[x86] Teach the AVX1 path of the new vector shuffle lowering one more

trick that I missed.

VPERMILPS has a non-immediate memory operand mode that allows it to do
asymetric shuffles in the two 128-bit lanes. Use this rather than two
shuffles and a blend.

However, it turns out the variable shuffle path to VPERMILPS (and
VPERMILPD, although that one offers no functional differenc from the
immediate operand other than variability) wasn't even plumbed through
codegen. Do such plumbing so that we can reasonably emit
a variable-masked VPERMILP instruction. Also plumb basic comment parsing
and printing through so that the tests are reasonable.

There are still a few tests which don't show the shuffle pattern. These
are tests with undef lanes. I'll teach the shuffle decoding and printing
to handle undef mask entries in a follow-up. I've looked at the masks
and they seem reasonable.

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218300 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/X86/Utils/X86ShuffleDecode.cpp b/lib/Target/X86/Utils/X86ShuffleDecode.cpp
index 6d42a101b0e5bb50addbbd18d48960288106f8d4..9aca2da49020fc2c3f84af10cd176f87ccbe4c2b 100644 (file)
--- a/lib/Target/X86/Utils/X86ShuffleDecode.cpp
+++ b/lib/Target/X86/Utils/X86ShuffleDecode.cpp
@@ -287,4 +287,27 @@ void DecodeVPERMMask(unsigned Imm, SmallVectorImpl<int> &ShuffleMask) {
   }
 }
 
+void DecodeVPERMILPMask(const ConstantDataSequential *C,
+                        SmallVectorImpl<int> &ShuffleMask) {
+  Type *MaskTy = C->getType();
+  assert(MaskTy->isVectorTy() && "Expected a vector constant mask!");
+  assert(MaskTy->getVectorElementType()->isIntegerTy() &&
+         "Expected integer constant mask elements!");
+  int ElementBits = MaskTy->getScalarSizeInBits();
+  int NumElements = MaskTy->getVectorNumElements();
+  assert((NumElements == 2 || NumElements == 4 || NumElements == 8) &&
+         "Unexpected number of vector elements.");
+  assert((unsigned)NumElements == C->getNumElements() &&
+         "Constant mask has a different number of elements!");
+
+  ShuffleMask.reserve(NumElements);
+  for (int i = 0; i < NumElements; ++i) {
+    int Base = (i * ElementBits / 128) * (128 / ElementBits);
+    uint64_t Element = C->getElementAsInteger(i);
+    // Only the least significant 2 bits of the integer are used.
+    int Index = Base + (Element & 0x3);
+    ShuffleMask.push_back(Index);
+  }
+}
+
 } // llvm namespace

diff --git a/lib/Target/X86/Utils/X86ShuffleDecode.h b/lib/Target/X86/Utils/X86ShuffleDecode.h
index 03a843e7b8df145e2c6abe1e11e01eeff59f4cdd..8034d209ac3890dfa8b6edbb59a3908e8f6b3b8a 100644 (file)
--- a/lib/Target/X86/Utils/X86ShuffleDecode.h
+++ b/lib/Target/X86/Utils/X86ShuffleDecode.h
@@ -84,6 +84,10 @@ void DecodeVPERM2X128Mask(MVT VT, unsigned Imm,
 /// No VT provided since it only works on 256-bit, 4 element vectors.
 void DecodeVPERMMask(unsigned Imm, SmallVectorImpl<int> &ShuffleMask);
 
+/// \brief Decode a VPERMILP variable mask from an IR-level vector constant.
+void DecodeVPERMILPMask(const ConstantDataSequential *C,
+                        SmallVectorImpl<int> &ShuffleMask);
+
 } // llvm namespace
 
 #endif

diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index 2e195080f8b185e49ff1247adb598d3a547fbfe6..40ab77aaaa03852d850887b560f7929e27961360 100644 (file)
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -9395,26 +9395,15 @@ static SDValue lowerV8F32VectorShuffle(SDValue Op, SDValue V1, SDValue V2,
   }
 
