[x86] Teach the 128-bit vector shuffle lowering routines to take

advantage of the existence of a reasonable blend instruction.

The 256-bit vector shuffle lowering has leveraged the general technique
of decomposed shuffles and blends for quite some time, but this never
made it back into the 128-bit code, and there are a large number of
patterns where this is substantially better. For example, this removes
almost all domain crossing in vector shuffles that involve some blend
and some permutation with SSE4.1 and later. See the massive reduction
in 'shufps' for integer test cases in this commit.

This isn't perfect yet for a few reasons:

1) The v8i16 shuffle lowering continues to plague me. We don't always
   form an unpack-based blend when that would be better. But the wins
   pretty drastically outstrip the losses here.
2) The v16i8 shuffle lowering is just a disaster here. I never went and
   implemented blend support here for some terrible reason. I'll do
   that next probably. I've not updated it for now.

More variations on this technique are coming as well -- we don't
shuffle-into-unpack or shuffle-into-palignr, both of which would also be
profitable.

Note that some test cases grow significantly in the number of
instructions, but I expect to actually be faster. We use
pshufd+pshufd+blendw instead of a single shufps, but the pshufd's are
very likely to pipeline well (two ports on most modern intel chips) and
the blend is a *very* fast instruction. The domain switch penalty will
essentially always be more than a blend instruction, which is the only
increase in tree height.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@229350 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index e259c84562a7fec600a358b4c6efc8e1600667cf..7d79bb218c23831cc2a575513ab87e0e77d82cc1 100644 (file)
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -8580,7 +8580,10 @@ static SDValue lowerV2I64VectorShuffle(SDValue Op, SDValue V1, SDValue V2,
       return Insertion;
   }
 
-  if (Subtarget->hasSSE41())
+  // We have different paths for blend lowering, but they all must use the
+  // *exact* same predicate.
+  bool IsBlendSupported = Subtarget->hasSSE41();
+  if (IsBlendSupported)
     if (SDValue Blend = lowerVectorShuffleAsBlend(DL, MVT::v2i64, V1, V2, Mask,
                                                   Subtarget, DAG))
       return Blend;
@@ -8598,6 +8601,12 @@ static SDValue lowerV2I64VectorShuffle(SDValue Op, SDValue V1, SDValue V2,
             DL, MVT::v2i64, V1, V2, Mask, Subtarget, DAG))
       return Rotate;
 
+  // If we have direct support for blends, we should lower by decomposing into
+  // a permute. That will be faster than the domain cross.
+  if (IsBlendSupported)
+    return lowerVectorShuffleAsDecomposedShuffleBlend(DL, MVT::v2i64, V1, V2,
+                                                      Mask, DAG);
+
   // We implement this with SHUFPD which is pretty lame because it will likely
   // incur 2 cycles of stall for integer vectors on Nehalem and older chips.
   // However, all the alternatives are still more cycles and newer chips don't
@@ -8860,7 +8869,10 @@ static SDValue lowerV4I32VectorShuffle(SDValue Op, SDValue V1, SDValue V2,
                                                          Mask, Subtarget, DAG))
       return V;
 
-  if (Subtarget->hasSSE41())
+  // We have different paths for blend lowering, but they all must use the
+  // *exact* same predicate.
+  bool IsBlendSupported = Subtarget->hasSSE41();
+  if (IsBlendSupported)
     if (SDValue Blend = lowerVectorShuffleAsBlend(DL, MVT::v4i32, V1, V2, Mask,
                                                   Subtarget, DAG))
       return Blend;
@@ -8882,6 +8894,12 @@ static SDValue lowerV4I32VectorShuffle(SDValue Op, SDValue V1, SDValue V2,
             DL, MVT::v4i32, V1, V2, Mask, Subtarget, DAG))
       return Rotate;
 
+  // If we have direct support for blends, we should lower by decomposing into
+  // a permute. That will be faster than the domain cross.
+  if (IsBlendSupported)
+    return lowerVectorShuffleAsDecomposedShuffleBlend(DL, MVT::v4i32, V1, V2,
+                                                      Mask, DAG);
+
   // We implement this with SHUFPS because it can blend from two vectors.
   // Because we're going to eventually use SHUFPS, we use SHUFPS even to build
   // up the inputs, bypassing domain shift penalties that we would encur if we
@@ -9576,7 +9594,10 @@ static SDValue lowerV8I16VectorShuffle(SDValue Op, SDValue V1, SDValue V2,
                                                          Mask, Subtarget, DAG))
       return V;
 
-  if (Subtarget->hasSSE41())
+  // We have different paths for blend lowering, but they all must use the
+  // *exact* same predicate.
+  bool IsBlendSupported = Subtarget->hasSSE41();
+  if (IsBlendSupported)
     if (SDValue Blend = lowerVectorShuffleAsBlend(DL, MVT::v8i16, V1, V2, Mask,
                                                   Subtarget, DAG))
       return Blend;
@@ -9622,6 +9643,12 @@ static SDValue lowerV8I16VectorShuffle(SDValue Op, SDValue V1, SDValue V2,
     return DAG.getNode(X86ISD::UNPCKL, DL, MVT::v8i16, Evens, Odds);
   }
 
+  // If we have direct support for blends, we should lower by decomposing into
+  // a permute.
+  if (IsBlendSupported)
+    return lowerVectorShuffleAsDecomposedShuffleBlend(DL, MVT::v8i16, V1, V2,
+                                                      Mask, DAG);
+
   int LoBlendMask[8] = {-1, -1, -1, -1, -1, -1, -1, -1};
   int HiBlendMask[8] = {-1, -1, -1, -1, -1, -1, -1, -1};
 

diff --git a/test/CodeGen/X86/SwizzleShuff.ll b/test/CodeGen/X86/SwizzleShuff.ll
index a435272dca4432fbeb70020d6aa3ffcaff5e7fd9..d387850abb91908426be117f2d4a943c2626dc05 100644 (file)
--- a/test/CodeGen/X86/SwizzleShuff.ll
+++ b/test/CodeGen/X86/SwizzleShuff.ll
@@ -14,11 +14,12 @@ define void @pull_bitcast (<4 x i8>* %pA, <4 x i8>* %pB) {
 }
 
 ; CHECK: multi_use_swizzle
-; CHECK: mov
-; CHECK-NEXT: shuf
-; CHECK-NEXT: shuf
-; CHECK-NEXT: shuf
-; CHECK-NEXT: xor
+; CHECK: pshufd
+; CHECK-NEXT: pshufd
+; CHECK-NEXT: pblendw
+; CHECK-NEXT: pshufd
+; CHECK-NEXT: pshufd
+; CHECK-NEXT: pxor
 ; CHECK-NEXT: ret
 define <4 x i32> @multi_use_swizzle (<4 x i32>* %pA, <4 x i32>* %pB) {
   %A = load <4 x i32>* %pA
@@ -45,7 +46,7 @@ define <4 x i8> @pull_bitcast2 (<4 x i8>* %pA, <4 x i8>* %pB, <4 x i8>* %pC) {
 
 
 ; CHECK: reverse_1
-; CHECK-NOT: shuf
+; CHECK-NOT: pshufd
 ; CHECK: ret
 define <4 x i32> @reverse_1 (<4 x i32>* %pA, <4 x i32>* %pB) {
   %A = load <4 x i32>* %pA
@@ -57,7 +58,7 @@ define <4 x i32> @reverse_1 (<4 x i32>* %pA, <4 x i32>* %pB) {
 
 
 ; CHECK: no_reverse_shuff
-; CHECK: shuf
+; CHECK: pshufd
 ; CHECK: ret
 define <4 x i32> @no_reverse_shuff (<4 x i32>* %pA, <4 x i32>* %pB) {
   %A = load <4 x i32>* %pA

diff --git a/test/CodeGen/X86/avx-trunc.ll b/test/CodeGen/X86/avx-trunc.ll
index bf8d9a7f1a40e0976c48ce14fef265c3604891b9..27be9fd2fcd14778ba792cf6c2f44cd0c44b0e2b 100644 (file)
--- a/test/CodeGen/X86/avx-trunc.ll
+++ b/test/CodeGen/X86/avx-trunc.ll
@@ -2,9 +2,9 @@
 
 define <4 x i32> @trunc_64_32(<4 x i64> %A) nounwind uwtable readnone ssp{
 ; CHECK-LABEL: trunc_64_32
-; CHECK: shufps
-; CHECK-NOT: pshufd
-; CHECK-NOT: movlhps 
+; CHECK: pshufd
+; CHECK: pshufd
+; CHECK: pblendw
   %B = trunc <4 x i64> %A to <4 x i32>
   ret <4 x i32>%B
 }

