
-; SSE41-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm2[0,0]
-; SSE41-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm2[2,0]
-; SSE41-NEXT:    movaps %xmm1, %xmm0
-; SSE41-NEXT:    retq
+; SSE-LABEL: combine_test1b:
+; SSE:       # BB#0:
+; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,1,2,0]
+; SSE-NEXT:    movaps %xmm1, %xmm0
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_test1b:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vblendps {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2],xmm0[3]
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[2,0],xmm0[2,0]
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm1[0,1,2,0]
 ; AVX-NEXT:    retq
   %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
   %2 = shufflevector <4 x float> %1, <4 x float> %b, <4 x i32> <i32 0, i32 5, i32 2, i32 0>
@@ -1647,35 +1638,23 @@ define <4 x float> @combine_test1b(<4 x float> %a, <4 x float> %b) {
 define <4 x float> @combine_test2b(<4 x float> %a, <4 x float> %b) {
 ; SSE2-LABEL: combine_test2b:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movaps %xmm1, %xmm2
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,0],xmm1[1,1]
-; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,1,3]
+; SSE2-NEXT:    movlhps {{.*#+}} xmm1 = xmm1[0,0]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_test2b:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movaps %xmm1, %xmm2
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,0],xmm1[1,1]
-; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,1,3]
+; SSSE3-NEXT:    movddup {{.*#+}} xmm0 = xmm1[0,0]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_test2b:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps %xmm1, %xmm2
-; SSE41-NEXT:    blendps {{.*#+}} xmm2 = xmm2[0],xmm0[1],xmm2[2],xmm0[3]
-; SSE41-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,0],xmm1[1,1]
-; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm2[0,2,1,3]
+; SSE41-NEXT:    movddup {{.*#+}} xmm0 = xmm1[0,0]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_test2b:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vblendps {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2],xmm0[3]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[1,1]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX-NEXT:    vmovddup {{.*#+}} xmm0 = xmm1[0,0]
 ; AVX-NEXT:    retq
   %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
   %2 = shufflevector <4 x float> %1, <4 x float> %b, <4 x i32> <i32 0, i32 5, i32 0, i32 5>
@@ -1683,21 +1662,28 @@ define <4 x float> @combine_test2b(<4 x float> %a, <4 x float> %b) {
 }
 
 define <4 x float> @combine_test3b(<4 x float> %a, <4 x float> %b) {
-; SSE-LABEL: combine_test3b:
-; SSE:       # BB#0:
-; SSE-NEXT:    movaps %xmm1, %xmm2
-; SSE-NEXT:    shufps {{.*#+}} xmm2 = xmm2[2,0],xmm0[3,0]
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,0],xmm2[0,2]
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[3,3]
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_test3b:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[3,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[2,3]
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: combine_test3b:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,0],xmm1[3,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[2,3]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_test3b:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    blendpd {{.*#+}} xmm0 = xmm0[0],xmm1[1]
+; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,3,2,3]
+; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_test3b:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm2 = xmm1[2,0],xmm0[3,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,0],xmm2[0,2]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[3,3]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX-NEXT:    vblendpd {{.*#+}} xmm0 = xmm0[0],xmm1[1]
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,3,2,3]
 ; AVX-NEXT:    retq
   %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 0, i32 0, i32 6, i32 3>
   %2 = shufflevector <4 x float> %1, <4 x float> %b, <4 x i32> <i32 0, i32 7, i32 2, i32 7>
@@ -1705,40 +1691,15 @@ define <4 x float> @combine_test3b(<4 x float> %a, <4 x float> %b) {
 }
 
 define <4 x float> @combine_test4b(<4 x float> %a, <4 x float> %b) {
-; SSE2-LABEL: combine_test4b:
-; SSE2:       # BB#0:
-; SSE2-NEXT:    movaps %xmm1, %xmm2
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[2,0],xmm1[3,0]
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,1],xmm2[0,2]
-; SSE2-NEXT:    movaps %xmm1, %xmm0