   // If we have a single input shuffle with different shuffle patterns in the
-  // two 128-bit lanes, just do two shuffles and blend them together. This will
-  // be faster than extracting the high 128-bit lane, shuffling it, and
-  // re-inserting it. Especially on newer processors where blending is *the*
-  // fastest operation.
+  // two 128-bit lanes use the variable mask to VPERMILPS.
   if (isSingleInputShuffleMask(Mask)) {
-    int LoMask[4] = {Mask[0], Mask[1], Mask[2], Mask[3]};
-    int HiMask[4] = {Mask[4], Mask[5], Mask[6], Mask[7]};
-    for (int &M : HiMask)
-      if (M >= 0)
-        M -= 4;
-    SDValue Lo = V1, Hi = V1;
-    if (!isNoopShuffleMask(LoMask))
-      Lo = DAG.getNode(X86ISD::VPERMILPI, DL, MVT::v8f32, Lo,
-                       getV4X86ShuffleImm8ForMask(LoMask, DAG));
-    if (!isNoopShuffleMask(HiMask))
-      Hi = DAG.getNode(X86ISD::VPERMILPI, DL, MVT::v8f32, Hi,
-                       getV4X86ShuffleImm8ForMask(HiMask, DAG));
-    unsigned BlendMask = 1 << 4 | 1 << 5 | 1 << 6 | 1 << 7;
-    return DAG.getNode(X86ISD::BLENDI, DL, MVT::v8f32, Lo, Hi,
-                       DAG.getConstant(BlendMask, MVT::i8));
+    SDValue VPermMask[8];
+    for (int i = 0; i < 8; ++i)
+      VPermMask[i] = Mask[i] < 0 ? DAG.getUNDEF(MVT::i32)
+                                 : DAG.getConstant(Mask[i], MVT::i32);
+    return DAG.getNode(
+        X86ISD::VPERMILPV, DL, MVT::v8f32, V1,
+        DAG.getNode(ISD::BUILD_VECTOR, DL, MVT::v8i32, VPermMask));
   }
 
   // Shuffle the input elements into the desired positions in V1 and V2 and

diff --git a/lib/Target/X86/X86ISelLowering.h b/lib/Target/X86/X86ISelLowering.h
index a624fa25dab6b654c6c9796fba1c4bb95bf10503..a16cf4a0b64ebcdcfd48923cbf286e0f6caf8b73 100644 (file)
--- a/lib/Target/X86/X86ISelLowering.h
+++ b/lib/Target/X86/X86ISelLowering.h
@@ -343,6 +343,7 @@ namespace llvm {
       MOVSS,
       UNPCKL,
       UNPCKH,
+      VPERMILPV,
       VPERMILPI,
       VPERMV,
       VPERMV3,

diff --git a/lib/Target/X86/X86InstrFragmentsSIMD.td b/lib/Target/X86/X86InstrFragmentsSIMD.td
index 455991e46814e00f3a7e685dc512890fa769e6fe..2badbb7d76b800a87d76e835e35909c3eb9e5b31 100644 (file)
--- a/lib/Target/X86/X86InstrFragmentsSIMD.td
+++ b/lib/Target/X86/X86InstrFragmentsSIMD.td
@@ -188,6 +188,8 @@ def SDTShuff2Op : SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisSameAs<0,1>,
 def SDTShuff3Op : SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0,1>,
                                 SDTCisSameAs<0,2>, SDTCisSameAs<0,3>]>;
 
+def SDTShuff2OpM : SDTypeProfile<1, 2, [SDTCisVec<0>, SDTCisSameAs<0,1>,
+                                        SDTCisVec<2>]>;
 def SDTShuff2OpI : SDTypeProfile<1, 2, [SDTCisVec<0>,
                                  SDTCisSameAs<0,1>, SDTCisInt<2>]>;
 def SDTShuff3OpI : SDTypeProfile<1, 3, [SDTCisVec<0>, SDTCisSameAs<0,1>,
@@ -232,6 +234,7 @@ def X86Packus : SDNode<"X86ISD::PACKUS", SDTPack>;
 def X86Unpckl : SDNode<"X86ISD::UNPCKL", SDTShuff2Op>;
 def X86Unpckh : SDNode<"X86ISD::UNPCKH", SDTShuff2Op>;
 