diff --git a/test/CodeGen/X86/combine-or.ll b/test/CodeGen/X86/combine-or.ll
index ac1d5ef70926631726d451076c29dbf8f28b34d6..f9e917a937cc67d87c47bca89eccf7e47f691936 100644 (file)
--- a/test/CodeGen/X86/combine-or.ll
+++ b/test/CodeGen/X86/combine-or.ll
@@ -153,7 +153,8 @@ define <4 x i32> @test12(<4 x i32> %a, <4 x i32> %b) {
 define <4 x i32> @test13(<4 x i32> %a, <4 x i32> %b) {
 ; CHECK-LABEL: test13:
 ; CHECK:       # BB#0:
-; CHECK-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,1],xmm1[2,3]
+; CHECK-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,1,2,3]
+; CHECK-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
 ; CHECK-NEXT:    retq
   %shuf1 = shufflevector <4 x i32> %a, <4 x i32> zeroinitializer, <4 x i32><i32 1, i32 1, i32 4, i32 4>
   %shuf2 = shufflevector <4 x i32> %b, <4 x i32> zeroinitializer, <4 x i32><i32 4, i32 4, i32 2, i32 3>
@@ -177,8 +178,9 @@ define <2 x i64> @test14(<2 x i64> %a, <2 x i64> %b) {
 define <4 x i32> @test15(<4 x i32> %a, <4 x i32> %b) {
 ; CHECK-LABEL: test15:
 ; CHECK:       # BB#0:
-; CHECK-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,1],xmm0[2,1]
-; CHECK-NEXT:    movaps %xmm1, %xmm0
+; CHECK-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[0,1,2,1]
+; CHECK-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[2,1,2,3]
+; CHECK-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm2[4,5,6,7]
 ; CHECK-NEXT:    retq
   %shuf1 = shufflevector <4 x i32> %a, <4 x i32> zeroinitializer, <4 x i32><i32 4, i32 4, i32 2, i32 1>
   %shuf2 = shufflevector <4 x i32> %b, <4 x i32> zeroinitializer, <4 x i32><i32 2, i32 1, i32 4, i32 4>
@@ -236,12 +238,12 @@ define <4 x i32> @test18(<4 x i32> %a, <4 x i32> %b) {
 define <4 x i32> @test19(<4 x i32> %a, <4 x i32> %b) {
 ; CHECK-LABEL: test19:
 ; CHECK:       # BB#0:
-; CHECK-NEXT:    xorps %xmm2, %xmm2
-; CHECK-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[0,3]
-; CHECK-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; CHECK-NEXT:    insertps {{.*#+}} xmm1 = xmm1[0],zero,xmm1[2,2]
-; CHECK-NEXT:    orps %xmm1, %xmm2
-; CHECK-NEXT:    movaps %xmm2, %xmm0
+; CHECK-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[0,0,2,3]
+; CHECK-NEXT:    pxor %xmm3, %xmm3
+; CHECK-NEXT:    pblendw {{.*#+}} xmm2 = xmm3[0,1],xmm2[2,3],xmm3[4,5],xmm2[6,7]
+; CHECK-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,1,2,2]
+; CHECK-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm3[2,3],xmm0[4,5,6,7]
+; CHECK-NEXT:    por %xmm2, %xmm0
 ; CHECK-NEXT:    retq
   %shuf1 = shufflevector <4 x i32> %a, <4 x i32> zeroinitializer, <4 x i32><i32 4, i32 0, i32 4, i32 3>
   %shuf2 = shufflevector <4 x i32> %b, <4 x i32> zeroinitializer, <4 x i32><i32 0, i32 4, i32 2, i32 2>

diff --git a/test/CodeGen/X86/masked_memop.ll b/test/CodeGen/X86/masked_memop.ll
index b5ff630f497a44b6ad229272972e4e7bcf74d0aa..f268c57f65bbd969ad4b06a7aa431d17936713d6 100644 (file)
--- a/test/CodeGen/X86/masked_memop.ll
+++ b/test/CodeGen/X86/masked_memop.ll
@@ -150,7 +150,8 @@ define void @test13(<16 x i32> %trigger, <16 x float>* %addr, <16 x float> %val)
 }
 
 ; AVX2-LABEL: test14
-; AVX2: vinsertps {{.*#+}} xmm0 = xmm0[0,2],zero,zero
+; AVX2: vpshufd
+; AVX2: vmovq
 ; AVX2: vmaskmovps
 define void @test14(<2 x i32> %trigger, <2 x float>* %addr, <2 x float> %val) {
   %mask = icmp eq <2 x i32> %trigger, zeroinitializer

diff --git a/test/CodeGen/X86/sse41.ll b/test/CodeGen/X86/sse41.ll
index 5b72d79783599ce3c0d14604622941d4dbade745..189b4abc0e731fc019f2fb78e2a65a4e3e16ca75 100644 (file)
--- a/test/CodeGen/X86/sse41.ll
+++ b/test/CodeGen/X86/sse41.ll
@@ -356,12 +356,14 @@ define <4 x i32> @pinsrd_from_shufflevector_i32(<4 x i32> %a, <4 x i32>* nocaptu
 ; X32-LABEL: pinsrd_from_shufflevector_i32:
 ; X32:       ## BB#0: ## %entry
 ; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; X32-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0,1,2],mem[0]
+; X32-NEXT:    pshufd {{.*#+}} xmm1 = mem[0,1,2,0]
+; X32-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5],xmm1[6,7]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: pinsrd_from_shufflevector_i32:
 ; X64:       ## BB#0: ## %entry
-; X64-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0,1,2],mem[0]
+; X64-NEXT:    pshufd {{.*#+}} xmm1 = mem[0,1,2,0]
+; X64-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5],xmm1[6,7]
 ; X64-NEXT:    retq
 entry:
   %0 = load <4 x i32>* %pb, align 16
@@ -372,12 +374,14 @@ entry:
 define <4 x i32> @insertps_from_shufflevector_i32_2(<4 x i32> %a, <4 x i32> %b) {
 ; X32-LABEL: insertps_from_shufflevector_i32_2:
 ; X32:       ## BB#0: ## %entry
-; X32-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0],xmm1[3],xmm0[2,3]
+; X32-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
+; X32-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: insertps_from_shufflevector_i32_2:
 ; X64:       ## BB#0: ## %entry
-; X64-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0],xmm1[3],xmm0[2,3]
+; X64-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
+; X64-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
 ; X64-NEXT:    retq
 entry:
   %vecinit6 = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 0, i32 7, i32 2, i32 3>
@@ -407,13 +411,15 @@ define <4 x i32> @insertps_from_load_ins_elt_undef_i32(<4 x i32> %a, i32* %b) {
 ; X32:       ## BB#0:
 ; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
 ; X32-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
-; X32-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0],xmm0[3]
+; X32-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,0,1]
+; X32-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5],xmm0[6,7]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: insertps_from_load_ins_elt_undef_i32:
 ; X64:       ## BB#0:
 ; X64-NEXT:    movd {{.*#+}} xmm1 = mem[0],zero,zero,zero
-; X64-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0],xmm0[3]
+; X64-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,0,1]
+; X64-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5],xmm0[6,7]
 ; X64-NEXT:    retq
   %1 = load i32* %b, align 4
   %2 = insertelement <4 x i32> undef, i32 %1, i32 0
@@ -695,14 +701,16 @@ define <4 x i32> @i32_shuf_X00A(<4 x i32> %x, <4 x i32> %a) {
 ; X32:       ## BB#0:
 ; X32-NEXT:    pxor %xmm2, %xmm2
 ; X32-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3,4,5,6,7]
-; X32-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[0]
+; X32-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,2,0]
+; X32-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5],xmm1[6,7]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: i32_shuf_X00A:
 ; X64:       ## BB#0:
 ; X64-NEXT:    pxor %xmm2, %xmm2
 ; X64-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3,4,5,6,7]
-; X64-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[0]
+; X64-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,2,0]
+; X64-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5],xmm1[6,7]
 ; X64-NEXT:    retq
   %vecext = extractelement <4 x i32> %x, i32 0
   %vecinit = insertelement <4 x i32> undef, i32 %vecext, i32 0
@@ -717,16 +725,16 @@ define <4 x i32> @i32_shuf_X00X(<4 x i32> %x, <4 x i32> %a) {
 ; X32:       ## BB#0:
 ; X32-NEXT:    pxor %xmm1, %xmm1
 ; X32-NEXT:    pblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3,4,5,6,7]
-; X32-NEXT:    insertps {{.*#+}} xmm1 = xmm1[0,1,2],xmm0[0]
-; X32-NEXT:    movaps %xmm1, %xmm0
+; X32-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,0]
+; X32-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1,2,3,4,5],xmm0[6,7]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: i32_shuf_X00X:
 ; X64:       ## BB#0:
 ; X64-NEXT:    pxor %xmm1, %xmm1
 ; X64-NEXT:    pblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3,4,5,6,7]
-; X64-NEXT:    insertps {{.*#+}} xmm1 = xmm1[0,1,2],xmm0[0]
-; X64-NEXT:    movaps %xmm1, %xmm0
+; X64-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,0]
+; X64-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1,2,3,4,5],xmm0[6,7]
 ; X64-NEXT:    retq
   %vecext = extractelement <4 x i32> %x, i32 0
   %vecinit = insertelement <4 x i32> undef, i32 %vecext, i32 0
@@ -740,17 +748,17 @@ define <4 x i32> @i32_shuf_X0YC(<4 x i32> %x, <4 x i32> %a) {
 ; X32-LABEL: i32_shuf_X0YC:
 ; X32:       ## BB#0:
 ; X32-NEXT:    pmovzxdq {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero
-; X32-NEXT:    insertps {{.*#+}} xmm2 = xmm2[0,1],xmm0[1],zero
-; X32-NEXT:    insertps {{.*#+}} xmm2 = xmm2[0,1,2],xmm1[2]
-; X32-NEXT:    movaps %xmm2, %xmm0
+; X32-NEXT:    pblendw {{.*#+}} xmm2 = xmm2[0,1,2,3,4,5,6,7]
+; X32-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,1,2,2]
+; X32-NEXT:    pblendw {{.*#+}} xmm0 = xmm2[0,1,2,3,4,5],xmm0[6,7]
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: i32_shuf_X0YC:
 ; X64:       ## BB#0:
 ; X64-NEXT:    pmovzxdq {{.*#+}} xmm2 = xmm0[0],zero,xmm0[1],zero
-; X64-NEXT:    insertps {{.*#+}} xmm2 = xmm2[0,1],xmm0[1],zero
-; X64-NEXT:    insertps {{.*#+}} xmm2 = xmm2[0,1,2],xmm1[2]
-; X64-NEXT:    movaps %xmm2, %xmm0
+; X64-NEXT:    pblendw {{.*#+}} xmm2 = xmm2[0,1,2,3,4,5,6,7]
+; X64-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[0,1,2,2]
+; X64-NEXT:    pblendw {{.*#+}} xmm0 = xmm2[0,1,2,3,4,5],xmm0[6,7]
 ; X64-NEXT:    retq
   %vecext = extractelement <4 x i32> %x, i32 0
   %vecinit = insertelement <4 x i32> undef, i32 %vecext, i32 0
@@ -1003,16 +1011,16 @@ define void @insertps_pr20411(i32* noalias nocapture %RET) #1 {
 ; X32-LABEL: insertps_pr20411:
 ; X32:       ## BB#0:
 ; X32-NEXT:    movl {{[0-9]+}}(%esp), %eax
-; X32-NEXT:    movaps {{.*#+}} xmm0 = [3,3,3,3]
-; X32-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0],mem[3],zero,zero
-; X32-NEXT:    movups %xmm0, (%eax)
+; X32-NEXT:    pshufd {{.*#+}} xmm0 = mem[2,3,0,1]
+; X32-NEXT:    pblendw {{.*#+}} xmm0 = mem[0,1],xmm0[2,3],mem[4,5,6,7]
+; X32-NEXT:    movdqu %xmm0, (%eax)
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: insertps_pr20411:
 ; X64:       ## BB#0:
-; X64-NEXT:    movaps {{.*#+}} xmm0 = [3,3,3,3]
-; X64-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0],mem[3],zero,zero
-; X64-NEXT:    movups %xmm0, (%rdi)
+; X64-NEXT:    pshufd {{.*#+}} xmm0 = mem[2,3,0,1]
+; X64-NEXT:    pblendw {{.*#+}} xmm0 = mem[0,1],xmm0[2,3],mem[4,5,6,7]
+; X64-NEXT:    movdqu %xmm0, (%rdi)
 ; X64-NEXT:    retq
   %gather_load = shufflevector <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>, <8 x i32> undef, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   %shuffle109 = shufflevector <4 x i32> <i32 4, i32 5, i32 6, i32 7>, <4 x i32> undef, <4 x i32> <i32 0, i32 1, i32 2, i32 3>  ; 4 5 6 7