-; SSE2-NEXT:    retq
-;
-; SSSE3-LABEL: combine_test4b:
-; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movaps %xmm1, %xmm2
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[2,0],xmm1[3,0]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,1],xmm2[0,2]
-; SSSE3-NEXT:    movaps %xmm1, %xmm0
-; SSSE3-NEXT:    retq
-;
-; SSE41-LABEL: combine_test4b:
-; SSE41:       # BB#0:
-; SSE41-NEXT:    movaps %xmm1, %xmm2
-; SSE41-NEXT:    blendps {{.*#+}} xmm2 = xmm2[0],xmm0[1],xmm2[2],xmm0[3]
-; SSE41-NEXT:    shufps {{.*#+}} xmm2 = xmm2[2,0],xmm1[3,0]
-; SSE41-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,1],xmm2[0,2]
-; SSE41-NEXT:    movaps %xmm1, %xmm0
-; SSE41-NEXT:    retq
+; SSE-LABEL: combine_test4b:
+; SSE:       # BB#0:
+; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,1,2,3]
+; SSE-NEXT:    movaps %xmm1, %xmm0
+; SSE-NEXT:    retq
 ;
 ; AVX-LABEL: combine_test4b:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vblendps {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2],xmm0[3]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[3,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[1,1],xmm0[0,2]
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm1[1,1,2,3]
 ; AVX-NEXT:    retq
   %1 = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
   %2 = shufflevector <4 x float> %1, <4 x float> %b, <4 x i32> <i32 5, i32 5, i32 2, i32 7>
@@ -1751,50 +1712,44 @@ define <4 x float> @combine_test4b(<4 x float> %a, <4 x float> %b) {
 define <4 x i8> @combine_test1c(<4 x i8>* %a, <4 x i8>* %b) {
 ; SSE2-LABEL: combine_test1c:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movd (%rdi), %xmm0
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSE2-NEXT:    movd (%rsi), %xmm1
+; SSE2-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[3,0]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2]
+; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
+; SSE2-NEXT:    movss {{.*#+}} xmm0 = xmm1[0],xmm0[1,2,3]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_test1c:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movd (%rdi), %xmm0
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSSE3-NEXT:    movd (%rsi), %xmm1
+; SSSE3-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
 ; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
 ; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[3,0]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2]
+; SSSE3-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
+; SSSE3-NEXT:    movss {{.*#+}} xmm0 = xmm1[0],xmm0[1,2,3]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_test1c:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    pmovzxbd (%rdi), %xmm1
-; SSE41-NEXT:    pmovzxbd (%rsi), %xmm0
+; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
 ; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3,4,5,6,7]
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: combine_test1c:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vpmovzxbd (%rdi), %xmm0
-; AVX1-NEXT:    vpmovzxbd (%rsi), %xmm1
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3,4,5,6,7]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: combine_test1c:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpmovzxbd (%rdi), %xmm0
-; AVX2-NEXT:    vpmovzxbd (%rsi), %xmm1
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
 ; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1,2,3]
 ; AVX2-NEXT:    retq
   %A = load <4 x i8>* %a
@@ -1807,10 +1762,10 @@ define <4 x i8> @combine_test1c(<4 x i8>* %a, <4 x i8>* %b) {
 define <4 x i8> @combine_test2c(<4 x i8>* %a, <4 x i8>* %b) {
 ; SSE2-LABEL: combine_test2c:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movd (%rdi), %xmm0
+; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSE2-NEXT:    movd (%rsi), %xmm1
+; SSE2-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
 ; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
@@ -1818,10 +1773,10 @@ define <4 x i8> @combine_test2c(<4 x i8>* %a, <4 x i8>* %b) {
 ;
 ; SSSE3-LABEL: combine_test2c:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movd (%rdi), %xmm0
+; SSSE3-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
-; SSSE3-NEXT:    movd (%rsi), %xmm1
+; SSSE3-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
 ; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
 ; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
 ; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
@@ -1829,15 +1784,15 @@ define <4 x i8> @combine_test2c(<4 x i8>* %a, <4 x i8>* %b) {
 ;
 ; SSE41-LABEL: combine_test2c:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    pmovzxbd (%rdi), %xmm0