+def X86VPermilpv  : SDNode<"X86ISD::VPERMILPV", SDTShuff2OpM>;
 def X86VPermilpi  : SDNode<"X86ISD::VPERMILPI", SDTShuff2OpI>;
 def X86VPermv     : SDNode<"X86ISD::VPERMV",    SDTShuff2Op>;
 def X86VPermi     : SDNode<"X86ISD::VPERMI",    SDTShuff2OpI>;

diff --git a/lib/Target/X86/X86InstrSSE.td b/lib/Target/X86/X86InstrSSE.td
index 7a7ca8548a15bbe2cfc1e1ce76fdfe80b449d43d..a186899d23104ff7b64cafdcf5e6e183737bdadc 100644 (file)
--- a/lib/Target/X86/X86InstrSSE.td
+++ b/lib/Target/X86/X86InstrSSE.td
@@ -8418,6 +8418,15 @@ let ExeDomain = SSEPackedDouble in {
 }
 
 let Predicates = [HasAVX] in {
+def : Pat<(v8f32 (X86VPermilpv VR256:$src1, (v8i32 VR256:$src2))),
+          (VPERMILPSYrr VR256:$src1, VR256:$src2)>;
+def : Pat<(v8f32 (X86VPermilpv VR256:$src1, (bc_v8i32 (loadv4i64 addr:$src2)))),
+          (VPERMILPSYrm VR256:$src1, addr:$src2)>;
+def : Pat<(v4f64 (X86VPermilpv VR256:$src1, (v4i64 VR256:$src2))),
+          (VPERMILPDYrr VR256:$src1, VR256:$src2)>;
+def : Pat<(v4f64 (X86VPermilpv VR256:$src1, (loadv4i64 addr:$src2))),
+          (VPERMILPDYrm VR256:$src1, addr:$src2)>;
+
 def : Pat<(v8i32 (X86VPermilpi VR256:$src1, (i8 imm:$imm))),
           (VPERMILPSYri VR256:$src1, imm:$imm)>;
 def : Pat<(v4i64 (X86VPermilpi VR256:$src1, (i8 imm:$imm))),
@@ -8428,6 +8437,15 @@ def : Pat<(v8i32 (X86VPermilpi (bc_v8i32 (loadv4i64 addr:$src1)),
 def : Pat<(v4i64 (X86VPermilpi (loadv4i64 addr:$src1), (i8 imm:$imm))),
           (VPERMILPDYmi addr:$src1, imm:$imm)>;
 
+def : Pat<(v4f32 (X86VPermilpv VR128:$src1, (v4i32 VR128:$src2))),
+          (VPERMILPSrr VR128:$src1, VR128:$src2)>;
+def : Pat<(v4f32 (X86VPermilpv VR128:$src1, (bc_v4i32 (loadv2i64 addr:$src2)))),
+          (VPERMILPSrm VR128:$src1, addr:$src2)>;
+def : Pat<(v2f64 (X86VPermilpv VR128:$src1, (v2i64 VR128:$src2))),
+          (VPERMILPDrr VR128:$src1, VR128:$src2)>;
+def : Pat<(v2f64 (X86VPermilpv VR128:$src1, (loadv2i64 addr:$src2))),
+          (VPERMILPDrm VR128:$src1, addr:$src2)>;
+
 def : Pat<(v2i64 (X86VPermilpi VR128:$src1, (i8 imm:$imm))),
           (VPERMILPDri VR128:$src1, imm:$imm)>;
 def : Pat<(v2i64 (X86VPermilpi (loadv2i64 addr:$src1), (i8 imm:$imm))),

diff --git a/lib/Target/X86/X86MCInstLower.cpp b/lib/Target/X86/X86MCInstLower.cpp
index ded84fc28f10df5edf2a1f21fd4efdbb4f7d5d0a..5665a01260645f61a18b5442c3a5e0a4a8ca50c0 100644 (file)
--- a/lib/Target/X86/X86MCInstLower.cpp
+++ b/lib/Target/X86/X86MCInstLower.cpp
@@ -1022,15 +1022,19 @@ void X86AsmPrinter::EmitInstruction(const MachineInstr *MI) {
 