diff --git a/test/CodeGen/X86/vector-idiv.ll b/test/CodeGen/X86/vector-idiv.ll
index a98ed322136a261e462f122e23571dbab71a3238..5d1fe96b254b82ec06346644476db2232b966ac2 100644 (file)
--- a/test/CodeGen/X86/vector-idiv.ll
+++ b/test/CodeGen/X86/vector-idiv.ll
@@ -8,16 +8,15 @@ define <4 x i32> @test1(<4 x i32> %a) {
 ; SSE41-LABEL: test1:
 ; SSE41:       # BB#0:
 ; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [613566757,613566757,613566757,613566757]
-; SSE41-NEXT:    movdqa %xmm0, %xmm2
-; SSE41-NEXT:    pmuludq %xmm1, %xmm2
-; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
 ; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
-; SSE41-NEXT:    pmuludq %xmm1, %xmm3
-; SSE41-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,3],xmm3[1,3]
-; SSE41-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSE41-NEXT:    psubd %xmm2, %xmm0
+; SSE41-NEXT:    pmuludq %xmm2, %xmm3
+; SSE41-NEXT:    pmuludq %xmm0, %xmm1
+; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm3[2,3],xmm1[4,5],xmm3[6,7]
+; SSE41-NEXT:    psubd %xmm1, %xmm0
 ; SSE41-NEXT:    psrld $1, %xmm0
-; SSE41-NEXT:    paddd %xmm2, %xmm0
+; SSE41-NEXT:    paddd %xmm1, %xmm0
 ; SSE41-NEXT:    psrld $2, %xmm0
 ; SSE41-NEXT:    retq
 ;
@@ -40,12 +39,12 @@ define <4 x i32> @test1(<4 x i32> %a) {
 ; AVX-LABEL: test1:
 ; AVX:       # BB#0:
 ; AVX-NEXT:    vpbroadcastd {{.*}}(%rip), %xmm1
-; AVX-NEXT:    vpmuludq %xmm1, %xmm0, %xmm2
-; AVX-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; AVX-NEXT:    vpshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
 ; AVX-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
-; AVX-NEXT:    vpmuludq %xmm1, %xmm3, %xmm1
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm2[1,3],xmm1[1,3]
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,2,1,3]
+; AVX-NEXT:    vpmuludq %xmm2, %xmm3, %xmm2
+; AVX-NEXT:    vpmuludq %xmm1, %xmm0, %xmm1
+; AVX-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; AVX-NEXT:    vpblendd {{.*#+}} xmm1 = xmm1[0],xmm2[1],xmm1[2],xmm2[3]
 ; AVX-NEXT:    vpsubd %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    vpsrld $1, %xmm0, %xmm0
 ; AVX-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
@@ -59,22 +58,22 @@ define <8 x i32> @test2(<8 x i32> %a) {
 ; SSE41-LABEL: test2:
 ; SSE41:       # BB#0:
 ; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [613566757,613566757,613566757,613566757]
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    pmuludq %xmm2, %xmm3
-; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[1,1,3,3]
-; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm0[1,1,3,3]
-; SSE41-NEXT:    pmuludq %xmm4, %xmm5
-; SSE41-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,3],xmm5[1,3]
-; SSE41-NEXT:    shufps {{.*#+}} xmm3 = xmm3[0,2,1,3]
-; SSE41-NEXT:    psubd %xmm3, %xmm0
+; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm0[1,1,3,3]
+; SSE41-NEXT:    pmuludq %xmm3, %xmm4
+; SSE41-NEXT:    movdqa %xmm0, %xmm5
+; SSE41-NEXT:    pmuludq %xmm2, %xmm5
+; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm5 = xmm5[0,1],xmm4[2,3],xmm5[4,5],xmm4[6,7]
+; SSE41-NEXT:    psubd %xmm5, %xmm0
 ; SSE41-NEXT:    psrld $1, %xmm0
-; SSE41-NEXT:    paddd %xmm3, %xmm0
+; SSE41-NEXT:    paddd %xmm5, %xmm0
 ; SSE41-NEXT:    psrld $2, %xmm0
+; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[1,1,3,3]
+; SSE41-NEXT:    pmuludq %xmm3, %xmm4
 ; SSE41-NEXT:    pmuludq %xmm1, %xmm2
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[1,1,3,3]
-; SSE41-NEXT:    pmuludq %xmm4, %xmm3
-; SSE41-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,3],xmm3[1,3]
-; SSE41-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm2 = xmm2[0,1],xmm4[2,3],xmm2[4,5],xmm4[6,7]
 ; SSE41-NEXT:    psubd %xmm2, %xmm1
 ; SSE41-NEXT:    psrld $1, %xmm1
 ; SSE41-NEXT:    paddd %xmm2, %xmm1
@@ -822,14 +821,13 @@ define <16 x i8> @test7(<16 x i8> %a) {
 define <4 x i32> @test8(<4 x i32> %a) {
 ; SSE41-LABEL: test8:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [2454267027,2454267027,2454267027,2454267027]
-; SSE41-NEXT:    movdqa %xmm0, %xmm1
-; SSE41-NEXT:    pmuldq %xmm2, %xmm1
-; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [2454267027,2454267027,2454267027,2454267027]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
 ; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
 ; SSE41-NEXT:    pmuldq %xmm2, %xmm3
-; SSE41-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,3],xmm3[1,3]
-; SSE41-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,1,3]
+; SSE41-NEXT:    pmuldq %xmm0, %xmm1
+; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm3[2,3],xmm1[4,5],xmm3[6,7]
 ; SSE41-NEXT:    paddd %xmm0, %xmm1
 ; SSE41-NEXT:    movdqa %xmm1, %xmm0
 ; SSE41-NEXT:    psrld $31, %xmm0
@@ -866,12 +864,12 @@ define <4 x i32> @test8(<4 x i32> %a) {
 ; AVX-LABEL: test8:
 ; AVX:       # BB#0:
 ; AVX-NEXT:    vpbroadcastd {{.*}}(%rip), %xmm1
-; AVX-NEXT:    vpmuldq %xmm1, %xmm0, %xmm2
-; AVX-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; AVX-NEXT:    vpshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
 ; AVX-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
-; AVX-NEXT:    vpmuldq %xmm1, %xmm3, %xmm1
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm2[1,3],xmm1[1,3]
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,2,1,3]
+; AVX-NEXT:    vpmuldq %xmm2, %xmm3, %xmm2
+; AVX-NEXT:    vpmuldq %xmm1, %xmm0, %xmm1
+; AVX-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
+; AVX-NEXT:    vpblendd {{.*#+}} xmm1 = xmm1[0],xmm2[1],xmm1[2],xmm2[3]
 ; AVX-NEXT:    vpaddd %xmm0, %xmm1, %xmm0
 ; AVX-NEXT:    vpsrld $31, %xmm0, %xmm1
 ; AVX-NEXT:    vpsrad $2, %xmm0, %xmm0
@@ -884,31 +882,31 @@ define <4 x i32> @test8(<4 x i32> %a) {
 define <8 x i32> @test9(<8 x i32> %a) {
 ; SSE41-LABEL: test9:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    movdqa %xmm1, %xmm2
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [2454267027,2454267027,2454267027,2454267027]
-; SSE41-NEXT:    # kill: XMM0<def> XMM3<kill>
-; SSE41-NEXT:    pmuldq %xmm1, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[1,1,3,3]
-; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm3[1,1,3,3]
+; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [2454267027,2454267027,2454267027,2454267027]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm3[1,1,3,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm0[1,1,3,3]
 ; SSE41-NEXT:    pmuldq %xmm4, %xmm5
-; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,3],xmm5[1,3]
-; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; SSE41-NEXT:    paddd %xmm3, %xmm0
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    psrld $31, %xmm3
-; SSE41-NEXT:    psrad $2, %xmm0
-; SSE41-NEXT:    paddd %xmm3, %xmm0
-; SSE41-NEXT:    pmuldq %xmm2, %xmm1
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
-; SSE41-NEXT:    pmuldq %xmm4, %xmm3
-; SSE41-NEXT:    shufps {{.*#+}} xmm1 = xmm1[1,3],xmm3[1,3]
-; SSE41-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2,1,3]
-; SSE41-NEXT:    paddd %xmm2, %xmm1
-; SSE41-NEXT:    movdqa %xmm1, %xmm2
-; SSE41-NEXT:    psrld $31, %xmm2
-; SSE41-NEXT:    psrad $2, %xmm1
-; SSE41-NEXT:    paddd %xmm2, %xmm1
+; SSE41-NEXT:    movdqa %xmm0, %xmm2
+; SSE41-NEXT:    pmuldq %xmm3, %xmm2
+; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm2 = xmm2[0,1],xmm5[2,3],xmm2[4,5],xmm5[6,7]
+; SSE41-NEXT:    paddd %xmm0, %xmm2
+; SSE41-NEXT:    movdqa %xmm2, %xmm0
+; SSE41-NEXT:    psrld $31, %xmm0
