[X86] Fix two more places to prefer VPERMQ/PD over VPERM2X128 when AVX2 is enabled

[craig.topper]

Summary

The shuffle combining and lowerVectorShuffleAsLanePermuteAndBlend were both still trying to use VPERM2XF128 for unary shuffles when AVX2 is enabled. VPERM2X128 takes two inputs meaning when we use it for a unary shuffle one of those inputs is left undefined creating a false dependency on whatever register gets allocated there.

If we have VPERMQ/PD we should prefer those since they only have a single input.

[RKSimon]

lib/Target/X86/X86ISelLowering.cpp

12150 ↗

(On Diff #115542)

What happens if we just replace this with a generic DAG.getVectorShuffle? I'm thinking about when we finally get around to supporting 512-bit in lowerVectorShuffleAsLanePermuteAndBlend.

[craig.topper]

Use getVectorShuffle

[craig.topper]

Not sure what happened there. That was clearly the wrong patch

[RKSimon]

LGTM

llvm/trunk/lib/Target/X86/X86ISelLowering.cpp

static SDValue lowerVectorShuffleAsLanePermuteAndBlend(const SDLoc &DL, MVT VT,
   SmallVector<int, 32> FlippedBlendMask(Size);
   for (int i = 0; i < Size; ++i)
     FlippedBlendMask[i] =
         Mask[i] < 0 ? -1 : (((Mask[i] % Size) / LaneSize == i / LaneSize)
                                 ? Mask[i]
                                 : Mask[i] % LaneSize +
                                       (i / LaneSize) * LaneSize + Size);
 
-  // Flip the vector, and blend the results which should now be in-lane. The
-  // VPERM2X128 mask uses the low 2 bits for the low source and bits 4 and
-  // 5 for the high source. The value 3 selects the high half of source 2 and
-  // the value 2 selects the low half of source 2. We only use source 2 to
-  // allow folding it into a memory operand.
-  unsigned PERMMask = 3 | 2 << 4;
+  // Flip the vector, and blend the results which should now be in-lane.
   MVT PVT = VT.isFloatingPoint() ? MVT::v4f64 : MVT::v4i64;
-  SDValue Flipped = DAG.getNode(X86ISD::VPERM2X128, DL, PVT, DAG.getUNDEF(VT),
-                                DAG.getBitcast(PVT, V1),
-                                DAG.getConstant(PERMMask, DL, MVT::i8));
+  SDValue Flipped = DAG.getBitcast(PVT, V1);
+  Flipped = DAG.getVectorShuffle(PVT, DL, Flipped, DAG.getUNDEF(PVT),
+                                 { 2, 3, 0, 1 });
   Flipped = DAG.getBitcast(VT, Flipped);
   return DAG.getVectorShuffle(VT, DL, V1, Flipped, FlippedBlendMask);
 }
 
 /// \brief Handle lowering 2-lane 128-bit shuffles.
 static SDValue lowerV2X128VectorShuffle(const SDLoc &DL, MVT VT, SDValue V1,
                                         SDValue V2, ArrayRef<int> Mask,
                                         const APInt &Zeroable,
static SDValue combineX86ShuffleChain(ArrayRef<SDValue> Inputs, SDValue Root,
   bool IsEVEXShuffle =
       RootSizeInBits == 512 || (Subtarget.hasVLX() && RootSizeInBits >= 128);
   if (IsEVEXShuffle && (RootVT.getScalarSizeInBits() != BaseMaskEltSizeInBits))
     return SDValue();
 
   // TODO - handle 128/256-bit lane shuffles of 512-bit vectors.
 
   // Handle 128-bit lane shuffles of 256-bit vectors.
+  // If we have AVX2, prefer to use VPERMQ/VPERMPD for unary shuffles unless
+  // we need to use the zeroing feature.
   // TODO - this should support binary shuffles.
   if (UnaryShuffle && RootVT.is256BitVector() && NumBaseMaskElts == 2 &&
+      !(Subtarget.hasAVX2() && BaseMask[0] >= -1 && BaseMask[1] >= -1) &&
       !isSequentialOrUndefOrZeroInRange(BaseMask, 0, 2, 0)) {
     if (Depth == 1 && Root.getOpcode() == X86ISD::VPERM2X128)
       return SDValue(); // Nothing to do!
     MVT ShuffleVT = (FloatDomain ? MVT::v4f64 : MVT::v4i64);
     unsigned PermMask = 0;
     PermMask |= ((BaseMask[0] < 0 ? 0x8 : (BaseMask[0] & 1)) << 0);
     PermMask |= ((BaseMask[1] < 0 ? 0x8 : (BaseMask[1] & 1)) << 4);
 

llvm/trunk/test/CodeGen/X86/avx-vperm2x128.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx -disable-peephole | FileCheck %s --check-prefix=ALL --check-prefix=AVX1
 ; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx2 -disable-peephole | FileCheck %s --check-prefix=ALL --check-prefix=AVX2
 
 define <8 x float> @shuffle_v8f32_45670123(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
-; ALL-LABEL: shuffle_v8f32_45670123:
-; ALL:       # BB#0: # %entry
-; ALL-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
-; ALL-NEXT:    retq
+; AVX1-LABEL: shuffle_v8f32_45670123:
+; AVX1:       # BB#0: # %entry
+; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v8f32_45670123:
+; AVX2:       # BB#0: # %entry
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX2-NEXT:    retq
 entry:
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_45670123_mem(<8 x float>* %pa, <8 x float>* %pb) nounwind uwtable readnone ssp {
-; ALL-LABEL: shuffle_v8f32_45670123_mem:
-; ALL:       # BB#0: # %entry
-; ALL-NEXT:    vperm2f128 {{.*#+}} ymm0 = mem[2,3,0,1]
-; ALL-NEXT:    retq
+; AVX1-LABEL: shuffle_v8f32_45670123_mem:
+; AVX1:       # BB#0: # %entry
+; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = mem[2,3,0,1]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v8f32_45670123_mem:
+; AVX2:       # BB#0: # %entry
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = mem[2,3,0,1]
+; AVX2-NEXT:    retq
 entry:
   %a = load <8 x float>, <8 x float>* %pa
   %b = load <8 x float>, <8 x float>* %pb
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_0123cdef(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {

 define <8 x float> @shuffle_v8f32_01230123(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
 ; AVX1-LABEL: shuffle_v8f32_01230123:
 ; AVX1:       # BB#0: # %entry
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v8f32_01230123:
 ; AVX2:       # BB#0: # %entry
-; AVX2-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,1,0,1]
 ; AVX2-NEXT:    retq
 entry:
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_01230123_mem(<8 x float>* %pa, <8 x float>* %pb) nounwind uwtable readnone ssp {
-; ALL-LABEL: shuffle_v8f32_01230123_mem:
-; ALL:       # BB#0: # %entry
-; ALL-NEXT:    vperm2f128 {{.*#+}} ymm0 = mem[0,1,0,1]
-; ALL-NEXT:    retq
+; AVX1-LABEL: shuffle_v8f32_01230123_mem:
+; AVX1:       # BB#0: # %entry
+; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = mem[0,1,0,1]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v8f32_01230123_mem:
+; AVX2:       # BB#0: # %entry
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = mem[0,1,0,1]
+; AVX2-NEXT:    retq
 entry:
   %a = load <8 x float>, <8 x float>* %pa
   %b = load <8 x float>, <8 x float>* %pb
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 1, i32 2, i32 3, i32 0, i32 1, i32 2, i32 3>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_45674567(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
-; ALL-LABEL: shuffle_v8f32_45674567:
-; ALL:       # BB#0: # %entry
-; ALL-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
-; ALL-NEXT:    retq
+; AVX1-LABEL: shuffle_v8f32_45674567:
+; AVX1:       # BB#0: # %entry
+; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v8f32_45674567:
+; AVX2:       # BB#0: # %entry
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX2-NEXT:    retq
 entry:
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 4, i32 5, i32 6, i32 7>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_45674567_mem(<8 x float>* %pa, <8 x float>* %pb) nounwind uwtable readnone ssp {
-; ALL-LABEL: shuffle_v8f32_45674567_mem:
-; ALL:       # BB#0: # %entry
-; ALL-NEXT:    vperm2f128 {{.*#+}} ymm0 = mem[2,3,2,3]
-; ALL-NEXT:    retq
+; AVX1-LABEL: shuffle_v8f32_45674567_mem:
+; AVX1:       # BB#0: # %entry
+; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = mem[2,3,2,3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v8f32_45674567_mem:
+; AVX2:       # BB#0: # %entry
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = mem[2,3,2,3]
+; AVX2-NEXT:    retq
 entry:
   %a = load <8 x float>, <8 x float>* %pa
   %b = load <8 x float>, <8 x float>* %pb
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 4, i32 5, i32 6, i32 7>
   ret <8 x float> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_2323(<32 x i8> %a, <32 x i8> %b) nounwind uwtable readnone ssp {
-; ALL-LABEL: shuffle_v32i8_2323:
-; ALL:       # BB#0: # %entry
-; ALL-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
-; ALL-NEXT:    retq
+; AVX1-LABEL: shuffle_v32i8_2323:
+; AVX1:       # BB#0: # %entry
+; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v32i8_2323:
+; AVX2:       # BB#0: # %entry
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX2-NEXT:    retq
 entry:
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_2323_domain(<32 x i8> %a, <32 x i8> %b) nounwind uwtable readnone ssp {
 ; AVX1-LABEL: shuffle_v32i8_2323_domain:
 ; AVX1:       # BB#0: # %entry
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm0
 ; AVX1-NEXT:    vpcmpeqd %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpsubb %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v32i8_2323_domain:
 ; AVX2:       # BB#0: # %entry
 ; AVX2-NEXT:    vpcmpeqd %ymm1, %ymm1, %ymm1
 ; AVX2-NEXT:    vpsubb %ymm1, %ymm0, %ymm0
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,2,3]
 ; AVX2-NEXT:    retq
 entry:
   ; add forces execution domain
   %a2 = add <32 x i8> %a, <i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1, i8 1>
   %shuffle = shufflevector <32 x i8> %a2, <32 x i8> %b, <32 x i32> <i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31, i32 16, i32 17, i32 18, i32 19, i32 20, i32 21, i32 22, i32 23, i32 24, i32 25, i32 26, i32 27, i32 28, i32 29, i32 30, i32 31>
   ret <32 x i8> %shuffle
 }
 

 ; ALL-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3],ymm1[0,1]
 ; ALL-NEXT:    retq
 entry:
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 undef, i32 undef, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_4567uu67(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
-; ALL-LABEL: shuffle_v8f32_4567uu67:
-; ALL:       # BB#0: # %entry
-; ALL-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
-; ALL-NEXT:    retq
+; AVX1-LABEL: shuffle_v8f32_4567uu67:
+; AVX1:       # BB#0: # %entry
+; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: shuffle_v8f32_4567uu67:
+; AVX2:       # BB#0: # %entry
+; AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX2-NEXT:    retq
 entry:
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 undef, i32 undef, i32 6, i32 7>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_4567uuef(<8 x float> %a, <8 x float> %b) nounwind uwtable readnone ssp {
 ; ALL-LABEL: shuffle_v8f32_4567uuef:
 ; ALL:       # BB#0: # %entry

llvm/trunk/test/CodeGen/X86/avx2-intrinsics-x86-upgrade.ll

 ; CHECK-NEXT:    retl
   %res = call <16 x i16> @llvm.x86.avx2.pabs.w(<16 x i16> %a0) ; <<16 x i16>> [#uses=1]
   ret <16 x i16> %res
 }
 declare <16 x i16> @llvm.x86.avx2.pabs.w(<16 x i16>) nounwind readnone
 
 
 define <4 x i64> @test_x86_avx2_vperm2i128(<4 x i64> %a0, <4 x i64> %a1) {
-; AVX2-LABEL: test_x86_avx2_vperm2i128:
-; AVX2:       ## BB#0:
-; AVX2-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
-; AVX2-NEXT:    retl
-;
-; AVX512-LABEL: test_x86_avx2_vperm2i128:
-; AVX512:       ## BB#0:
-; AVX512-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
-; AVX512-NEXT:    retl
+; CHECK-LABEL: test_x86_avx2_vperm2i128:
+; CHECK:       ## BB#0:
+; CHECK-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; CHECK-NEXT:    retl
   %res = call <4 x i64> @llvm.x86.avx2.vperm2i128(<4 x i64> %a0, <4 x i64> %a1, i8 1) ; <<4 x i64>> [#uses=1]
   ret <4 x i64> %res
 }
 declare <4 x i64> @llvm.x86.avx2.vperm2i128(<4 x i64>, <4 x i64>, i8) nounwind readonly

llvm/trunk/test/CodeGen/X86/oddshuffles.ll

 ;
 ; AVX2-LABEL: interleave_24i16_in:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vmovdqu (%rsi), %xmm0
 ; AVX2-NEXT:    vmovdqu (%rdx), %xmm1
 ; AVX2-NEXT:    vmovdqu (%rcx), %xmm2
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm3
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm4 = ymm3[0,1,2,3,6,7,2,3,8,9,8,9,4,5,6,7,16,17,18,19,22,23,18,19,24,25,24,25,20,21,22,23]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm3 = ymm3[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[2,3,0,1]
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm3 = ymm3[u,u,0,1,u,u,u,u,2,3,u,u,u,u,4,5,u,u,22,23,u,u,u,u,24,25,u,u,u,u,26,27]
 ; AVX2-NEXT:    vpblendw {{.*#+}} ymm3 = ymm4[0],ymm3[1],ymm4[2,3],ymm3[4],ymm4[5,6],ymm3[7],ymm4[8],ymm3[9],ymm4[10,11],ymm3[12],ymm4[13,14],ymm3[15]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = <u,0,0,u,1,1,u,2>
 ; AVX2-NEXT:    vpermd %ymm2, %ymm4, %ymm4
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm5 = [255,255,255,255,0,0,255,255,255,255,0,0,255,255,255,255,0,0,255,255,255,255,0,0,255,255,255,255,0,0,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm5, %ymm3, %ymm4, %ymm3
 ; AVX2-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
 ; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[4,5,10,11,10,11,8,9,8,9,14,15,12,13,14,15]