   case X86::PSHUFBrm:
   case X86::VPSHUFBrm:
-    // Lower PSHUFB normally but add a comment if we can find a constant
-    // shuffle mask. We won't be able to do this at the MC layer because the
-    // mask isn't an immediate.
+  case X86::VPERMILPSrm:
+  case X86::VPERMILPDrm:
+  case X86::VPERMILPSYrm:
+  case X86::VPERMILPDYrm:
+    // Lower PSHUFB and VPERMILP normally but add a comment if we can find
+    // a constant shuffle mask. We won't be able to do this at the MC layer
+    // because the mask isn't an immediate.
     std::string Comment;
     raw_string_ostream CS(Comment);
     SmallVector<int, 16> Mask;
 
-    assert(MI->getNumOperands() >= 6 &&
-           "Wrong number of operands for PSHUFBrm or VPSHUFBrm");
+    // All of these instructions accept a constant pool operand as their fifth.
+    assert(MI->getNumOperands() > 5 && "We should always have at least 5 operands!");
     const MachineOperand &DstOp = MI->getOperand(0);
     const MachineOperand &SrcOp = MI->getOperand(1);
     const MachineOperand &MaskOp = MI->getOperand(5);
@@ -1061,7 +1065,18 @@ void X86AsmPrinter::EmitInstruction(const MachineInstr *MI) {
           assert(MaskTy == C->getType() &&
                  "Expected a constant of the same type!");
 
-          DecodePSHUFBMask(C, Mask);
+          switch (MI->getOpcode()) {
+          case X86::PSHUFBrm:
+          case X86::VPSHUFBrm:
+            DecodePSHUFBMask(C, Mask);
+            break;
+          case X86::VPERMILPSrm:
+          case X86::VPERMILPDrm:
+          case X86::VPERMILPSYrm:
+          case X86::VPERMILPDYrm:
+            DecodeVPERMILPMask(C, Mask);
+          }
+
           assert(Mask.size() == MaskTy->getVectorNumElements() &&
                  "Shuffle mask has a different size than its type!");
         }