+; SSE41-NEXT:    psrad $2, %xmm2
+; SSE41-NEXT:    paddd %xmm0, %xmm2
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm1[1,1,3,3]
+; SSE41-NEXT:    pmuldq %xmm4, %xmm0
+; SSE41-NEXT:    pmuldq %xmm1, %xmm3
+; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm3[1,1,3,3]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm3 = xmm3[0,1],xmm0[2,3],xmm3[4,5],xmm0[6,7]
+; SSE41-NEXT:    paddd %xmm1, %xmm3
+; SSE41-NEXT:    movdqa %xmm3, %xmm0
+; SSE41-NEXT:    psrld $31, %xmm0
+; SSE41-NEXT:    psrad $2, %xmm3
+; SSE41-NEXT:    paddd %xmm0, %xmm3
+; SSE41-NEXT:    movdqa %xmm2, %xmm0
+; SSE41-NEXT:    movdqa %xmm3, %xmm1
 ; SSE41-NEXT:    retq
 ;
 ; SSE-LABEL: test9:
@@ -977,33 +975,33 @@ define <8 x i32> @test10(<8 x i32> %a) {
 ; SSE41-LABEL: test10:
 ; SSE41:       # BB#0:
 ; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [613566757,613566757,613566757,613566757]
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    pmuludq %xmm2, %xmm3
-; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[1,1,3,3]
-; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm0[1,1,3,3]
-; SSE41-NEXT:    pmuludq %xmm4, %xmm5
-; SSE41-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,3],xmm5[1,3]
-; SSE41-NEXT:    shufps {{.*#+}} xmm3 = xmm3[0,2,1,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm0[1,1,3,3]
+; SSE41-NEXT:    pmuludq %xmm3, %xmm4
 ; SSE41-NEXT:    movdqa %xmm0, %xmm5
-; SSE41-NEXT:    psubd %xmm3, %xmm5
-; SSE41-NEXT:    psrld $1, %xmm5
-; SSE41-NEXT:    paddd %xmm3, %xmm5
-; SSE41-NEXT:    psrld $2, %xmm5
-; SSE41-NEXT:    movdqa {{.*#+}} xmm3 = [7,7,7,7]
-; SSE41-NEXT:    pmulld %xmm3, %xmm5
-; SSE41-NEXT:    psubd %xmm5, %xmm0
-; SSE41-NEXT:    pmuludq %xmm1, %xmm2
-; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm1[1,1,3,3]
-; SSE41-NEXT:    pmuludq %xmm4, %xmm5
-; SSE41-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,3],xmm5[1,3]
-; SSE41-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
-; SSE41-NEXT:    movdqa %xmm1, %xmm4
-; SSE41-NEXT:    psubd %xmm2, %xmm4
+; SSE41-NEXT:    pmuludq %xmm2, %xmm5
+; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm5 = xmm5[0,1],xmm4[2,3],xmm5[4,5],xmm4[6,7]
+; SSE41-NEXT:    movdqa %xmm0, %xmm4
+; SSE41-NEXT:    psubd %xmm5, %xmm4
 ; SSE41-NEXT:    psrld $1, %xmm4
-; SSE41-NEXT:    paddd %xmm2, %xmm4
+; SSE41-NEXT:    paddd %xmm5, %xmm4
 ; SSE41-NEXT:    psrld $2, %xmm4
-; SSE41-NEXT:    pmulld %xmm3, %xmm4
-; SSE41-NEXT:    psubd %xmm4, %xmm1
+; SSE41-NEXT:    movdqa {{.*#+}} xmm5 = [7,7,7,7]
+; SSE41-NEXT:    pmulld %xmm5, %xmm4
+; SSE41-NEXT:    psubd %xmm4, %xmm0
+; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm1[1,1,3,3]
+; SSE41-NEXT:    pmuludq %xmm3, %xmm4
+; SSE41-NEXT:    pmuludq %xmm1, %xmm2
+; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm2 = xmm2[0,1],xmm4[2,3],xmm2[4,5],xmm4[6,7]
+; SSE41-NEXT:    movdqa %xmm1, %xmm3
+; SSE41-NEXT:    psubd %xmm2, %xmm3
+; SSE41-NEXT:    psrld $1, %xmm3
+; SSE41-NEXT:    paddd %xmm2, %xmm3
+; SSE41-NEXT:    psrld $2, %xmm3
+; SSE41-NEXT:    pmulld %xmm5, %xmm3
+; SSE41-NEXT:    psubd %xmm3, %xmm1
 ; SSE41-NEXT:    retq
 ;
 ; SSE-LABEL: test10:
@@ -1071,32 +1069,32 @@ define <8 x i32> @test11(<8 x i32> %a) {
 ; SSE41-LABEL: test11:
 ; SSE41:       # BB#0:
 ; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [2454267027,2454267027,2454267027,2454267027]
-; SSE41-NEXT:    movdqa %xmm0, %xmm3
-; SSE41-NEXT:    pmuldq %xmm2, %xmm3
-; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm2[1,1,3,3]
-; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm0[1,1,3,3]
-; SSE41-NEXT:    pmuldq %xmm4, %xmm5
-; SSE41-NEXT:    shufps {{.*#+}} xmm3 = xmm3[1,3],xmm5[1,3]
-; SSE41-NEXT:    shufps {{.*#+}} xmm3 = xmm3[0,2,1,3]
-; SSE41-NEXT:    paddd %xmm0, %xmm3
-; SSE41-NEXT:    movdqa %xmm3, %xmm5
-; SSE41-NEXT:    psrld $31, %xmm5
-; SSE41-NEXT:    psrad $2, %xmm3
-; SSE41-NEXT:    paddd %xmm5, %xmm3
-; SSE41-NEXT:    movdqa {{.*#+}} xmm5 = [7,7,7,7]
-; SSE41-NEXT:    pmulld %xmm5, %xmm3
-; SSE41-NEXT:    psubd %xmm3, %xmm0
+; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm2[1,1,3,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm4 = xmm0[1,1,3,3]
+; SSE41-NEXT:    pmuldq %xmm3, %xmm4
+; SSE41-NEXT:    movdqa %xmm0, %xmm5
+; SSE41-NEXT:    pmuldq %xmm2, %xmm5
+; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm5[1,1,3,3]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm5 = xmm5[0,1],xmm4[2,3],xmm5[4,5],xmm4[6,7]
+; SSE41-NEXT:    paddd %xmm0, %xmm5
+; SSE41-NEXT:    movdqa %xmm5, %xmm4
+; SSE41-NEXT:    psrld $31, %xmm4
+; SSE41-NEXT:    psrad $2, %xmm5
+; SSE41-NEXT:    paddd %xmm4, %xmm5
+; SSE41-NEXT:    movdqa {{.*#+}} xmm4 = [7,7,7,7]
+; SSE41-NEXT:    pmulld %xmm4, %xmm5
+; SSE41-NEXT:    psubd %xmm5, %xmm0
+; SSE41-NEXT:    pshufd {{.*#+}} xmm5 = xmm1[1,1,3,3]
+; SSE41-NEXT:    pmuldq %xmm3, %xmm5
 ; SSE41-NEXT:    pmuldq %xmm1, %xmm2
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[1,1,3,3]
-; SSE41-NEXT:    pmuldq %xmm4, %xmm3
-; SSE41-NEXT:    shufps {{.*#+}} xmm2 = xmm2[1,3],xmm3[1,3]
-; SSE41-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2,1,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm2[1,1,3,3]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm2 = xmm2[0,1],xmm5[2,3],xmm2[4,5],xmm5[6,7]
 ; SSE41-NEXT:    paddd %xmm1, %xmm2
 ; SSE41-NEXT:    movdqa %xmm2, %xmm3
 ; SSE41-NEXT:    psrld $31, %xmm3
 ; SSE41-NEXT:    psrad $2, %xmm2
 ; SSE41-NEXT:    paddd %xmm3, %xmm2
-; SSE41-NEXT:    pmulld %xmm5, %xmm2
+; SSE41-NEXT:    pmulld %xmm4, %xmm2
 ; SSE41-NEXT:    psubd %xmm2, %xmm1
 ; SSE41-NEXT:    retq
 ;
@@ -1201,16 +1199,15 @@ define <4 x i32> @PR20355(<4 x i32> %a) {
 ; SSE41-LABEL: PR20355:
 ; SSE41:       # BB#0: # %entry
 ; SSE41-NEXT:    movdqa {{.*#+}} xmm1 = [1431655766,1431655766,1431655766,1431655766]
-; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[1,1,3,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
+; SSE41-NEXT:    pmuldq %xmm2, %xmm3
 ; SSE41-NEXT:    pmuldq %xmm1, %xmm0
-; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; SSE41-NEXT:    pmuldq %xmm2, %xmm1
-; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[1,3],xmm1[1,3]
-; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
-; SSE41-NEXT:    movaps %xmm0, %xmm1
-; SSE41-NEXT:    psrld $31, %xmm1
-; SSE41-NEXT:    paddd %xmm0, %xmm1
+; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[1,1,3,3]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm1 = xmm1[0,1],xmm3[2,3],xmm1[4,5],xmm3[6,7]
 ; SSE41-NEXT:    movdqa %xmm1, %xmm0
+; SSE41-NEXT:    psrld $31, %xmm0
+; SSE41-NEXT:    paddd %xmm1, %xmm0
 ; SSE41-NEXT:    retq
 ;
 ; SSE-LABEL: PR20355:
@@ -1239,12 +1236,12 @@ define <4 x i32> @PR20355(<4 x i32> %a) {
 ; AVX-LABEL: PR20355:
 ; AVX:       # BB#0: # %entry
 ; AVX-NEXT:    vpbroadcastd {{.*}}(%rip), %xmm1
-; AVX-NEXT:    vpmuldq %xmm1, %xmm0, %xmm2
-; AVX-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,1,3,3]
-; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; AVX-NEXT:    vpshufd {{.*#+}} xmm2 = xmm1[1,1,3,3]
+; AVX-NEXT:    vpshufd {{.*#+}} xmm3 = xmm0[1,1,3,3]
+; AVX-NEXT:    vpmuldq %xmm2, %xmm3, %xmm2
 ; AVX-NEXT:    vpmuldq %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm2[1,3],xmm0[1,3]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,1,3,3]
+; AVX-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm2[1],xmm0[2],xmm2[3]
 ; AVX-NEXT:    vpsrld $31, %xmm0, %xmm1
 ; AVX-NEXT:    vpaddd %xmm1, %xmm0, %xmm0
 ; AVX-NEXT:    retq