-; SSE41-NEXT:    pmovzxbd (%rsi), %xmm1
+; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
 ; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_test2c:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vpmovzxbd (%rdi), %xmm0
-; AVX-NEXT:    vpmovzxbd (%rsi), %xmm1
+; AVX-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
 ; AVX-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
 ; AVX-NEXT:    retq
   %A = load <4 x i8>* %a
@@ -1850,10 +1805,10 @@ define <4 x i8> @combine_test2c(<4 x i8>* %a, <4 x i8>* %b) {
 define <4 x i8> @combine_test3c(<4 x i8>* %a, <4 x i8>* %b) {
 ; SSE2-LABEL: combine_test3c:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movd (%rdi), %xmm1
+; SSE2-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSE2-NEXT:    movd (%rsi), %xmm0
+; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
 ; SSE2-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
@@ -1861,10 +1816,10 @@ define <4 x i8> @combine_test3c(<4 x i8>* %a, <4 x i8>* %b) {
 ;
 ; SSSE3-LABEL: combine_test3c:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movd (%rdi), %xmm1
+; SSSE3-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
 ; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
 ; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSSE3-NEXT:    movd (%rsi), %xmm0
+; SSSE3-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
 ; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
 ; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
 ; SSSE3-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
@@ -1872,15 +1827,15 @@ define <4 x i8> @combine_test3c(<4 x i8>* %a, <4 x i8>* %b) {
 ;
 ; SSE41-LABEL: combine_test3c:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    pmovzxbd (%rdi), %xmm1
-; SSE41-NEXT:    pmovzxbd (%rsi), %xmm0
+; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
 ; SSE41-NEXT:    punpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_test3c:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vpmovzxbd (%rdi), %xmm0
-; AVX-NEXT:    vpmovzxbd (%rsi), %xmm1
+; AVX-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
 ; AVX-NEXT:    vpunpckhqdq {{.*#+}} xmm0 = xmm1[1],xmm0[1]
 ; AVX-NEXT:    retq
   %A = load <4 x i8>* %a
@@ -1893,52 +1848,46 @@ define <4 x i8> @combine_test3c(<4 x i8>* %a, <4 x i8>* %b) {
 define <4 x i8> @combine_test4c(<4 x i8>* %a, <4 x i8>* %b) {
 ; SSE2-LABEL: combine_test4c:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movd (%rdi), %xmm1
+; SSE2-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSE2-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
+; SSE2-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
 ; SSE2-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
 ; SSE2-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSE2-NEXT:    movd (%rsi), %xmm2
-; SSE2-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSE2-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSE2-NEXT:    movdqa %xmm2, %xmm0
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[3,0],xmm0[2,0]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm1[0,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[2,3]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_test4c:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movd (%rdi), %xmm1
+; SSSE3-NEXT:    movd {{.*#+}} xmm0 = mem[0],zero,zero,zero
+; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3,4,4,5,5,6,6,7,7]
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0,0,1,1,2,2,3,3]
+; SSSE3-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
 ; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3],xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
 ; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm1 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
-; SSSE3-NEXT:    movd (%rsi), %xmm2
-; SSSE3-NEXT:    punpcklbw {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3],xmm2[4],xmm0[4],xmm2[5],xmm0[5],xmm2[6],xmm0[6],xmm2[7],xmm0[7]
-; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm2 = xmm2[0],xmm0[0],xmm2[1],xmm0[1],xmm2[2],xmm0[2],xmm2[3],xmm0[3]
-; SSSE3-NEXT:    movdqa %xmm2, %xmm0
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[3,0],xmm0[2,0]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,0],xmm1[0,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0],xmm1[2,3]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_test4c:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    pmovzxbd (%rdi), %xmm1
-; SSE41-NEXT:    pmovzxbd (%rsi), %xmm0
+; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; SSE41-NEXT:    pmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
 ; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: combine_test4c:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vpmovzxbd (%rdi), %xmm0
-; AVX1-NEXT:    vpmovzxbd (%rsi), %xmm1
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX1-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3],xmm1[4,5,6,7]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: combine_test4c:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpmovzxbd (%rdi), %xmm0
-; AVX2-NEXT:    vpmovzxbd (%rsi), %xmm1
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm0 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
+; AVX2-NEXT:    vpmovzxbd {{.*#+}} xmm1 = mem[0],zero,zero,zero,mem[1],zero,zero,zero,mem[2],zero,zero,zero,mem[3],zero,zero,zero
 ; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2,3]
 ; AVX2-NEXT:    retq
   %A = load <4 x i8>* %a
@@ -1981,24 +1930,17 @@ define <4 x i8> @combine_test4c(<4 x i8>* %a, <4 x i8>* %b) {
 define <4 x float> @combine_blend_01(<4 x float> %a, <4 x float> %b) {
 ; SSE2-LABEL: combine_blend_01:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movsd %xmm1, %xmm0
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[2,3]
-; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_blend_01:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movsd %xmm1, %xmm0
-; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,0],xmm0[0,0]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[2,3]
-; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_blend_01:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    blendpd {{.*#+}} xmm1 = xmm1[0],xmm0[1]
-; SSE41-NEXT:    movapd %xmm1, %xmm0
+; SSE41-NEXT:    blendpd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_blend_01:
@@ -2013,22 +1955,21 @@ define <4 x float> @combine_blend_01(<4 x float> %a, <4 x float> %b) {
 define <4 x float> @combine_blend_02(<4 x float> %a, <4 x float> %b) {
 ; SSE2-LABEL: combine_blend_02:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movss %xmm1, %xmm0
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[3,0]
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2]
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[1,3]
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,1,3]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_blend_02:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movss %xmm1, %xmm0
-; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[3,0]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[1,3]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,1,3]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_blend_02:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    blendps {{.*#+}} xmm1 = xmm1[0],xmm0[1],xmm1[2],xmm0[3]
-; SSE41-NEXT:    movaps %xmm1, %xmm0
+; SSE41-NEXT:    blendps {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2],xmm0[3]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_blend_02:
@@ -2043,26 +1984,19 @@ define <4 x float> @combine_blend_02(<4 x float> %a, <4 x float> %b) {
 define <4 x float> @combine_blend_123(<4 x float> %a, <4 x float> %b) {
 ; SSE2-LABEL: combine_blend_123:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movaps %xmm1, %xmm2
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,0],xmm0[0,0]
-; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[2,0],xmm0[2,3]
-; SSE2-NEXT:    movsd %xmm2, %xmm1
+; SSE2-NEXT:    movss {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
 ; SSE2-NEXT:    movaps %xmm1, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_blend_123:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movaps %xmm1, %xmm2
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,0],xmm0[0,0]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[2,0],xmm0[2,3]
-; SSSE3-NEXT:    movsd %xmm2, %xmm1
+; SSSE3-NEXT:    movss {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
 ; SSSE3-NEXT:    movaps %xmm1, %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_blend_123:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    blendps {{.*#+}} xmm1 = xmm0[0],xmm1[1,2,3]
-; SSE41-NEXT:    movaps %xmm1, %xmm0
+; SSE41-NEXT:    blendps {{.*#+}} xmm0 = xmm0[0],xmm1[1,2,3]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_blend_123:
@@ -2130,22 +2064,17 @@ define <4 x i32> @combine_test_movhl_3(<4 x i32> %a, <4 x i32> %b) {
 define <4 x float> @combine_undef_input_test1(<4 x float> %a, <4 x float> %b) {
 ; SSE2-LABEL: combine_undef_input_test1:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,3,1]
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,1],xmm0[1,2]
-; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_undef_input_test1:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,3,1]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,1],xmm0[1,2]
-; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_undef_input_test1:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    blendpd {{.*#+}} xmm1 = xmm1[0],xmm0[1]
-; SSE41-NEXT:    movapd %xmm1, %xmm0
+; SSE41-NEXT:    blendpd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_undef_input_test1:
@@ -2206,14 +2135,14 @@ define <4 x float> @combine_undef_input_test4(<4 x float> %a, <4 x float> %b) {
 define <4 x float> @combine_undef_input_test5(<4 x float> %a, <4 x float> %b) {
 ; SSE2-LABEL: combine_undef_input_test5:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movsd %xmm0, %xmm1
-; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
+; SSE2-NEXT:    movapd %xmm1, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_undef_input_test5:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movsd %xmm0, %xmm1
-; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
+; SSSE3-NEXT:    movapd %xmm1, %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_undef_input_test5:
@@ -2251,17 +2180,17 @@ define <4 x float> @combine_undef_input_test7(<4 x float> %a) {
 ;
 ; SSSE3-LABEL: combine_undef_input_test7:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    unpcklpd {{.*#+}} xmm0 = xmm0[0,0]
+; SSSE3-NEXT:    movddup {{.*#+}} xmm0 = xmm0[0,0]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_undef_input_test7:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    unpcklpd {{.*#+}} xmm0 = xmm0[0,0]
+; SSE41-NEXT:    movddup {{.*#+}} xmm0 = xmm0[0,0]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_undef_input_test7:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vunpcklpd {{.*#+}} xmm0 = xmm0[0,0]
+; AVX-NEXT:    vmovddup {{.*#+}} xmm0 = xmm0[0,0]
 ; AVX-NEXT:    retq
   %1 = shufflevector <4 x float> %a, <4 x float> undef, <4 x i32> <i32 6, i32 0, i32 1, i32 7>
   %2 = shufflevector <4 x float> %1, <4 x float> %a, <4 x i32> <i32 1, i32 2, i32 4, i32 5>
@@ -2276,17 +2205,17 @@ define <4 x float> @combine_undef_input_test8(<4 x float> %a) {
 ;
 ; SSSE3-LABEL: combine_undef_input_test8:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    unpcklpd {{.*#+}} xmm0 = xmm0[0,0]
+; SSSE3-NEXT:    movddup {{.*#+}} xmm0 = xmm0[0,0]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_undef_input_test8:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    unpcklpd {{.*#+}} xmm0 = xmm0[0,0]
+; SSE41-NEXT:    movddup {{.*#+}} xmm0 = xmm0[0,0]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_undef_input_test8:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vunpcklpd {{.*#+}} xmm0 = xmm0[0,0]
+; AVX-NEXT:    vmovddup {{.*#+}} xmm0 = xmm0[0,0]
 ; AVX-NEXT:    retq
   %1 = shufflevector <4 x float> %a, <4 x float> undef, <4 x i32> <i32 0, i32 5, i32 1, i32 7>
   %2 = shufflevector <4 x float> %1, <4 x float> %a, <4 x i32> <i32 0, i32 2, i32 4, i32 1>
@@ -2320,22 +2249,17 @@ define <4 x float> @combine_undef_input_test10(<4 x float> %a) {
 define <4 x float> @combine_undef_input_test11(<4 x float> %a, <4 x float> %b) {
 ; SSE2-LABEL: combine_undef_input_test11:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,3,1]
-; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,1],xmm0[1,2]
-; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_undef_input_test11:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,3,1]
-; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,1],xmm0[1,2]
-; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    movsd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_undef_input_test11:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    blendpd {{.*#+}} xmm1 = xmm1[0],xmm0[1]
-; SSE41-NEXT:    movapd %xmm1, %xmm0
+; SSE41-NEXT:    blendpd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_undef_input_test11:
@@ -2396,14 +2320,14 @@ define <4 x float> @combine_undef_input_test14(<4 x float> %a, <4 x float> %b) {
 define <4 x float> @combine_undef_input_test15(<4 x float> %a, <4 x float> %b) {
 ; SSE2-LABEL: combine_undef_input_test15:
 ; SSE2:       # BB#0:
-; SSE2-NEXT:    movsd %xmm0, %xmm1
-; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
+; SSE2-NEXT:    movapd %xmm1, %xmm0
 ; SSE2-NEXT:    retq
 ;
 ; SSSE3-LABEL: combine_undef_input_test15:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    movsd %xmm0, %xmm1
-; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    movsd {{.*#+}} xmm1 = xmm0[0],xmm1[1]
+; SSSE3-NEXT:    movapd %xmm1, %xmm0
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_undef_input_test15:
@@ -2447,17 +2371,17 @@ define <4 x float> @combine_undef_input_test17(<4 x float> %a) {
 ;
 ; SSSE3-LABEL: combine_undef_input_test17:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    unpcklpd {{.*#+}} xmm0 = xmm0[0,0]
+; SSSE3-NEXT:    movddup {{.*#+}} xmm0 = xmm0[0,0]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_undef_input_test17:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    unpcklpd {{.*#+}} xmm0 = xmm0[0,0]
+; SSE41-NEXT:    movddup {{.*#+}} xmm0 = xmm0[0,0]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_undef_input_test17:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vunpcklpd {{.*#+}} xmm0 = xmm0[0,0]
+; AVX-NEXT:    vmovddup {{.*#+}} xmm0 = xmm0[0,0]
 ; AVX-NEXT:    retq
   %1 = shufflevector <4 x float> %a, <4 x float> undef, <4 x i32> <i32 6, i32 0, i32 1, i32 7>
   %2 = shufflevector <4 x float> %a, <4 x float> %1, <4 x i32> <i32 5, i32 6, i32 0, i32 1>
@@ -2472,17 +2396,17 @@ define <4 x float> @combine_undef_input_test18(<4 x float> %a) {
 ;
 ; SSSE3-LABEL: combine_undef_input_test18:
 ; SSSE3:       # BB#0:
-; SSSE3-NEXT:    unpcklpd {{.*#+}} xmm0 = xmm0[0,0]
+; SSSE3-NEXT:    movddup {{.*#+}} xmm0 = xmm0[0,0]
 ; SSSE3-NEXT:    retq
 ;
 ; SSE41-LABEL: combine_undef_input_test18:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    unpcklpd {{.*#+}} xmm0 = xmm0[0,0]
+; SSE41-NEXT:    movddup {{.*#+}} xmm0 = xmm0[0,0]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_undef_input_test18:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vunpcklpd {{.*#+}} xmm0 = xmm0[0,0]
+; AVX-NEXT:    vmovddup {{.*#+}} xmm0 = xmm0[0,0]
 ; AVX-NEXT:    retq
   %1 = shufflevector <4 x float> %a, <4 x float> undef, <4 x i32> <i32 0, i32 5, i32 1, i32 7>
   %2 = shufflevector <4 x float> %a, <4 x float> %1, <4 x i32> <i32 4, i32 6, i32 0, i32 5>
@@ -2558,21 +2482,197 @@ define <8 x i32> @combine_unneeded_subvector2(<8 x i32> %a, <8 x i32> %b) {
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
 ; AVX1-NEXT:    vpaddd {{.*}}(%rip), %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
-; AVX1-NEXT:    vpermilps {{.*#+}} xmm1 = xmm1[3,2,1,0]
+; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm1[2,3],ymm0[2,3]
 ; AVX1-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX1-NEXT:    vblendpd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: combine_unneeded_subvector2:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpaddd {{.*}}(%rip), %ymm0, %ymm0
-; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <7,6,5,4,u,u,u,u>
-; AVX2-NEXT:    vpermd %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm1[2,3],ymm0[2,3]
 ; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
 ; AVX2-NEXT:    retq
   %c = add <8 x i32> %a, <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
   %d = shufflevector <8 x i32> %b, <8 x i32> %c, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 15, i32 14, i32 13, i32 12>
   ret <8 x i32> %d
 }
+
+define <4 x float> @combine_insertps1(<4 x float> %a, <4 x float> %b) {
+; SSE2-LABEL: combine_insertps1:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[1,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[2,3]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: combine_insertps1:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[1,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[2,3]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_insertps1:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    insertps {{.*#+}} xmm0 = xmm1[2],xmm0[1,2,3]
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: combine_insertps1:
+; AVX:       # BB#0:
+; AVX-NEXT:    vinsertps {{.*#+}} xmm0 = xmm1[2],xmm0[1,2,3]
+; AVX-NEXT:    retq
+
+  %c = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32><i32 0, i32 6, i32 2, i32 4>
+  %d = shufflevector <4 x float> %a, <4 x float> %c, <4 x i32> <i32 5, i32 1, i32 6, i32 3>
+  ret <4 x float> %d
+}
+
+define <4 x float> @combine_insertps2(<4 x float> %a, <4 x float> %b) {
+; SSE2-LABEL: combine_insertps2:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[0,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[2,3]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: combine_insertps2:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[0,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[2,3]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_insertps2:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0],xmm1[2],xmm0[2,3]
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: combine_insertps2:
+; AVX:       # BB#0:
+; AVX-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0],xmm1[2],xmm0[2,3]