diff --git a/test/CodeGen/X86/vector-shuffle-256-v8.ll b/test/CodeGen/X86/vector-shuffle-256-v8.ll
index e2f731b2af585409ae19854e3d304aa9a270d4b7..df40df2a3258200c56015033ffd847c537e97071 100644 (file)
--- a/test/CodeGen/X86/vector-shuffle-256-v8.ll
+++ b/test/CodeGen/X86/vector-shuffle-256-v8.ll
@@ -381,9 +381,7 @@ define <8 x float> @shuffle_v8f32_10225466(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_00015444(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_00015444
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,0,0,0,5,4,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,0,1,4,4,4,5]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,0,1,5,4,4,4]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 1, i32 5, i32 4, i32 4, i32 4>
   ret <8 x float> %shuffle
@@ -392,9 +390,7 @@ define <8 x float> @shuffle_v8f32_00015444(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_00204644(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_00204644
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[0,2,0,0,4,6,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,0,4,4,6,4]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,0,4,6,4,4]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 2, i32 0, i32 4, i32 6, i32 4, i32 4>
   ret <8 x float> %shuffle
@@ -403,9 +399,7 @@ define <8 x float> @shuffle_v8f32_00204644(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_03004474(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_03004474
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[0,0,3,0,4,4,7,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,3,0,0,4,7,4,4]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,3,0,0,4,4,7,4]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 3, i32 0, i32 0, i32 4, i32 4, i32 7, i32 4>
   ret <8 x float> %shuffle
@@ -414,9 +408,7 @@ define <8 x float> @shuffle_v8f32_03004474(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_10004444(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_10004444
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[0,0,0,0,4,4,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[1,0,0,0,5,4,4,4]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[1,0,0,0,4,4,4,4]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 1, i32 0, i32 0, i32 0, i32 4, i32 4, i32 4, i32 4>
   ret <8 x float> %shuffle
@@ -425,9 +417,7 @@ define <8 x float> @shuffle_v8f32_10004444(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_22006446(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_22006446
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[2,0,0,2,6,4,4,6]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[2,2,0,0,6,6,4,4]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[2,2,0,0,6,4,4,6]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 2, i32 2, i32 0, i32 0, i32 6, i32 4, i32 4, i32 6>
   ret <8 x float> %shuffle
@@ -436,9 +426,7 @@ define <8 x float> @shuffle_v8f32_22006446(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_33307474(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_33307474
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[3,0,3,0,7,4,7,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[3,3,3,0,7,7,7,4]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[3,3,3,0,7,4,7,4]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 3, i32 3, i32 3, i32 0, i32 7, i32 4, i32 7, i32 4>
   ret <8 x float> %shuffle
@@ -447,8 +435,7 @@ define <8 x float> @shuffle_v8f32_33307474(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_32104567(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_32104567
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[3,2,1,0,7,6,5,4]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[3,2,1,0,4,5,6,7]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 4, i32 5, i32 6, i32 7>
   ret <8 x float> %shuffle
@@ -457,9 +444,7 @@ define <8 x float> @shuffle_v8f32_32104567(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_00236744(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_00236744
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[2,3,0,0,6,7,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,3,4,4,6,7]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,3,6,7,4,4]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 2, i32 3, i32 6, i32 7, i32 4, i32 4>
   ret <8 x float> %shuffle
@@ -468,9 +453,7 @@ define <8 x float> @shuffle_v8f32_00236744(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_00226644(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_00226644
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[2,2,0,0,6,6,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,2,4,4,6,6]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,2,6,6,4,4]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 2, i32 2, i32 6, i32 6, i32 4, i32 4>
   ret <8 x float> %shuffle
@@ -479,8 +462,7 @@ define <8 x float> @shuffle_v8f32_00226644(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_10324567(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_10324567
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,0,3,2,5,4,7,6]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[1,0,3,2,4,5,6,7]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 4, i32 5, i32 6, i32 7>
   ret <8 x float> %shuffle
@@ -489,8 +471,7 @@ define <8 x float> @shuffle_v8f32_10324567(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_11334567(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_11334567
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,1,3,3,5,5,7,7]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[1,1,3,3,4,5,6,7]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 1, i32 1, i32 3, i32 3, i32 4, i32 5, i32 6, i32 7>
   ret <8 x float> %shuffle
@@ -499,8 +480,7 @@ define <8 x float> @shuffle_v8f32_11334567(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_01235467(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_01235467
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,0,2,3,5,4,6,7]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,1,2,3,5,4,6,7]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 4, i32 6, i32 7>
   ret <8 x float> %shuffle
@@ -509,8 +489,7 @@ define <8 x float> @shuffle_v8f32_01235467(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_01235466(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_01235466