diff --git a/test/CodeGen/X86/vector-shuffle-128-v4.ll b/test/CodeGen/X86/vector-shuffle-128-v4.ll
index e0267b29c7d718ed694f1c107a698d50ee3307d8..261d7c9eb0b0d2bddf3e551b7b9559b85f8ec60b 100644 (file)
--- a/test/CodeGen/X86/vector-shuffle-128-v4.ll
+++ b/test/CodeGen/X86/vector-shuffle-128-v4.ll
@@ -322,60 +322,150 @@ define <4 x i32> @shuffle_v4i32_0124(<4 x i32> %a, <4 x i32> %b) {
 ;
 ; SSE41-LABEL: shuffle_v4i32_0124:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    insertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[0]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,2,0]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5],xmm1[6,7]
 ; SSE41-NEXT:    retq
 ;
-; AVX-LABEL: shuffle_v4i32_0124:
-; AVX:       # BB#0:
-; AVX-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[0]
-; AVX-NEXT:    retq
+; AVX1-LABEL: shuffle_v4i32_0124:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,1,2,0]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5],xmm1[6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4i32_0124:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpbroadcastd %xmm1, %xmm1
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 0, i32 1, i32 2, i32 4>
   ret <4 x i32> %shuffle
 }
 define <4 x i32> @shuffle_v4i32_0142(<4 x i32> %a, <4 x i32> %b) {
-; SSE-LABEL: shuffle_v4i32_0142:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[2,0]
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2]
-; SSE-NEXT:    retq
+; SSE2-LABEL: shuffle_v4i32_0142:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[2,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2]
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: shuffle_v4i32_0142:
-; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[2,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2]
-; AVX-NEXT:    retq
+; SSE3-LABEL: shuffle_v4i32_0142:
+; SSE3:       # BB#0:
+; SSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[2,0]
+; SSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2]
+; SSE3-NEXT:    retq
+;
+; SSSE3-LABEL: shuffle_v4i32_0142:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[2,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,2]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: shuffle_v4i32_0142:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,0,1]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,2,2]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5],xmm0[6,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: shuffle_v4i32_0142:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,1,0,1]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,2,2]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5],xmm0[6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4i32_0142:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpbroadcastq %xmm1, %xmm1
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,2,2]
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2],xmm0[3]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 2>
   ret <4 x i32> %shuffle
 }
 define <4 x i32> @shuffle_v4i32_0412(<4 x i32> %a, <4 x i32> %b) {
-; SSE-LABEL: shuffle_v4i32_0412:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[1,2]
-; SSE-NEXT:    movaps %xmm1, %xmm0
-; SSE-NEXT:    retq
+; SSE2-LABEL: shuffle_v4i32_0412:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[1,2]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: shuffle_v4i32_0412:
-; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[2,0],xmm0[1,2]
-; AVX-NEXT:    retq
+; SSE3-LABEL: shuffle_v4i32_0412:
+; SSE3:       # BB#0:
+; SSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
+; SSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[1,2]
+; SSE3-NEXT:    movaps %xmm1, %xmm0
+; SSE3-NEXT:    retq
+;
+; SSSE3-LABEL: shuffle_v4i32_0412:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[1,2]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: shuffle_v4i32_0412:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,1,2]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: shuffle_v4i32_0412:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,1,2]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4i32_0412:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpbroadcastd %xmm1, %xmm1
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,1,2]
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 0, i32 4, i32 1, i32 2>
   ret <4 x i32> %shuffle
 }
 define <4 x i32> @shuffle_v4i32_4012(<4 x i32> %a, <4 x i32> %b) {
-; SSE-LABEL: shuffle_v4i32_4012:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[1,2]
-; SSE-NEXT:    movaps %xmm1, %xmm0
-; SSE-NEXT:    retq
+; SSE2-LABEL: shuffle_v4i32_4012:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[1,2]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: shuffle_v4i32_4012:
-; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[0,2],xmm0[1,2]
-; AVX-NEXT:    retq
+; SSE3-LABEL: shuffle_v4i32_4012:
+; SSE3:       # BB#0:
+; SSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
+; SSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[1,2]
+; SSE3-NEXT:    movaps %xmm1, %xmm0
+; SSE3-NEXT:    retq
+;
+; SSSE3-LABEL: shuffle_v4i32_4012:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[1,2]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: shuffle_v4i32_4012:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,0,1,2]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3,4,5,6,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: shuffle_v4i32_4012:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,2]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3,4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4i32_4012:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,2]
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0],xmm0[1,2,3]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 2>
   ret <4 x i32> %shuffle
 }
@@ -393,17 +483,44 @@ define <4 x i32> @shuffle_v4i32_0145(<4 x i32> %a, <4 x i32> %b) {
   ret <4 x i32> %shuffle
 }
 define <4 x i32> @shuffle_v4i32_0451(<4 x i32> %a, <4 x i32> %b) {
-; SSE-LABEL: shuffle_v4i32_0451:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,1]
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,3,1]
-; SSE-NEXT:    retq
+; SSE2-LABEL: shuffle_v4i32_0451:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,1]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,3,1]
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: shuffle_v4i32_0451:
-; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,1]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2,3,1]
-; AVX-NEXT:    retq
+; SSE3-LABEL: shuffle_v4i32_0451:
+; SSE3:       # BB#0:
+; SSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,1]
+; SSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,3,1]
+; SSE3-NEXT:    retq
+;
+; SSSE3-LABEL: shuffle_v4i32_0451:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,1]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2,3,1]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: shuffle_v4i32_0451:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3,4,5],xmm0[6,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: shuffle_v4i32_0451:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3,4,5],xmm0[6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4i32_0451:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
+; AVX2-NEXT:    vpbroadcastq %xmm0, %xmm0
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1,2],xmm0[3]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 0, i32 4, i32 5, i32 1>
   ret <4 x i32> %shuffle
 }
@@ -422,17 +539,44 @@ define <4 x i32> @shuffle_v4i32_4501(<4 x i32> %a, <4 x i32> %b) {
   ret <4 x i32> %shuffle
 }
 define <4 x i32> @shuffle_v4i32_4015(<4 x i32> %a, <4 x i32> %b) {
-; SSE-LABEL: shuffle_v4i32_4015:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,1]
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0,1,3]
-; SSE-NEXT:    retq
+; SSE2-LABEL: shuffle_v4i32_4015:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,1]
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0,1,3]
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: shuffle_v4i32_4015:
-; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,1]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[2,0,1,3]
-; AVX-NEXT:    retq
+; SSE3-LABEL: shuffle_v4i32_4015:
+; SSE3:       # BB#0:
+; SSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,1]
+; SSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0,1,3]
+; SSE3-NEXT:    retq
+;
+; SSSE3-LABEL: shuffle_v4i32_4015:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,1],xmm1[0,1]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,0,1,3]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: shuffle_v4i32_4015:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,0,1]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3,4,5],xmm1[6,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: shuffle_v4i32_4015:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,1,0,1]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3,4,5],xmm1[6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4i32_4015:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpbroadcastq %xmm1, %xmm1
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0],xmm0[1,2],xmm1[3]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 4, i32 0, i32 1, i32 5>
   ret <4 x i32> %shuffle
 }
@@ -977,13 +1121,24 @@ define <4 x i32> @shuffle_v4i32_z6zz(<4 x i32> %a) {
 ;
 ; SSE41-LABEL: shuffle_v4i32_z6zz:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    insertps {{.*#+}} xmm0 = zero,xmm0[2],zero,zero
+; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[2,2,3,3]
+; SSE41-NEXT:    pxor %xmm0, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
 ; SSE41-NEXT:    retq
 ;
-; AVX-LABEL: shuffle_v4i32_z6zz:
-; AVX:       # BB#0:
-; AVX-NEXT:    vinsertps {{.*#+}} xmm0 = zero,xmm0[2],zero,zero
-; AVX-NEXT:    retq
+; AVX1-LABEL: shuffle_v4i32_z6zz:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,2,3,3]
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3],xmm1[4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4i32_z6zz:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,2,3,3]
+; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2,3]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x i32> zeroinitializer, <4 x i32> %a, <4 x i32> <i32 0, i32 6, i32 2, i32 3>
   ret <4 x i32> %shuffle
 }
@@ -1150,18 +1305,44 @@ define <4 x i32> @shuffle_v4i32_2345(<4 x i32> %a, <4 x i32> %b) {
 }
 