 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: interleave_24i32_in:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vmovdqu (%rsi), %ymm0
 ; AVX2-NEXT:    vmovdqu (%rdx), %ymm1
 ; AVX2-NEXT:    vmovdqu (%rcx), %ymm2
 ; AVX2-NEXT:    vpshufd {{.*#+}} xmm3 = xmm1[1,0,2,2]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm3 = ymm3[0,1,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[0,1,0,1]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm4 = ymm0[0,0,2,1]
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm3 = ymm4[0],ymm3[1],ymm4[2,3],ymm3[4],ymm4[5,6],ymm3[7]
 ; AVX2-NEXT:    vpbroadcastq %xmm2, %ymm4
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm3 = ymm3[0,1],ymm4[2],ymm3[3,4],ymm4[5],ymm3[6,7]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm4 = ymm2[2,1,3,3]
 ; AVX2-NEXT:    vpshufd {{.*#+}} ymm5 = ymm1[1,2,3,3,5,6,7,7]
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm5 = ymm5[2,2,2,3]
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm4 = ymm5[0],ymm4[1],ymm5[2,3],ymm4[4],ymm5[5,6],ymm4[7]

llvm/trunk/test/CodeGen/X86/vector-shuffle-256-v16.ll

 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_00_00_00_00_00_00_08_00_00_00_00_00_00_00_00:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpbroadcastw %xmm0, %xmm1
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpshuflw {{.*#+}} ymm0 = ymm0[0,0,0,0,4,5,6,7,8,8,8,8,12,13,14,15]
 ; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,0,1,1,4,4,5,5]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_00_00_00_00_00_00_08_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,0,1,0,1,0,1,0,1,0,1,6,7,0,1]
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_00_00_00_00_00_09_00_00_00_00_00_00_00_00_00:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,0,255,255,u,u,u,u,u,u,u,u,u,u,u,u,255,255,u,u,u,u,u,u,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,0,1,0,1,0,1,0,1,0,1,2,3,0,1,16,17,16,17,16,17,16,17,16,17,16,17,16,17,16,17]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_00_00_00_00_00_09_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [0,0,0,0,0,0,9,0,0,0,0,0,0,0,0,0]

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,0,1,0,1,0,1,0,1,10,11,0,1,0,1]
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_00_00_00_00_10_00_00_00_00_00_00_00_00_00_00:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3,4,5,6,7]
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,0,1,0,1,0,1,0,1,4,5,0,1,0,1,16,17,16,17,16,17,16,17,16,17,16,17,16,17,16,17]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_00_00_00_00_10_00_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [0,0,0,0,0,10,0,0,0,0,0,0,0,0,0,0]
 ; AVX512VL-NEXT:    vpermw %ymm0, %ymm1, %ymm0

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,0,1,0,1,0,1,14,15,0,1,0,1,0,1]
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,0,0,0,4,5,6,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,0,1,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_00_00_00_11_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3,4,5,6,7]
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,0,1,0,1,0,1,6,7,0,1,0,1,0,1,16,17,16,17,16,17,16,17,16,17,16,17,16,17,16,17]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_00_00_00_11_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [0,0,0,0,11,0,0,0,0,0,0,0,0,0,0,0]
 ; AVX512VL-NEXT:    vpermw %ymm0, %ymm1, %ymm0

 
 define <16 x i16> @shuffle_v16i16_00_00_00_00_00_00_01_00_00_00_00_00_00_00_01_00(<16 x i16> %a, <16 x i16> %b) {
 ; AVX1-LABEL: shuffle_v16i16_00_00_00_00_00_00_01_00_00_00_00_00_00_00_01_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,2,3,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-LABEL: shuffle_v16i16_00_00_00_00_00_00_01_00_00_00_00_00_00_00_01_00:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,2,3,0,1]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX2-NEXT:    retq
-;
-; AVX512VL-LABEL: shuffle_v16i16_00_00_00_00_00_00_01_00_00_00_00_00_00_00_01_00:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,2,3,0,1]
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v16i16_00_00_00_00_00_00_01_00_00_00_00_00_00_00_01_00:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,2,3,0,1]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 0>
   ret <16 x i16> %shuffle
 }
 
 define <16 x i16> @shuffle_v16i16_00_00_00_00_00_02_00_00_00_00_00_00_00_02_00_00(<16 x i16> %a, <16 x i16> %b) {
 ; AVX1-LABEL: shuffle_v16i16_00_00_00_00_00_02_00_00_00_00_00_00_00_02_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,4,5,0,1,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-LABEL: shuffle_v16i16_00_00_00_00_00_02_00_00_00_00_00_00_00_02_00_00:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,4,5,0,1,0,1]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX2-NEXT:    retq
-;
-; AVX512VL-LABEL: shuffle_v16i16_00_00_00_00_00_02_00_00_00_00_00_00_00_02_00_00:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,4,5,0,1,0,1]
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v16i16_00_00_00_00_00_02_00_00_00_00_00_00_00_02_00_00:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,4,5,0,1,0,1]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 2, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 2, i32 0, i32 0>
   ret <16 x i16> %shuffle
 }
 
 define <16 x i16> @shuffle_v16i16_00_00_00_00_03_00_00_00_00_00_00_00_03_00_00_00(<16 x i16> %a, <16 x i16> %b) {
 ; AVX1-LABEL: shuffle_v16i16_00_00_00_00_03_00_00_00_00_00_00_00_03_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,6,7,0,1,0,1,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-LABEL: shuffle_v16i16_00_00_00_00_03_00_00_00_00_00_00_00_03_00_00_00:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,6,7,0,1,0,1,0,1]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX2-NEXT:    retq
-;
-; AVX512VL-LABEL: shuffle_v16i16_00_00_00_00_03_00_00_00_00_00_00_00_03_00_00_00:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,6,7,0,1,0,1,0,1]
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v16i16_00_00_00_00_03_00_00_00_00_00_00_00_03_00_00_00:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,6,7,0,1,0,1,0,1]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 0, i32 0, i32 0, i32 3, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 3, i32 0, i32 0, i32 0>
   ret <16 x i16> %shuffle
 }
 