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,0,2,2,5,4,6,6]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,1,2,3,5,4,6,6]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 4, i32 6, i32 6>
   ret <8 x float> %shuffle
@@ -519,9 +498,7 @@ define <8 x float> @shuffle_v8f32_01235466(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_002u6u44(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_002u6u44
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[2,1,0,0,6,5,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,3,4,4,6,7]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}}, %ymm0, %ymm0
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 2, i32 undef, i32 6, i32 undef, i32 4, i32 4>
   ret <8 x float> %shuffle
@@ -530,9 +507,7 @@ define <8 x float> @shuffle_v8f32_002u6u44(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_00uu66uu(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_00uu66uu
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[2,2,2,3,6,6,6,7]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,3,4,4,6,7]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}}, %ymm0, %ymm0
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 undef, i32 undef, i32 6, i32 6, i32 undef, i32 undef>
   ret <8 x float> %shuffle
@@ -541,8 +516,7 @@ define <8 x float> @shuffle_v8f32_00uu66uu(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_103245uu(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_103245uu
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,0,3,2,5,4,7,6]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}}, %ymm0, %ymm0
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 4, i32 5, i32 undef, i32 undef>
   ret <8 x float> %shuffle
@@ -551,8 +525,7 @@ define <8 x float> @shuffle_v8f32_103245uu(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_1133uu67(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_1133uu67
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,1,3,3,5,5,7,7]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}}, %ymm0, %ymm0
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 1, i32 1, i32 3, i32 3, i32 undef, i32 undef, i32 6, i32 7>
   ret <8 x float> %shuffle
@@ -561,8 +534,7 @@ define <8 x float> @shuffle_v8f32_1133uu67(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_0uu354uu(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_0uu354uu
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,0,2,3,5,4,6,7]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}}, %ymm0, %ymm0
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 undef, i32 undef, i32 3, i32 5, i32 4, i32 undef, i32 undef>
   ret <8 x float> %shuffle
@@ -571,8 +543,7 @@ define <8 x float> @shuffle_v8f32_0uu354uu(<8 x float> %a, <8 x float> %b) {
 define <8 x float> @shuffle_v8f32_uuu3uu66(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: @shuffle_v8f32_uuu3uu66
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[0,1,2,2,4,5,6,6]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}}, %ymm0, %ymm0
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 undef, i32 3, i32 undef, i32 undef, i32 6, i32 6>
   ret <8 x float> %shuffle
@@ -956,9 +927,7 @@ define <8 x i32> @shuffle_v8i32_10225466(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_00015444(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_00015444
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,0,0,0,5,4,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,0,1,4,4,4,5]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,0,1,5,4,4,4]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 1, i32 5, i32 4, i32 4, i32 4>
   ret <8 x i32> %shuffle
@@ -967,9 +936,7 @@ define <8 x i32> @shuffle_v8i32_00015444(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_00204644(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_00204644
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[0,2,0,0,4,6,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,0,4,4,6,4]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,0,4,6,4,4]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 0, i32 2, i32 0, i32 4, i32 6, i32 4, i32 4>
   ret <8 x i32> %shuffle
@@ -978,9 +945,7 @@ define <8 x i32> @shuffle_v8i32_00204644(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_03004474(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_03004474
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[0,0,3,0,4,4,7,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,3,0,0,4,7,4,4]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,3,0,0,4,4,7,4]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 3, i32 0, i32 0, i32 4, i32 4, i32 7, i32 4>
   ret <8 x i32> %shuffle
@@ -989,9 +954,7 @@ define <8 x i32> @shuffle_v8i32_03004474(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_10004444(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_10004444
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[0,0,0,0,4,4,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[1,0,0,0,5,4,4,4]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[1,0,0,0,4,4,4,4]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 1, i32 0, i32 0, i32 0, i32 4, i32 4, i32 4, i32 4>
   ret <8 x i32> %shuffle
@@ -1000,9 +963,7 @@ define <8 x i32> @shuffle_v8i32_10004444(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_22006446(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_22006446
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[2,0,0,2,6,4,4,6]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[2,2,0,0,6,6,4,4]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[2,2,0,0,6,4,4,6]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 2, i32 2, i32 0, i32 0, i32 6, i32 4, i32 4, i32 6>
   ret <8 x i32> %shuffle
@@ -1011,9 +972,7 @@ define <8 x i32> @shuffle_v8i32_22006446(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_33307474(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_33307474
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[3,0,3,0,7,4,7,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[3,3,3,0,7,7,7,4]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[3,3,3,0,7,4,7,4]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 3, i32 3, i32 3, i32 0, i32 7, i32 4, i32 7, i32 4>