 define <4 x i32> @shuffle_v4i32_40u1(<4 x i32> %a, <4 x i32> %b) {
-; SSE-LABEL: shuffle_v4i32_40u1:
-; SSE:       # BB#0:
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[2,1]
-; SSE-NEXT:    movaps %xmm1, %xmm0
-; SSE-NEXT:    retq
+; SSE2-LABEL: shuffle_v4i32_40u1:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[2,1]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: shuffle_v4i32_40u1:
-; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[0,2],xmm0[2,1]
-; AVX-NEXT:    retq
+; SSE3-LABEL: shuffle_v4i32_40u1:
+; SSE3:       # BB#0:
+; SSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
+; SSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[2,1]
+; SSE3-NEXT:    movaps %xmm1, %xmm0
+; SSE3-NEXT:    retq
+;
+; SSSE3-LABEL: shuffle_v4i32_40u1:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[0,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,2],xmm0[2,1]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: shuffle_v4i32_40u1:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3,4,5,6,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: shuffle_v4i32_40u1:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3,4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v4i32_40u1:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0],xmm0[1,2,3]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x i32> %a, <4 x i32> %b, <4 x i32> <i32 4, i32 0, i32 undef, i32 1>
   ret <4 x i32> %shuffle
 }

diff --git a/test/CodeGen/X86/vector-shuffle-128-v8.ll b/test/CodeGen/X86/vector-shuffle-128-v8.ll
index e73e6cadd79b53e6edd1824b92d9080f2d7ba2f8..6241e84bd18cbf3ad801bf650ce75778477ee27c 100644 (file)
--- a/test/CodeGen/X86/vector-shuffle-128-v8.ll
+++ b/test/CodeGen/X86/vector-shuffle-128-v8.ll
@@ -949,24 +949,46 @@ define <8 x i16> @shuffle_v8i16_0c1d6879(<8 x i16> %a, <8 x i16> %b) {
 }
 
 define <8 x i16> @shuffle_v8i16_109832ba(<8 x i16> %a, <8 x i16> %b) {
-; SSE-LABEL: shuffle_v8i16_109832ba:
-; SSE:       # BB#0:
-; SSE-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm0[2,0,3,1,4,5,6,7]
-; SSE-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,0,3,1,4,5,6,7]
-; SSE-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm0[0]
-; SSE-NEXT:    movdqa %xmm1, %xmm0
-; SSE-NEXT:    retq
+; SSE2-LABEL: shuffle_v8i16_109832ba:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm1 = xmm0[2,0,3,1,4,5,6,7]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,0,3,1,4,5,6,7]
+; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm0[0]
+; SSE2-NEXT:    movdqa %xmm1, %xmm0
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: shuffle_v8i16_109832ba:
-; AVX:       # BB#0:
-; AVX-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm0[2,0,3,1,4,5,6,7]
-; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
-; AVX-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[2,0,3,1,4,5,6,7]
-; AVX-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: shuffle_v8i16_109832ba:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
+; SSSE3-NEXT:    pshuflw {{.*#+}} xmm1 = xmm0[2,0,3,1,4,5,6,7]
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
+; SSSE3-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[2,0,3,1,4,5,6,7]
+; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm0[0]
+; SSSE3-NEXT:    movdqa %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: shuffle_v8i16_109832ba:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,1,2,3,2,3,0,1,8,9,10,11,6,7,4,5]
+; SSE41-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[2,3,0,1,4,5,6,7,6,7,4,5,4,5,6,7]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5],xmm1[6,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: shuffle_v8i16_109832ba:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,2,3,2,3,0,1,8,9,10,11,6,7,4,5]
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[2,3,0,1,4,5,6,7,6,7,4,5,4,5,6,7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5],xmm1[6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v8i16_109832ba:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,2,3,2,3,0,1,8,9,10,11,6,7,4,5]
+; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[2,3,0,1,4,5,6,7,6,7,4,5,4,5,6,7]
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 1, i32 0, i32 9, i32 8, i32 3, i32 2, i32 11, i32 10>
   ret <8 x i16> %shuffle
 }
@@ -1001,21 +1023,42 @@ define <8 x i16> @shuffle_v8i16_c4d5e6f7(<8 x i16> %a, <8 x i16> %b) {
 }
 
 define <8 x i16> @shuffle_v8i16_0213cedf(<8 x i16> %a, <8 x i16> %b) {
-; SSE-LABEL: shuffle_v8i16_0213cedf:
-; SSE:       # BB#0:
-; SSE-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
-; SSE-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
-; SSE-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,2,1,3,4,5,6,7]
-; SSE-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; SSE-NEXT:    retq
+; SSE2-LABEL: shuffle_v8i16_0213cedf:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
+; SSE2-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
+; SSE2-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,2,1,3,4,5,6,7]
+; SSE2-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: shuffle_v8i16_0213cedf:
-; AVX:       # BB#0:
-; AVX-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
-; AVX-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
-; AVX-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,2,1,3,4,5,6,7]
-; AVX-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: shuffle_v8i16_0213cedf:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
+; SSSE3-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[2,3,2,3]
+; SSSE3-NEXT:    pshuflw {{.*#+}} xmm1 = xmm1[0,2,1,3,4,5,6,7]
+; SSSE3-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: shuffle_v8i16_0213cedf:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
+; SSE41-NEXT:    pshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
+; SSE41-NEXT:    pshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,5,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: shuffle_v8i16_0213cedf:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
+; AVX1-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,5,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v8i16_0213cedf:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
+; AVX2-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
+; AVX2-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,5,7]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 2, i32 1, i32 3, i32 12, i32 14, i32 13, i32 15>
   ret <8 x i16> %shuffle
 }
@@ -1257,35 +1300,20 @@ define <8 x i16> @shuffle_v8i16_012dcde3(<8 x i16> %a, <8 x i16> %b) {
 ;
 ; SSE41-LABEL: shuffle_v8i16_012dcde3:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    pshufd {{.*#+}} xmm2 = xmm0[0,1,0,1]
-; SSE41-NEXT:    pshufd {{.*#+}} xmm3 = xmm1[2,3,0,1]
-; SSE41-NEXT:    punpckhwd {{.*#+}} xmm1 = xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
-; SSE41-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,8,9,14,15,8,9,14,15,12,13,14,15]
-; SSE41-NEXT:    punpcklwd {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1],xmm0[2],xmm3[2],xmm0[3],xmm3[3]
-; SSE41-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,6,7,8,9,0,1,0,1,2,3]
-; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
+; SSE41-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,10,11,8,9,10,11,12,13,6,7]
 ; SSE41-NEXT:    retq
 ;
 ; AVX1-LABEL: shuffle_v8i16_012dcde3:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm0[0,1,0,1]
-; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm2 = xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm2 = xmm2[0,1,4,5,8,9,14,15,8,9,14,15,12,13,14,15]
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,6,7,8,9,0,1,0,1,2,3]
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,10,11,8,9,10,11,12,13,6,7]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v8i16_012dcde3:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vpbroadcastq %xmm0, %xmm2
-; AVX2-NEXT:    vpunpckhwd {{.*#+}} xmm2 = xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm2 = xmm2[0,1,4,5,8,9,14,15,8,9,14,15,12,13,14,15]
-; AVX2-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
-; AVX2-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,6,7,8,9,0,1,0,1,2,3]
-; AVX2-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm2[0]
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3]
+; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,10,11,8,9,10,11,12,13,6,7]
 ; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x i16> %a, <8 x i16> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 13, i32 12, i32 13, i32 14, i32 3>
   ret <8 x i16> %shuffle

diff --git a/test/CodeGen/X86/vector-shuffle-256-v16.ll b/test/CodeGen/X86/vector-shuffle-256-v16.ll
index e2fbde55b65af0e17c15407fc48bbd3e3ba1998c..deb3da1685c2c4aae13079a08401bd22748f3fc1 100644 (file)
--- a/test/CodeGen/X86/vector-shuffle-256-v16.ll
+++ b/test/CodeGen/X86/vector-shuffle-256-v16.ll
@@ -150,11 +150,12 @@ define <16 x i16> @shuffle_v16i16_00_00_00_00_00_00_00_08_00_00_00_00_00_00_00_0
 ; AVX1-LABEL: shuffle_v16i16_00_00_00_00_00_00_00_08_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,1,4,5,6,7]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm0[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
-; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
+; AVX1-NEXT:    vpslldq {{.*#+}} xmm1 = zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,zero,xmm1[0,1]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,3]
+; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
+; AVX1-NEXT:    vpshufhw {{.*#+}} xmm2 = xmm0[0,1,2,3,4,4,4,7]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0,1,2,3,4,5,6],xmm1[7]
+; AVX1-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,4,4]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
@@ -174,10 +175,8 @@ define <16 x i16> @shuffle_v16i16_00_00_00_00_00_00_09_00_00_00_00_00_00_00_00_0
 ; AVX1-LABEL: shuffle_v16i16_00_00_00_00_00_00_09_00_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,3,0,4,5,6,7]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm0[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0],xmm1[1],xmm0[2,3,4,5,6,7]
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,0,1,0,1,0,1,0,1,0,1,2,3,0,1]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
@@ -197,11 +196,8 @@ define <16 x i16> @shuffle_v16i16_00_00_00_00_00_10_00_00_00_00_00_00_00_00_00_0
 ; AVX1-LABEL: shuffle_v16i16_00_00_00_00_00_10_00_00_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,3,0,0,4,5,6,7]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm0[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,0,1,0,1,0,1,0,1,4,5,0,1,0,1]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
@@ -220,11 +216,8 @@ define <16 x i16> @shuffle_v16i16_00_00_00_00_11_00_00_00_00_00_00_00_00_00_00_0
 ; AVX1-LABEL: shuffle_v16i16_00_00_00_00_11_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,3,2,3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[3,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm0[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm2[0],xmm1[0]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,0,1,0,1,0,1,6,7,0,1,0,1,0,1]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
@@ -243,11 +236,8 @@ define <16 x i16> @shuffle_v16i16_00_00_00_12_00_00_00_00_00_00_00_00_00_00_00_0
 ; AVX1-LABEL: shuffle_v16i16_00_00_00_12_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,1,4,5,6,7]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm0[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0,1,2,3],xmm1[4,5,6,7]
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,0,1,0,1,8,9,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
@@ -266,11 +256,8 @@ define <16 x i16> @shuffle_v16i16_00_00_13_00_00_00_00_00_00_00_00_00_00_00_00_0
 ; AVX1-LABEL: shuffle_v16i16_00_00_13_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,3,0,4,5,6,7]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm0[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0,1,2,3],xmm1[4,5,6,7]
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,0,1,10,11,0,1,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
@@ -289,12 +276,8 @@ define <16 x i16> @shuffle_v16i16_00_14_00_00_00_00_00_00_00_00_00_00_00_00_00_0
 ; AVX1-LABEL: shuffle_v16i16_00_14_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,3,0,0,4,5,6,7]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm0[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0,1,2,3],xmm1[4,5,6,7]
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,12,13,0,1,0,1,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
@@ -313,12 +296,8 @@ define <16 x i16> @shuffle_v16i16_15_00_00_00_00_00_00_00_00_00_00_00_00_00_00_0
 ; AVX1-LABEL: shuffle_v16i16_15_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[2,3,0,1]
-; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,3,2,3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[3,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm0[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm1 = xmm1[0],xmm2[0]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm0[0,1,2,3],xmm1[4,5,6,7]
+; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[14,15,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
@@ -755,13 +734,11 @@ define <16 x i16> @shuffle_v16i16_16_16_16_16_04_05_06_07_24_24_24_24_12_13_14_1
 ; AVX1-LABEL: shuffle_v16i16_16_16_16_16_04_05_06_07_24_24_24_24_12_13_14_15:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm2 = xmm2[2,3,0,1]
 ; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm3
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm3 = xmm3[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm3[0],xmm2[0]
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm3[0,1,2,3],xmm2[4,5,6,7]
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3],xmm0[4,5,6,7]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
@@ -777,17 +754,15 @@ define <16 x i16> @shuffle_v16i16_16_16_16_16_04_05_06_07_24_24_24_24_12_13_14_1
 define <16 x i16> @shuffle_v16i16_19_18_17_16_07_06_05_04_27_26_25_24_15_14_13_12(<16 x i16> %a, <16 x i16> %b) {
 ; AVX1-LABEL: shuffle_v16i16_19_18_17_16_07_06_05_04_27_26_25_24_15_14_13_12:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm1[0,1,2,3],xmm0[4,5,6,7]
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[3,2,1,0,4,5,6,7]
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm3
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm3[2,3,2,3]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm3 = xmm3[3,2,1,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm2[0],xmm3[0]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[3,2,1,0,4,5,6,7]
-; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,2,3]
+; AVX1-NEXT:    vpshufhw {{.*#+}} xmm2 = xmm2[0,1,2,3,7,6,5,4]
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3],xmm0[4,5,6,7]
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[3,2,1,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
-; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,6,5,4]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm2, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_19_18_17_16_07_06_05_04_27_26_25_24_15_14_13_12:
@@ -802,14 +777,16 @@ define <16 x i16> @shuffle_v16i16_19_18_17_16_07_06_05_04_27_26_25_24_15_14_13_1
 define <16 x i16> @shuffle_v16i16_19_18_17_16_03_02_01_00_27_26_25_24_11_10_09_08(<16 x i16> %a, <16 x i16> %b) {
 ; AVX1-LABEL: shuffle_v16i16_19_18_17_16_03_02_01_00_27_26_25_24_11_10_09_08:
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
+; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm2 = xmm2[3,2,1,0,4,5,6,7]
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm3
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm3 = xmm3[3,2,1,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm2 = xmm3[0],xmm2[0]
-; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[3,2,1,0,4,5,6,7]
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm3
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm3 = xmm3[0,1,0,1]
+; AVX1-NEXT:    vpshufhw {{.*#+}} xmm3 = xmm3[0,1,2,3,7,6,5,4]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm2 = xmm2[0,1,2,3],xmm3[4,5,6,7]
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[3,2,1,0,4,5,6,7]
-; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm1[0],xmm0[0]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
+; AVX1-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,7,6,5,4]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3],xmm0[4,5,6,7]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;