+; AVX-NEXT:    retq
+
+  %c = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32><i32 0, i32 1, i32 6, i32 7>
+  %d = shufflevector <4 x float> %a, <4 x float> %c, <4 x i32> <i32 4, i32 6, i32 2, i32 3>
+  ret <4 x float> %d
+}
+
+define <4 x float> @combine_insertps3(<4 x float> %a, <4 x float> %b) {
+; SSE2-LABEL: combine_insertps3:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[3,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2]
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: combine_insertps3:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[3,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_insertps3:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0],xmm0[3]
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: combine_insertps3:
+; AVX:       # BB#0:
+; AVX-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0],xmm0[3]
+; AVX-NEXT:    retq
+
+  %c = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32><i32 0, i32 4, i32 2, i32 5>
+  %d = shufflevector <4 x float> %a, <4 x float> %c, <4 x i32><i32 4, i32 1, i32 5, i32 3>
+  ret <4 x float> %d
+}
+
+define <4 x float> @combine_insertps4(<4 x float> %a, <4 x float> %b) {
+; SSE2-LABEL: combine_insertps4:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[2,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,0]
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: combine_insertps4:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[2,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,0]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_insertps4:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[0]
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: combine_insertps4:
+; AVX:       # BB#0:
+; AVX-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[0]
+; AVX-NEXT:    retq
+
+  %c = shufflevector <4 x float> %a, <4 x float> %b, <4 x i32><i32 0, i32 4, i32 2, i32 5>
+  %d = shufflevector <4 x float> %a, <4 x float> %c, <4 x i32><i32 4, i32 1, i32 6, i32 5>
+  ret <4 x float> %d
+}
+
+define <4 x float> @PR22377(<4 x float> %a, <4 x float> %b) {
+; SSE-LABEL: PR22377:
+; SSE:       # BB#0: # %entry
+; SSE-NEXT:    movaps %xmm0, %xmm1
+; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,3,1,3]
+; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,0,2]
+; SSE-NEXT:    addps %xmm0, %xmm1
+; SSE-NEXT:    unpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; SSE-NEXT:    retq
+;
+; AVX-LABEL: PR22377:
+; AVX:       # BB#0: # %entry
+; AVX-NEXT:    vpermilps {{.*#+}} xmm1 = xmm0[1,3,1,3]
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,0,2]
+; AVX-NEXT:    vaddps %xmm0, %xmm1, %xmm1
+; AVX-NEXT:    vunpcklps {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1]
+; AVX-NEXT:    retq
+entry:
+  %s1 = shufflevector <4 x float> %a, <4 x float> undef, <4 x i32> <i32 1, i32 3, i32 1, i32 3>
+  %s2 = shufflevector <4 x float> %a, <4 x float> undef, <4 x i32> <i32 0, i32 2, i32 0, i32 2>
+  %r2 = fadd <4 x float> %s1, %s2
+  %s3 = shufflevector <4 x float> %s2, <4 x float> %r2, <4 x i32> <i32 0, i32 4, i32 1, i32 5>
+  ret <4 x float> %s3
+}
+
+define <4 x float> @PR22390(<4 x float> %a, <4 x float> %b) {
+; SSE2-LABEL: PR22390:
+; SSE2:       # BB#0: # %entry
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,0,1,2]
+; SSE2-NEXT:    movaps %xmm0, %xmm2
+; SSE2-NEXT:    movss {{.*#+}} xmm2 = xmm1[0],xmm2[1,2,3]
+; SSE2-NEXT:    addps %xmm0, %xmm2
+; SSE2-NEXT:    movaps %xmm2, %xmm0
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: PR22390:
+; SSSE3:       # BB#0: # %entry
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,0,1,2]
+; SSSE3-NEXT:    movaps %xmm0, %xmm2
+; SSSE3-NEXT:    movss {{.*#+}} xmm2 = xmm1[0],xmm2[1,2,3]
+; SSSE3-NEXT:    addps %xmm0, %xmm2
+; SSSE3-NEXT:    movaps %xmm2, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: PR22390:
+; SSE41:       # BB#0: # %entry
+; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[3,0,1,2]
+; SSE41-NEXT:    blendps {{.*#+}} xmm1 = xmm1[0],xmm0[1,2,3]
+; SSE41-NEXT:    addps %xmm1, %xmm0
+; SSE41-NEXT:    retq
+;
+; AVX-LABEL: PR22390:
+; AVX:       # BB#0: # %entry
+; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[3,0,1,2]
+; AVX-NEXT:    vblendps {{.*#+}} xmm1 = xmm1[0],xmm0[1,2,3]
+; AVX-NEXT:    vaddps %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    retq
+entry:
+  %s1 = shufflevector <4 x float> %a, <4 x float> undef, <4 x i32> <i32 3, i32 0, i32 1, i32 2>
+  %s2 = shufflevector <4 x float> %s1, <4 x float> %b, <4 x i32> <i32 4, i32 1, i32 2, i32 3>
+  %r2 = fadd <4 x float> %s1, %s2
+  ret <4 x float> %r2
+}