 define <16 x i16> @shuffle_v16i16_00_00_00_04_00_00_00_00_00_00_00_04_00_00_00_00(<16 x i16> %a, <16 x i16> %b) {
 ; AVX1-LABEL: shuffle_v16i16_00_00_00_04_00_00_00_00_00_00_00_04_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,8,9,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-LABEL: shuffle_v16i16_00_00_00_04_00_00_00_00_00_00_00_04_00_00_00_00:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,8,9,0,1,0,1,0,1,0,1]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX2-NEXT:    retq
-;
-; AVX512VL-LABEL: shuffle_v16i16_00_00_00_04_00_00_00_00_00_00_00_04_00_00_00_00:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,8,9,0,1,0,1,0,1,0,1]
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v16i16_00_00_00_04_00_00_00_00_00_00_00_04_00_00_00_00:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,8,9,0,1,0,1,0,1,0,1]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 0, i32 0, i32 4, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 4, i32 0, i32 0, i32 0, i32 0>
   ret <16 x i16> %shuffle
 }
 
 define <16 x i16> @shuffle_v16i16_00_00_05_00_00_00_00_00_00_00_05_00_00_00_00_00(<16 x i16> %a, <16 x i16> %b) {
 ; AVX1-LABEL: shuffle_v16i16_00_00_05_00_00_00_00_00_00_00_05_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,10,11,0,1,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-LABEL: shuffle_v16i16_00_00_05_00_00_00_00_00_00_00_05_00_00_00_00_00:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,10,11,0,1,0,1,0,1,0,1,0,1]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX2-NEXT:    retq
-;
-; AVX512VL-LABEL: shuffle_v16i16_00_00_05_00_00_00_00_00_00_00_05_00_00_00_00_00:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,10,11,0,1,0,1,0,1,0,1,0,1]
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v16i16_00_00_05_00_00_00_00_00_00_00_05_00_00_00_00_00:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,10,11,0,1,0,1,0,1,0,1,0,1]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 0, i32 5, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 5, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <16 x i16> %shuffle
 }
 
 define <16 x i16> @shuffle_v16i16_00_06_00_00_00_00_00_00_00_06_00_00_00_00_00_00(<16 x i16> %a, <16 x i16> %b) {
 ; AVX1-LABEL: shuffle_v16i16_00_06_00_00_00_00_00_00_00_06_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,12,13,0,1,0,1,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-LABEL: shuffle_v16i16_00_06_00_00_00_00_00_00_00_06_00_00_00_00_00_00:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,12,13,0,1,0,1,0,1,0,1,0,1,0,1]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX2-NEXT:    retq
-;
-; AVX512VL-LABEL: shuffle_v16i16_00_06_00_00_00_00_00_00_00_06_00_00_00_00_00_00:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,12,13,0,1,0,1,0,1,0,1,0,1,0,1]
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v16i16_00_06_00_00_00_00_00_00_00_06_00_00_00_00_00_00:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,12,13,0,1,0,1,0,1,0,1,0,1,0,1]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 6, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 6, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <16 x i16> %shuffle
 }
 
 define <16 x i16> @shuffle_v16i16_07_00_00_00_00_00_00_00_07_00_00_00_00_00_00_00(<16 x i16> %a, <16 x i16> %b) {
 ; AVX1-LABEL: shuffle_v16i16_07_00_00_00_00_00_00_00_07_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[14,15,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-LABEL: shuffle_v16i16_07_00_00_00_00_00_00_00_07_00_00_00_00_00_00_00:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[14,15,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX2-NEXT:    retq
-;
-; AVX512VL-LABEL: shuffle_v16i16_07_00_00_00_00_00_00_00_07_00_00_00_00_00_00_00:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[14,15,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v16i16_07_00_00_00_00_00_00_00_07_00_00_00_00_00_00_00:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[14,15,0,1,0,1,0,1,0,1,0,1,0,1,0,1]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 7, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 7, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <16 x i16> %shuffle
 }
 
 define <16 x i16> @shuffle_v16i16_00_17_02_19_04_21_06_23_08_25_10_27_12_29_14_31(<16 x i16> %a, <16 x i16> %b) {
 ; AVX1-LABEL: shuffle_v16i16_00_17_02_19_04_21_06_23_08_25_10_27_12_29_14_31:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vmovaps {{.*#+}} ymm2 = [65535,0,65535,0,65535,0,65535,0,65535,0,65535,0,65535,0,65535,0]

 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,0,2,4,5,6,7]
 ; AVX1-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,6,4,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_01_00_01_02_03_02_11_08_09_08_09_10_11_10_11:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpbroadcastq %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,0,1,1,4,4,5,5]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_01_00_01_02_03_02_11_08_09_08_09_10_11_10_11:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0],xmm1[1],xmm0[2,3,4,5,6,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,2,1,0]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[3,2,1,0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_06_07_04_05_02_03_00_09_14_15_12_13_10_11_08_09:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_06_07_04_05_02_03_00_09_14_15_12_13_10_11_08_09:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [6,7,4,5,2,3,0,9,14,15,12,13,10,11,8,9]

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,4,5,6,7,0,1,4,5,8,9,14,15]
 ; AVX1-NEXT:    vpunpckhqdq {{.*#+}} xmm0 = xmm0[1],xmm1[1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_04_05_06_07_16_17_18_27_12_13_14_15_24_25_26_27:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5],ymm0[6,7]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[2,3,0,1,6,7,4,5]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_04_05_06_07_16_17_18_27_12_13_14_15_24_25_26_27:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm2 = [4,5,6,7,16,17,18,27,12,13,14,15,24,25,26,27]

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,0,1,0,1,0,1,0,1,0,1,2,3]
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,0,1,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08_08:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpbroadcastw %xmm1, %xmm1
 ; AVX2-NEXT:    vpshuflw {{.*#+}} ymm0 = ymm0[0,0,0,0,4,5,6,7,8,8,8,8,12,13,14,15]
 ; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,0,1,1,4,4,5,5]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_00_00_00_00_00_00_08_08_08_08_08_08_08_08_08:

 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,0,0,0,4,5,6,7]
 ; AVX1-NEXT:    vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,4,4,4,4]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_00_00_00_04_04_04_12_08_08_08_08_12_12_12_12:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpsllq $48, %xmm1, %xmm1
 ; AVX2-NEXT:    vpshuflw {{.*#+}} ymm0 = ymm0[0,0,0,0,4,5,6,7,8,8,8,8,12,13,14,15]
 ; AVX2-NEXT:    vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,4,4,4,4,8,9,10,11,12,12,12,12]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_00_00_00_04_04_04_12_08_08_08_08_12_12_12_12:

 ; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,0,2,2,4,5,6,7]
 ; AVX1-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,6,7]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_uu_00_uu_01_uu_02_uu_11_uu_08_uu_09_uu_10_uu_11:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <255,255,255,255,255,255,0,0,u,u,u,u,u,u,u,u,255,255,255,255,255,255,255,255,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpunpcklwd {{.*#+}} ymm0 = ymm0[0,0,1,1,2,2,3,3,8,8,9,9,10,10,11,11]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_uu_00_uu_01_uu_02_uu_11_uu_08_uu_09_uu_10_uu_11:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = <u,0,u,1,u,2,u,11,u,8,u,9,u,10,u,11>

 ; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,0,2,2,4,5,6,7]
 ; AVX1-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,4,6,7]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_uu_04_uu_05_uu_06_uu_15_uu_12_uu_13_uu_14_uu_15:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
-; AVX2-NEXT:    vpunpckhwd {{.*#+}} ymm0 = ymm0[4,4,5,5,6,6,7,7,12,12,13,13,14,14,15,15]
-; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3],ymm0[4,5,6,7]
+; AVX2-NEXT:    vpunpckhwd {{.*#+}} ymm1 = ymm0[4,4,5,5,6,6,7,7,12,12,13,13,14,14,15,15]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3],ymm1[4,5,6,7]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_uu_04_uu_05_uu_06_uu_15_uu_12_uu_13_uu_14_uu_15:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = <u,4,u,5,u,6,u,15,u,12,u,13,u,14,u,15>
 ; AVX512VL-NEXT:    vpermw %ymm0, %ymm1, %ymm0
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 undef, i32 4, i32 undef, i32 5, i32 undef, i32 6, i32 undef, i32 15, i32 undef, i32 12, i32 undef, i32 13, i32 undef, i32 14, i32 undef, i32 15>

 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,3,2]
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[3,1,2,0,4,5,6,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,1,3,2]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_03_01_02_00_06_07_04_13_11_09_10_08_14_15_12_13:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpshuflw {{.*#+}} ymm0 = ymm0[3,1,2,0,4,5,6,7,11,9,10,8,12,13,14,15]
 ; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,1,3,2,4,5,7,6]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_03_01_02_00_06_07_04_13_11_09_10_08_14_15_12_13:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,8,9,8,9,0,1,0,1,0,1,14,15]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[8,9,8,9,8,9,8,9,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_04_04_04_04_00_00_00_08_12_12_12_12_08_08_08_08:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpbroadcastw %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[8,9,8,9,8,9,8,9,0,1,0,1,0,1,0,1,24,25,24,25,24,25,24,25,16,17,16,17,16,17,16,17]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_04_04_04_04_00_00_00_08_12_12_12_12_08_08_08_08:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5],xmm0[6,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[1,0,3,2]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,0,3,2]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_02_03_00_01_06_07_04_13_10_11_08_09_14_15_12_13:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[1,0,3,2,5,4,7,6]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_02_03_00_01_06_07_04_13_10_11_08_09_14_15_12_13:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [2,3,0,1,6,7,4,13,10,11,8,9,14,15,12,13]

 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,1,3,2]
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[2,3,0,2,4,5,6,7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,1,3,2]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_02_03_00_02_06_07_04_13_10_11_08_10_14_15_12_13:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <255,255,u,u,255,255,255,255,255,255,0,0,255,255,255,255,255,255,u,u,255,255,255,255,255,255,255,255,255,255,255,255>
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpshuflw {{.*#+}} ymm0 = ymm0[2,3,0,2,4,5,6,7,10,11,8,10,12,13,14,15]
 ; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,1,3,2,4,5,7,6]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_02_03_00_02_06_07_04_13_10_11_08_10_14_15_12_13:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[1,0,2,3]
 ; AVX1-NEXT:    vpshufhw {{.*#+}} xmm1 = xmm1[0,1,2,3,6,7,4,7]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_02_03_00_01_06_07_04_15_10_11_08_09_14_15_12_15:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[1,0,2,3,5,4,6,7]
 ; AVX2-NEXT:    vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,6,7,4,7,8,9,10,11,14,15,12,15]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_02_03_00_01_06_07_04_15_10_11_08_09_14_15_12_15:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm3
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0],xmm0[1,2,3,4,5,6,7]
 ; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_07_05_06_04_03_01_02_08_15_13_14_12_11_09_10_08:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[14,15,10,11,12,13,8,9,6,7,2,3,4,5,0,1,30,31,26,27,28,29,24,25,22,23,18,19,20,21,16,17]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_07_05_06_04_03_01_02_08_15_13_14_12_11_09_10_08:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [7,5,6,4,3,1,2,8,15,13,14,12,11,9,10,8]

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[2,3,0,1,10,11,8,9,10,11,8,9,2,3,2,3]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[2,3,0,1,10,11,8,9,10,11,8,9,2,3,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_01_00_05_04_05_04_01_08_09_08_13_12_13_12_09_08:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpbroadcastw %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[2,3,0,1,10,11,8,9,10,11,8,9,2,3,0,1,18,19,16,17,26,27,24,25,26,27,24,25,18,19,16,17]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_01_00_05_04_05_04_01_08_09_08_13_12_13_12_09_08:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[10,11,8,9,2,3,0,1,10,11,8,9,2,3,2,3]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[10,11,8,9,2,3,0,1,10,11,8,9,2,3,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_05_04_01_00_05_04_01_08_13_12_09_08_13_12_09_08:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpbroadcastw %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[10,11,8,9,2,3,0,1,10,11,8,9,2,3,0,1,26,27,24,25,18,19,16,17,26,27,24,25,18,19,16,17]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_05_04_01_00_05_04_01_08_13_12_09_08_13_12_09_08:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[10,11,8,9,2,3,0,1,2,3,0,1,10,11,2,3]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[10,11,8,9,2,3,0,1,2,3,0,1,10,11,8,9]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_05_04_01_00_01_00_05_12_13_12_09_08_09_08_13_12:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpsllq $48, %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[10,11,8,9,2,3,0,1,2,3,0,1,10,11,8,9,26,27,24,25,18,19,16,17,18,19,16,17,26,27,24,25]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_05_04_01_00_01_00_05_12_13_12_09_08_09_08_13_12:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,8,9,8,9,2,3]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,0,1,8,9,8,9,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_04_04_00_00_04_04_08_08_12_12_08_08_12_12_08:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpbroadcastw %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,8,9,8,9,0,1,0,1,8,9,8,9,0,1,16,17,24,25,24,25,16,17,16,17,24,25,24,25,16,17]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_04_04_00_00_04_04_08_08_12_12_08_08_12_12_08:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[8,9,0,1,0,1,8,9,8,9,0,1,0,1,2,3]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[8,9,0,1,0,1,8,9,8,9,0,1,0,1,8,9]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_04_00_00_04_04_00_00_12_12_08_08_12_12_08_08_12:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpsllq $48, %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[8,9,0,1,0,1,8,9,8,9,0,1,0,1,8,9,24,25,16,17,16,17,24,25,24,25,16,17,16,17,24,25]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_04_00_00_04_04_00_00_12_12_08_08_12_12_08_08_12:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm3
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7]
 ; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_02_06_04_00_05_01_07_11_10_14_12_08_13_09_15_11:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[4,5,12,13,8,9,0,1,10,11,2,3,14,15,6,7,20,21,28,29,24,25,16,17,26,27,18,19,30,31,22,23]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_02_06_04_00_05_01_07_11_10_14_12_08_13_09_15_11:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [2,6,4,0,5,1,7,11,10,14,12,8,13,9,15,11]

 ; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm3
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7]
 ; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_02_00_06_04_05_01_07_11_10_08_14_12_13_09_15_11:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[4,5,0,1,12,13,8,9,10,11,2,3,14,15,6,7,20,21,16,17,28,29,24,25,26,27,18,19,30,31,22,23]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_02_00_06_04_05_01_07_11_10_08_14_12_13_09_15_11:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [2,0,6,4,5,1,7,11,10,8,14,12,13,9,15,11]

 ; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm3
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5],xmm0[6,7]
 ; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_02_06_04_00_01_03_07_13_10_14_12_08_09_11_15_13:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[4,5,12,13,8,9,0,1,2,3,6,7,14,15,10,11,20,21,28,29,24,25,16,17,18,19,22,23,30,31,26,27]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_02_06_04_00_01_03_07_13_10_14_12_08_09_11_15_13:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [2,6,4,0,1,3,7,13,10,14,12,8,9,11,15,13]

 ; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm3
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2,3],xmm0[4,5,6,7]
 ; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_06_06_07_05_01_06_04_11_14_14_15_13_09_14_12_11:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2,3,4,5,6,7]
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[12,13,12,13,14,15,10,11,2,3,12,13,8,9,6,7,28,29,28,29,30,31,26,27,18,19,28,29,24,25,22,23]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_06_06_07_05_01_06_04_11_14_14_15_13_09_14_12_11:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [6,6,7,5,1,6,4,11,14,14,15,13,9,14,12,11]
 ; AVX512VL-NEXT:    vpermw %ymm0, %ymm1, %ymm0

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,0,1,8,9,8,9,8,9,8,9,8,9,14,15]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,0,1,8,9,8,9,8,9,8,9,8,9,8,9]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_00_04_04_04_04_04_12_08_08_12_12_12_12_12_12:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpsllq $48, %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,0,1,8,9,8,9,8,9,8,9,8,9,8,9,16,17,16,17,24,25,24,25,24,25,24,25,24,25,24,25]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_00_04_04_04_04_04_12_08_08_12_12_12_12_12_12:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,0,1,0,1,8,9,8,9,8,9,14,15]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[8,9,8,9,0,1,0,1,8,9,8,9,8,9,8,9]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_04_04_00_00_04_04_04_12_12_12_08_08_12_12_12_12:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpsllq $48, %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[8,9,8,9,0,1,0,1,8,9,8,9,8,9,8,9,24,25,24,25,16,17,16,17,24,25,24,25,24,25,24,25]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_04_04_00_00_04_04_04_12_12_12_08_08_12_12_12_12:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,14,15]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_04_04_00_04_04_04_12_08_12_12_08_12_12_12_12:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpsllq $48, %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9,16,17,24,25,24,25,16,17,24,25,24,25,24,25,24,25]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_04_04_00_04_04_04_12_08_12_12_08_12_12_12_12:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,0,1,0,1,0,1,14,15]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,0,1,0,1,0,1,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_04_04_00_00_00_00_08_08_12_12_08_08_08_08_08:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpbroadcastw %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,8,9,8,9,0,1,0,1,0,1,0,1,0,1,16,17,24,25,24,25,16,17,16,17,16,17,16,17,16,17]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_04_04_00_00_00_00_08_08_12_12_08_08_08_08_08:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm1 = xmm1[0,2,2,3]
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm1 = xmm1[0,2,2,0,4,5,6,7]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_04_04_00_04_05_06_15_08_12_12_08_12_13_14_15:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[0,2,2,3,4,6,6,7]
 ; AVX2-NEXT:    vpshuflw {{.*#+}} ymm0 = ymm0[0,2,2,0,4,5,6,7,8,10,10,8,12,13,14,15]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_04_04_00_04_05_06_15_08_12_12_08_12_13_14_15:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,2,3,8,9,8,9,8,9,8,9,8,9,14,15]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,2,3,8,9,8,9,8,9,8,9,8,9,8,9]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_uu_04_04_04_04_04_12_08_uu_12_12_12_12_12_12:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpsllq $48, %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,2,3,8,9,8,9,8,9,8,9,8,9,8,9,16,17,18,19,24,25,24,25,24,25,24,25,24,25,24,25]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <255,255,u,u,255,255,255,255,255,255,255,255,255,255,0,0,255,255,u,u,255,255,255,255,255,255,255,255,255,255,255,255>
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_uu_04_04_04_04_04_12_08_uu_12_12_12_12_12_12:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[8,9,8,9,8,9,0,1,8,9,8,9,8,9,14,15]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[8,9,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_04_04_uu_00_04_04_04_12_12_12_uu_08_12_12_12_12:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpsllq $48, %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[8,9,8,9,8,9,0,1,8,9,8,9,8,9,8,9,24,25,24,25,24,25,16,17,24,25,24,25,24,25,24,25]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <255,255,255,255,u,u,255,255,255,255,255,255,255,255,0,0,255,255,255,255,u,u,255,255,255,255,255,255,255,255,255,255>
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_04_04_uu_00_04_04_04_12_12_12_uu_08_12_12_12_12:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,14,15]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_uu_04_04_00_04_04_04_12_uu_12_12_08_12_12_12_12:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpsllq $48, %xmm1, %xmm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,8,9,8,9,0,1,8,9,8,9,8,9,8,9,16,17,24,25,24,25,16,17,24,25,24,25,24,25,24,25]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <u,u,255,255,255,255,255,255,255,255,255,255,255,255,0,0,u,u,255,255,255,255,255,255,255,255,255,255,255,255,255,255>
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_uu_04_04_00_04_04_04_12_uu_12_12_08_12_12_12_12:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm3
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2],xmm1[3],xmm0[4,5,6,7]
 ; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_01_02_07_04_05_06_11_08_09_10_15_12_13_14_11:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,14,15,8,9,10,11,12,13,6,7,16,17,18,19,20,21,30,31,24,25,26,27,28,29,22,23]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_01_02_07_04_05_06_11_08_09_10_15_12_13_14_11:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [0,1,2,7,4,5,6,11,8,9,10,15,12,13,14,11]