diff --git a/test/CodeGen/X86/vector-shuffle-combining.ll b/test/CodeGen/X86/vector-shuffle-combining.ll
index a7dc193cae6eaa28f0b73f63a6ae089b649e69f5..048b8b40649175b078e5070b829884f6f2f31081 100644 (file)
--- a/test/CodeGen/X86/vector-shuffle-combining.ll
+++ b/test/CodeGen/X86/vector-shuffle-combining.ll
@@ -507,17 +507,38 @@ define <4 x i32> @combine_bitwise_ops_test6b(<4 x i32> %a, <4 x i32> %b, <4 x i3
 }
 
 define <4 x i32> @combine_bitwise_ops_test1c(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
-; SSE-LABEL: combine_bitwise_ops_test1c:
-; SSE:       # BB#0:
-; SSE-NEXT:    andps %xmm1, %xmm0
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_bitwise_ops_test1c:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    andps %xmm1, %xmm0
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_bitwise_ops_test1c:
-; AVX:       # BB#0:
-; AVX-NEXT:    vandps %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_bitwise_ops_test1c:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    andps %xmm1, %xmm0
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_bitwise_ops_test1c:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pand %xmm1, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3],xmm0[4,5],xmm2[6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_bitwise_ops_test1c:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3],xmm0[4,5],xmm2[6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_bitwise_ops_test1c:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm2[1],xmm0[2],xmm2[3]
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX2-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>
   %and = and <4 x i32> %shuf1, %shuf2
@@ -525,17 +546,38 @@ define <4 x i32> @combine_bitwise_ops_test1c(<4 x i32> %a, <4 x i32> %b, <4 x i3
 }
 
 define <4 x i32> @combine_bitwise_ops_test2c(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
-; SSE-LABEL: combine_bitwise_ops_test2c:
-; SSE:       # BB#0:
-; SSE-NEXT:    orps %xmm1, %xmm0
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_bitwise_ops_test2c:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    orps %xmm1, %xmm0
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_bitwise_ops_test2c:
-; AVX:       # BB#0:
-; AVX-NEXT:    vorps %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_bitwise_ops_test2c:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    orps %xmm1, %xmm0
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm2[1,3]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_bitwise_ops_test2c:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3],xmm0[4,5],xmm2[6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_bitwise_ops_test2c:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm2[2,3],xmm0[4,5],xmm2[6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_bitwise_ops_test2c:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpor %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm2[1],xmm0[2],xmm2[3]
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX2-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>
   %or = or <4 x i32> %shuf1, %shuf2
@@ -559,14 +601,16 @@ define <4 x i32> @combine_bitwise_ops_test3c(<4 x i32> %a, <4 x i32> %b, <4 x i3
 ;
 ; 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:    pxor %xmm1, %xmm0
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SSE41-NEXT:    movq {{.*#+}} xmm0 = xmm0[0],zero
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: combine_bitwise_ops_test3c:
 ; AVX:       # BB#0:
-; AVX-NEXT:    vxorps %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vinsertps {{.*#+}} xmm0 = xmm0[0,2],zero,zero
+; AVX-NEXT:    vpxor %xmm1, %xmm0, %xmm0
+; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX-NEXT:    vmovq {{.*#+}} xmm0 = xmm0[0],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>
@@ -575,18 +619,40 @@ define <4 x i32> @combine_bitwise_ops_test3c(<4 x i32> %a, <4 x i32> %b, <4 x i3
 }
 
 define <4 x i32> @combine_bitwise_ops_test4c(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
-; SSE-LABEL: combine_bitwise_ops_test4c:
-; SSE:       # BB#0:
-; SSE-NEXT:    andps %xmm1, %xmm0
-; SSE-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSE-NEXT:    movaps %xmm2, %xmm0
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_bitwise_ops_test4c:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    andps %xmm1, %xmm0
+; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
+; SSE2-NEXT:    movaps %xmm2, %xmm0
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_bitwise_ops_test4c:
-; AVX:       # BB#0:
-; AVX-NEXT:    vandps %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm2[0,2],xmm0[1,3]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_bitwise_ops_test4c:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    andps %xmm1, %xmm0
+; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
+; SSSE3-NEXT:    movaps %xmm2, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_bitwise_ops_test4c:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pand %xmm1, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm2[0,1],xmm0[2,3],xmm2[4,5],xmm0[6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_bitwise_ops_test4c:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm2[0,1],xmm0[2,3],xmm2[4,5],xmm0[6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_bitwise_ops_test4c:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpand %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm2[0],xmm0[1],xmm2[2],xmm0[3]
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX2-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>
   %and = and <4 x i32> %shuf1, %shuf2
@@ -594,18 +660,40 @@ define <4 x i32> @combine_bitwise_ops_test4c(<4 x i32> %a, <4 x i32> %b, <4 x i3
 }
 
 define <4 x i32> @combine_bitwise_ops_test5c(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
-; SSE-LABEL: combine_bitwise_ops_test5c:
-; SSE:       # BB#0:
-; SSE-NEXT:    orps %xmm1, %xmm0
-; SSE-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
-; SSE-NEXT:    movaps %xmm2, %xmm0
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_bitwise_ops_test5c:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    orps %xmm1, %xmm0
+; SSE2-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
+; SSE2-NEXT:    movaps %xmm2, %xmm0
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_bitwise_ops_test5c:
-; AVX:       # BB#0:
-; AVX-NEXT:    vorps %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm2[0,2],xmm0[1,3]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_bitwise_ops_test5c:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    orps %xmm1, %xmm0
+; SSSE3-NEXT:    shufps {{.*#+}} xmm2 = xmm2[0,2],xmm0[1,3]
+; SSSE3-NEXT:    movaps %xmm2, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_bitwise_ops_test5c:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    por %xmm1, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm2[0,1],xmm0[2,3],xmm2[4,5],xmm0[6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_bitwise_ops_test5c:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpor %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm2[0,1],xmm0[2,3],xmm2[4,5],xmm0[6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_bitwise_ops_test5c:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpor %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm2[0],xmm0[1],xmm2[2],xmm0[3]
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,1,3]
+; AVX2-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>
   %or = or <4 x i32> %shuf1, %shuf2
@@ -613,20 +701,45 @@ define <4 x i32> @combine_bitwise_ops_test5c(<4 x i32> %a, <4 x i32> %b, <4 x i3
 }
 
 define <4 x i32> @combine_bitwise_ops_test6c(<4 x i32> %a, <4 x i32> %b, <4 x i32> %c) {
-; SSE-LABEL: combine_bitwise_ops_test6c:
-; SSE:       # BB#0:
-; SSE-NEXT:    xorps %xmm1, %xmm0
-; SSE-NEXT:    xorps %xmm1, %xmm1
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,1],xmm0[1,3]
-; SSE-NEXT:    movaps %xmm1, %xmm0
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_bitwise_ops_test6c:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    xorps %xmm1, %xmm0
+; SSE2-NEXT:    xorps %xmm1, %xmm1
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,1],xmm0[1,3]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_bitwise_ops_test6c:
-; AVX:       # BB#0:
-; AVX-NEXT:    vxorps %xmm1, %xmm0, %xmm0
-; AVX-NEXT:    vxorps %xmm1, %xmm1, %xmm1
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[0,1],xmm0[1,3]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_bitwise_ops_test6c:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    xorps %xmm1, %xmm0
+; SSSE3-NEXT:    xorps %xmm1, %xmm1
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,1],xmm0[1,3]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_bitwise_ops_test6c:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pxor %xmm1, %xmm0
+; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm0[0,1,1,3]
+; SSE41-NEXT:    pxor %xmm0, %xmm0
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_bitwise_ops_test6c:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpxor %xmm1, %xmm0, %xmm0
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,1,3]
+; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3],xmm0[4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_bitwise_ops_test6c:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpxor %xmm1, %xmm0, %xmm0
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,1,3]
+; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3]
+; AVX2-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>
   %xor = xor <4 x i32> %shuf1, %shuf2
@@ -857,18 +970,40 @@ define <4 x i32> @combine_nested_undef_test14(<4 x i32> %A, <4 x i32> %B) {
 ; it.
 