   ret <8 x i32> %shuffle
@@ -1022,8 +981,7 @@ define <8 x i32> @shuffle_v8i32_33307474(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_32104567(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_32104567
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[3,2,1,0,7,6,5,4]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[3,2,1,0,4,5,6,7]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 4, i32 5, i32 6, i32 7>
   ret <8 x i32> %shuffle
@@ -1032,9 +990,7 @@ define <8 x i32> @shuffle_v8i32_32104567(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_00236744(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_00236744
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[2,3,0,0,6,7,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,3,4,4,6,7]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,3,6,7,4,4]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 0, i32 2, i32 3, i32 6, i32 7, i32 4, i32 4>
   ret <8 x i32> %shuffle
@@ -1043,9 +999,7 @@ define <8 x i32> @shuffle_v8i32_00236744(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_00226644(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_00226644
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[2,2,0,0,6,6,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,2,4,4,6,6]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,2,6,6,4,4]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 0, i32 2, i32 2, i32 6, i32 6, i32 4, i32 4>
   ret <8 x i32> %shuffle
@@ -1054,8 +1008,7 @@ define <8 x i32> @shuffle_v8i32_00226644(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_10324567(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_10324567
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,0,3,2,5,4,7,6]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[1,0,3,2,4,5,6,7]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 4, i32 5, i32 6, i32 7>
   ret <8 x i32> %shuffle
@@ -1064,8 +1017,7 @@ define <8 x i32> @shuffle_v8i32_10324567(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_11334567(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_11334567
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,1,3,3,5,5,7,7]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[1,1,3,3,4,5,6,7]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 1, i32 1, i32 3, i32 3, i32 4, i32 5, i32 6, i32 7>
   ret <8 x i32> %shuffle
@@ -1074,8 +1026,7 @@ define <8 x i32> @shuffle_v8i32_11334567(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_01235467(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_01235467
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,0,2,3,5,4,6,7]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,1,2,3,5,4,6,7]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 4, i32 6, i32 7>
   ret <8 x i32> %shuffle
@@ -1084,8 +1035,7 @@ define <8 x i32> @shuffle_v8i32_01235467(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_01235466(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_01235466
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,0,2,2,5,4,6,6]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,1,2,3,5,4,6,6]
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 5, i32 4, i32 6, i32 6>
   ret <8 x i32> %shuffle
@@ -1094,9 +1044,7 @@ define <8 x i32> @shuffle_v8i32_01235466(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_002u6u44(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_002u6u44
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[2,1,0,0,6,5,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,3,4,4,6,7]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}}, %ymm0, %ymm0
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 0, i32 2, i32 undef, i32 6, i32 undef, i32 4, i32 4>
   ret <8 x i32> %shuffle
@@ -1105,9 +1053,7 @@ define <8 x i32> @shuffle_v8i32_002u6u44(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_00uu66uu(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_00uu66uu
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[2,2,2,3,6,6,6,7]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,3,4,4,6,7]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}}, %ymm0, %ymm0
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 0, i32 undef, i32 undef, i32 6, i32 6, i32 undef, i32 undef>
   ret <8 x i32> %shuffle
@@ -1116,8 +1062,7 @@ define <8 x i32> @shuffle_v8i32_00uu66uu(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_103245uu(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_103245uu
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,0,3,2,5,4,7,6]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}}, %ymm0, %ymm0
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 1, i32 0, i32 3, i32 2, i32 4, i32 5, i32 undef, i32 undef>
   ret <8 x i32> %shuffle
@@ -1126,8 +1071,7 @@ define <8 x i32> @shuffle_v8i32_103245uu(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_1133uu67(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_1133uu67
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,1,3,3,5,5,7,7]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm1[0,1,2,3],ymm0[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}}, %ymm0, %ymm0
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 1, i32 1, i32 3, i32 3, i32 undef, i32 undef, i32 6, i32 7>
   ret <8 x i32> %shuffle
@@ -1136,8 +1080,7 @@ define <8 x i32> @shuffle_v8i32_1133uu67(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_0uu354uu(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_0uu354uu
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[1,0,2,3,5,4,6,7]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}}, %ymm0, %ymm0
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 0, i32 undef, i32 undef, i32 3, i32 5, i32 4, i32 undef, i32 undef>
   ret <8 x i32> %shuffle
@@ -1146,8 +1089,7 @@ define <8 x i32> @shuffle_v8i32_0uu354uu(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_uuu3uu66(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: @shuffle_v8i32_uuu3uu66
 ; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm0[0,1,2,2,4,5,6,6]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2,3],ymm1[4,5,6,7]
+; ALL-NEXT:    vpermilps {{.*}}, %ymm0, %ymm0
 ; ALL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 undef, i32 undef, i32 undef, i32 3, i32 undef, i32 undef, i32 6, i32 6>
   ret <8 x i32> %shuffle