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[8,9,10,11,12,13,6,7,0,1,2,3,4,5,2,3]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4,5,6],xmm1[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[8,9,10,11,12,13,6,7,0,1,2,3,4,5,14,15]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_04_05_06_03_00_01_02_15_12_13_14_11_08_09_10_15:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
-; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[8,9,10,11,12,13,6,7,0,1,2,3,4,5,14,15,24,25,26,27,28,29,22,23,16,17,18,19,20,21,30,31]
+; AVX2-NEXT:    vpshufb {{.*#+}} ymm1 = ymm0[8,9,10,11,12,13,6,7,0,1,2,3,4,5,14,15,24,25,26,27,28,29,22,23,16,17,18,19,20,21,30,31]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vpblendvb %ymm2, %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_04_05_06_03_00_01_02_15_12_13_14_11_08_09_10_15:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [4,5,6,3,0,1,2,15,12,13,14,11,8,9,10,15]
 ; AVX512VL-NEXT:    vpermw %ymm0, %ymm1, %ymm0
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 4, i32 5, i32 6, i32 3, i32 0, i32 1, i32 2, i32 15, i32 12, i32 13, i32 14, i32 11, i32 8, i32 9, i32 10, i32 15>
   ret <16 x i16> %shuffle
 }
 
 define <16 x i16> @shuffle_v16i16_03_07_01_00_02_07_03_13_11_15_09_08_10_15_11_13(<16 x i16> %a, <16 x i16> %b) {
 ; AVX1-LABEL: shuffle_v16i16_03_07_01_00_02_07_03_13_11_15_09_08_10_15_11_13:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [6,7,14,15,2,3,0,1,4,5,14,15,6,7,10,11]
 ; AVX1-NEXT:    vpshufb %xmm2, %xmm1, %xmm3
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4,5],xmm0[6,7]
 ; AVX1-NEXT:    vpshufb %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_03_07_01_00_02_07_03_13_11_15_09_08_10_15_11_13:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2],ymm0[3,4,5,6,7]
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[6,7,14,15,2,3,0,1,4,5,14,15,6,7,10,11,22,23,30,31,18,19,16,17,20,21,30,31,22,23,26,27]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_03_07_01_00_02_07_03_13_11_15_09_08_10_15_11_13:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [3,7,1,0,2,7,3,13,11,15,9,8,10,15,11,13]
 ; AVX512VL-NEXT:    vpermw %ymm0, %ymm1, %ymm0

 ; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm1 = xmm1[0],xmm2[0],xmm1[1],xmm2[1],xmm1[2],xmm2[2],xmm1[3],xmm2[3]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,8,9,14,15,14,15,8,9,12,13,14,15]
 ; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_16_01_17_02_18_03_27_08_24_09_25_10_26_11_27:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm1[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = <255,255,255,255,255,255,0,0,u,u,u,u,u,u,u,u,255,255,255,255,255,255,255,255,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    vpblendvb %ymm3, %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vpunpcklwd {{.*#+}} ymm0 = ymm0[0],ymm1[0],ymm0[1],ymm1[1],ymm0[2],ymm1[2],ymm0[3],ymm1[3],ymm0[8],ymm1[8],ymm0[9],ymm1[9],ymm0[10],ymm1[10],ymm0[11],ymm1[11]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_16_01_17_02_18_03_27_08_24_09_25_10_26_11_27:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,16,1,17,2,18,3,27,8,24,9,25,10,26,11,27]

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,4,5,8,9,14,15,14,15,8,9,12,13,14,15]
 ; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_20_01_21_02_22_03_31_08_28_09_29_10_30_11_31:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
-; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,8,9,2,3,10,11,4,5,12,13,6,7,14,15,16,17,24,25,18,19,26,27,20,21,28,29,22,23,30,31]
+; AVX2-NEXT:    vpshufb {{.*#+}} ymm1 = ymm0[0,1,8,9,2,3,10,11,4,5,12,13,6,7,14,15,16,17,24,25,18,19,26,27,20,21,28,29,22,23,30,31]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
-; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    vpblendvb %ymm2, %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_20_01_21_02_22_03_31_08_28_09_29_10_30_11_31:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,20,1,21,2,22,3,31,8,28,9,29,10,30,11,31]
 ; AVX512VL-NEXT:    vpermt2w %ymm1, %ymm2, %ymm0
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <16 x i16> %a, <16 x i16> %b, <16 x i32> <i32 0, i32 20, i32 1, i32 21, i32 2, i32 22, i32 3, i32 31, i32 8, i32 28, i32 9, i32 29, i32 10, i32 30, i32 11, i32 31>

 ; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm1 = xmm1[4],xmm2[4],xmm1[5],xmm2[5],xmm1[6],xmm2[6],xmm1[7],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[4,5,0,1,4,5,4,5,0,1,4,5,8,9,14,15]
 ; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_04_20_05_21_06_22_07_31_12_28_13_29_14_30_15_31:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
-; AVX2-NEXT:    vpunpckhwd {{.*#+}} ymm1 = ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15]
-; AVX2-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm2[3],ymm1[4,5,6,7]
 ; AVX2-NEXT:    vpunpckhwd {{.*#+}} ymm0 = ymm0[4,4,5,5,6,6,7,7,12,12,13,13,14,14,15,15]
+; AVX2-NEXT:    vpunpckhwd {{.*#+}} ymm2 = ymm0[4],ymm1[4],ymm0[5],ymm1[5],ymm0[6],ymm1[6],ymm0[7],ymm1[7],ymm0[12],ymm1[12],ymm0[13],ymm1[13],ymm0[14],ymm1[14],ymm0[15],ymm1[15]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[2,3,0,1]
+; AVX2-NEXT:    vpblendd {{.*#+}} ymm1 = ymm2[0,1,2],ymm1[3],ymm2[4,5,6,7]
 ; AVX2-NEXT:    vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7],ymm0[8],ymm1[9],ymm0[10],ymm1[11],ymm0[12],ymm1[13],ymm0[14],ymm1[15]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_04_20_05_21_06_22_07_31_12_28_13_29_14_30_15_31:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm2 = [4,20,5,21,6,22,7,31,12,28,13,29,14,30,15,31]
 ; AVX512VL-NEXT:    vpermt2w %ymm1, %ymm2, %ymm0
 ; AVX512VL-NEXT:    retq

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[4,5,0,1,4,5,4,5,0,1,4,5,8,9,14,15]
 ; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm1[4],xmm0[5],xmm1[5],xmm0[6],xmm1[6],xmm0[7],xmm1[7]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_04_16_05_17_06_18_07_27_12_24_13_25_14_26_15_27:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm1[0,1],ymm0[2,3],ymm1[4,5],ymm0[6,7]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[8,9,0,1,10,11,2,3,12,13,4,5,14,15,6,7,24,25,16,17,26,27,18,19,28,29,20,21,30,31,22,23]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_04_16_05_17_06_18_07_27_12_24_13_25_14_26_15_27:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm2 = [4,16,5,17,6,18,7,27,12,24,13,25,14,26,15,27]