 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[3,0]
-; SSE-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[0,1]
-; SSE-NEXT:    movaps %xmm1, %xmm0
-; SSE-NEXT:    retq
+; SSE2-LABEL: combine_nested_undef_test15:
+; SSE2:       # BB#0:
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[3,0]
+; SSE2-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[0,1]
+; SSE2-NEXT:    movaps %xmm1, %xmm0
+; SSE2-NEXT:    retq
 ;
-; AVX-LABEL: combine_nested_undef_test15:
-; AVX:       # BB#0:
-; AVX-NEXT:    vshufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[3,0]
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm1[2,0],xmm0[0,1]
-; AVX-NEXT:    retq
+; SSSE3-LABEL: combine_nested_undef_test15:
+; SSSE3:       # BB#0:
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[0,0],xmm0[3,0]
+; SSSE3-NEXT:    shufps {{.*#+}} xmm1 = xmm1[2,0],xmm0[0,1]
+; SSSE3-NEXT:    movaps %xmm1, %xmm0
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: combine_nested_undef_test15:
+; SSE41:       # BB#0:
+; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,0,1]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
+; SSE41-NEXT:    retq
+;
+; AVX1-LABEL: combine_nested_undef_test15:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,1,0,1]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_nested_undef_test15:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpbroadcastd %xmm1, %xmm1
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,1,0,1]
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
+; AVX2-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>
   ret <4 x i32> %2
@@ -889,15 +1024,21 @@ define <4 x i32> @combine_nested_undef_test16(<4 x i32> %A, <4 x i32> %B) {
 ;
 ; SSE41-LABEL: combine_nested_undef_test16:
 ; SSE41:       # BB#0:
-; SSE41-NEXT:    blendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
-; SSE41-NEXT:    shufps {{.*#+}} xmm0 = xmm0[2,1,0,3]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5],xmm1[6,7]
 ; SSE41-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
+; AVX1-LABEL: combine_nested_undef_test16:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5],xmm1[6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_nested_undef_test16:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[2,3,0,1]
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
+; AVX2-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
@@ -918,15 +1059,21 @@ define <4 x i32> @combine_nested_undef_test17(<4 x i32> %A, <4 x i32> %B) {
 ;
 ; 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:    pblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3,4,5,6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[3,1,0,1]
 ; SSE41-NEXT:    retq
 ;
-; AVX-LABEL: combine_nested_undef_test17:
-; AVX:       # BB#0:
-; AVX-NEXT:    vblendps {{.*#+}} xmm0 = xmm1[0],xmm0[1,2,3]
-; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[3,1,0,1]
-; AVX-NEXT:    retq
+; AVX1-LABEL: combine_nested_undef_test17:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3,4,5,6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,1,0,1]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_nested_undef_test17:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0],xmm0[1,2,3]
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,1,0,1]
+; AVX2-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>
   ret <4 x i32> %2
@@ -964,15 +1111,21 @@ define <4 x i32> @combine_nested_undef_test19(<4 x i32> %A, <4 x i32> %B) {
 ;
 ; 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:    pblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[1,0,0,0]
 ; SSE41-NEXT:    retq
 ;
-; AVX-LABEL: combine_nested_undef_test19:
-; AVX:       # BB#0:
-; AVX-NEXT:    vblendps {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
-; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[1,0,0,0]
-; AVX-NEXT:    retq
+; AVX1-LABEL: combine_nested_undef_test19:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,0,0,0]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_nested_undef_test19:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3]
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,0,0,0]
+; AVX2-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>
   ret <4 x i32> %2
@@ -995,15 +1148,21 @@ define <4 x i32> @combine_nested_undef_test20(<4 x i32> %A, <4 x i32> %B) {
 ;
 ; 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:    pblendw {{.*#+}} xmm0 = xmm1[0,1,2,3],xmm0[4,5,6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,3,0]
 ; SSE41-NEXT:    retq
 ;
-; AVX-LABEL: combine_nested_undef_test20:
-; AVX:       # BB#0:
-; AVX-NEXT:    vblendpd {{.*#+}} xmm0 = xmm1[0],xmm0[1]
-; AVX-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[0,2,3,0]
-; AVX-NEXT:    retq
+; AVX1-LABEL: combine_nested_undef_test20:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3],xmm0[4,5,6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,3,0]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_nested_undef_test20:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3]
+; AVX2-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,3,0]
+; AVX2-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>
   ret <4 x i32> %2
@@ -1026,15 +1185,21 @@ define <4 x i32> @combine_nested_undef_test21(<4 x i32> %A, <4 x i32> %B) {
 ;
 ; 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:    pblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3],xmm1[4,5,6,7]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
 ; SSE41-NEXT:    retq
 ;
-; AVX-LABEL: combine_nested_undef_test21:
-; AVX:       # BB#0:
-; AVX-NEXT:    vblendps {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2,3]
-; AVX-NEXT:    vmovddup {{.*#+}} xmm0 = xmm0[0,0]
-; AVX-NEXT:    retq
+; AVX1-LABEL: combine_nested_undef_test21:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2,3],xmm1[4,5,6,7]
+; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,0,1]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: combine_nested_undef_test21:
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpblendd {{.*#+}} xmm0 = xmm1[0],xmm0[1],xmm1[2,3]
+; AVX2-NEXT:    vpbroadcastq %xmm0, %xmm0
+; AVX2-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>
   ret <4 x i32> %2

diff --git a/test/CodeGen/X86/vector-trunc.ll b/test/CodeGen/X86/vector-trunc.ll
index 2850b0ad2703f274fe9f76027016e03421def102..ea0ff4c208d49bc351adc004a572808b3cb855e9 100644 (file)
--- a/test/CodeGen/X86/vector-trunc.ll
+++ b/test/CodeGen/X86/vector-trunc.ll
@@ -4,14 +4,28 @@
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mcpu=x86-64 -mattr=+avx | FileCheck %s --check-prefix=AVX --check-prefix=AVX1
 
 define <4 x i32> @trunc2x2i64(<2 x i64> %a, <2 x i64> %b) {
-; SSE-LABEL: trunc2x2i64:
-; SSE:       # BB#0: # %entry
-; SSE-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
-; SSE-NEXT:    retq
+; SSE2-LABEL: trunc2x2i64:
+; SSE2:       # BB#0: # %entry
+; SSE2-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
+; SSE2-NEXT:    retq
+;
+; SSSE3-LABEL: trunc2x2i64:
+; SSSE3:       # BB#0: # %entry
+; SSSE3-NEXT:    shufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
+; SSSE3-NEXT:    retq
+;
+; SSE41-LABEL: trunc2x2i64:
+; SSE41:       # BB#0: # %entry
+; SSE41-NEXT:    pshufd {{.*#+}} xmm1 = xmm1[0,1,0,2]
+; SSE41-NEXT:    pshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
+; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: trunc2x2i64:
 ; AVX:       # BB#0: # %entry
-; AVX-NEXT:    vshufps {{.*#+}} xmm0 = xmm0[0,2],xmm1[0,2]
+; AVX-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,1,0,2]
+; AVX-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,2,2,3]
+; AVX-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
 ; AVX-NEXT:    retq
 
 
@@ -64,18 +78,16 @@ define <8 x i16> @trunc2x4i32(<4 x i32> %a, <4 x i32> %b) {
 ;
 ; SSE41-LABEL: trunc2x4i32:
 ; SSE41:       # BB#0: # %entry
-; SSE41-NEXT:    movdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; SSE41-NEXT:    pshufb %xmm2, %xmm1
-; SSE41-NEXT:    pshufb %xmm2, %xmm0
-; SSE41-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; SSE41-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,4,5,6,7,0,1,4,5,8,9,12,13]
+; SSE41-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; SSE41-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
 ; SSE41-NEXT:    retq
 ;
 ; AVX-LABEL: trunc2x4i32:
 ; AVX:       # BB#0: # %entry
-; AVX-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; AVX-NEXT:    vpshufb %xmm2, %xmm1, %xmm1
-; AVX-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
-; AVX-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; AVX-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,4,5,6,7,0,1,4,5,8,9,12,13]
+; AVX-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; AVX-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
 ; AVX-NEXT:    retq
 
 

diff --git a/test/CodeGen/X86/widen_shuffle-1.ll b/test/CodeGen/X86/widen_shuffle-1.ll
index 2aa870f16eb4a6f557a112204c6bde70ba58e51b..546ce4502f025f28f1db2f1f698e7b3c26f38404 100644 (file)
--- a/test/CodeGen/X86/widen_shuffle-1.ll
+++ b/test/CodeGen/X86/widen_shuffle-1.ll
@@ -68,10 +68,9 @@ entry:
 define <8 x i8> @shuf4(<4 x i8> %a, <4 x i8> %b) nounwind readnone {
 ; CHECK-LABEL: shuf4:
 ; CHECK:       # BB#0:
-; CHECK-NEXT:    movdqa {{.*#+}} xmm2 = [0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
-; CHECK-NEXT:    pshufb %xmm2, %xmm1
-; CHECK-NEXT:    pshufb %xmm2, %xmm0
-; CHECK-NEXT:    punpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
+; CHECK-NEXT:    pshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,4,5,6,7,0,1,4,5,8,9,12,13]
+; CHECK-NEXT:    pshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,12,13,8,9,12,13,12,13,14,15]
+; CHECK-NEXT:    pblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5,6,7]
 ; CHECK-NEXT:    retl
   %vshuf = shufflevector <4 x i8> %a, <4 x i8> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7>
   ret <8 x i8> %vshuf