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[2,3,6,7,10,11,14,15,14,15,10,11,12,13,14,15]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,3,2,3]
 ; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_16_01_17_06_22_07_31_08_24_09_25_14_30_15_31:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm1[2,3,0,1]
 ; AVX2-NEXT:    vpshuflw {{.*#+}} ymm1 = ymm1[0,0,2,1,4,5,6,7,8,8,10,9,12,13,14,15]
 ; AVX2-NEXT:    vpshufhw {{.*#+}} ymm1 = ymm1[0,1,2,3,4,6,6,7,8,9,10,11,12,14,14,15]
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm1 = ymm1[0,1,2],ymm2[3],ymm1[4,5,6,7]
 ; AVX2-NEXT:    vpshuflw {{.*#+}} ymm0 = ymm0[0,1,1,3,4,5,6,7,8,9,9,11,12,13,14,15]
 ; AVX2-NEXT:    vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,6,5,7,7,8,9,10,11,14,13,15,15]
 ; AVX2-NEXT:    vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7],ymm0[8],ymm1[9],ymm0[10],ymm1[11],ymm0[12],ymm1[13],ymm0[14],ymm1[15]
 ; AVX2-NEXT:    retq
 ;

 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[8,9,10,11,0,1,2,3,2,3,0,1,12,13,2,3]
 ; AVX1-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[0,3,2,3]
 ; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm1[0],xmm0[1],xmm1[1],xmm0[2],xmm1[2],xmm0[3],xmm1[3]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_20_01_21_06_16_07_25_08_28_09_29_14_24_15_25:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm1[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = <255,255,0,0,u,u,u,u,255,255,255,255,u,u,u,u,255,255,255,255,u,u,u,u,255,255,255,255,u,u,u,u>
 ; AVX2-NEXT:    vpblendvb %ymm3, %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[8,9,8,9,4,5,10,11,0,1,0,1,12,13,2,3,24,25,24,25,20,21,26,27,16,17,16,17,28,29,18,19]
 ; AVX2-NEXT:    vpshuflw {{.*#+}} ymm0 = ymm0[0,1,1,3,4,5,6,7,8,9,9,11,12,13,14,15]
 ; AVX2-NEXT:    vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,6,5,7,7,8,9,10,11,14,13,15,15]
 ; AVX2-NEXT:    vpblendw {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7],ymm0[8],ymm1[9],ymm0[10],ymm1[11],ymm0[12],ymm1[13],ymm0[14],ymm1[15]
 ; AVX2-NEXT:    retq
 ;

 ; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[2,0,3,1,4,5,6,7]
 ; AVX1-NEXT:    vpshufhw {{.*#+}} xmm0 = xmm0[0,1,2,3,6,4,7,5]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_01_00_17_16_03_02_19_26_09_08_25_24_11_10_27_26:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm1[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = <255,255,255,255,0,0,255,255,u,u,u,u,u,u,u,u,255,255,255,255,255,255,255,255,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    vpblendvb %ymm3, %ymm1, %ymm2, %ymm1
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[0,1,2,3,2,3,0,1,8,9,10,11,6,7,4,5,16,17,18,19,18,19,16,17,24,25,26,27,22,23,20,21]
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[2,3,0,1,4,5,6,7,6,7,4,5,4,5,6,7,18,19,16,17,20,21,22,23,22,23,20,21,20,21,22,23]
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_01_00_17_16_03_02_19_26_09_08_25_24_11_10_27_26:

 ; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,1,4,5,8,9,14,15,14,15,8,9,12,13,14,15]
 ; AVX1-NEXT:    vpunpcklwd {{.*#+}} xmm0 = xmm1[0],xmm0[0],xmm1[1],xmm0[1],xmm1[2],xmm0[2],xmm1[3],xmm0[3]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_16_00_17_01_18_02_19_11_24_08_25_09_26_10_27_11:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm2 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = <255,255,255,255,255,255,0,0,u,u,u,u,u,u,u,u,255,255,255,255,255,255,255,255,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    vpblendvb %ymm3, %ymm0, %ymm2, %ymm0
 ; AVX2-NEXT:    vpunpcklwd {{.*#+}} ymm0 = ymm1[0],ymm0[0],ymm1[1],ymm0[1],ymm1[2],ymm0[2],ymm1[3],ymm0[3],ymm1[8],ymm0[8],ymm1[9],ymm0[9],ymm1[10],ymm0[10],ymm1[11],ymm0[11]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_16_00_17_01_18_02_19_11_24_08_25_09_26_10_27_11:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,16,1,17,2,18,3,27,8,24,9,25,10,26,11,27]

 ; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[4,5,0,1,4,5,4,5,0,1,4,5,8,9,14,15]
 ; AVX1-NEXT:    vpunpckhwd {{.*#+}} xmm0 = xmm1[4],xmm0[4],xmm1[5],xmm0[5],xmm1[6],xmm0[6],xmm1[7],xmm0[7]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_20_04_21_05_22_06_23_15_28_12_29_13_30_14_31_15:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
-; AVX2-NEXT:    vpunpckhwd {{.*#+}} ymm0 = ymm0[4,4,5,5,6,6,7,7,12,12,13,13,14,14,15,15]
-; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2],ymm2[3],ymm0[4,5,6,7]
 ; AVX2-NEXT:    vpunpckhwd {{.*#+}} ymm1 = ymm1[4],ymm0[4],ymm1[5],ymm0[5],ymm1[6],ymm0[6],ymm1[7],ymm0[7],ymm1[12],ymm0[12],ymm1[13],ymm0[13],ymm1[14],ymm0[14],ymm1[15],ymm0[15]
+; AVX2-NEXT:    vpunpckhwd {{.*#+}} ymm2 = ymm0[4,4,5,5,6,6,7,7,12,12,13,13,14,14,15,15]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm2[0,1,2],ymm0[3],ymm2[4,5,6,7]
 ; AVX2-NEXT:    vpblendw {{.*#+}} ymm0 = ymm1[0],ymm0[1],ymm1[2],ymm0[3],ymm1[4],ymm0[5],ymm1[6],ymm0[7],ymm1[8],ymm0[9],ymm1[10],ymm0[11],ymm1[12],ymm0[13],ymm1[14],ymm0[15]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_20_04_21_05_22_06_23_15_28_12_29_13_30_14_31_15:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm2 = [4,20,5,21,6,22,7,31,12,28,13,29,14,30,15,31]
 ; AVX512VL-NEXT:    vpermi2w %ymm0, %ymm1, %ymm2
 ; AVX512VL-NEXT:    vmovdqa %ymm2, %ymm0

 ; AVX1-NEXT:    vpshuflw {{.*#+}} xmm0 = xmm0[0,2,1,3,4,5,6,7]
 ; AVX1-NEXT:    vpunpcklqdq {{.*#+}} xmm0 = xmm0[0],xmm1[0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_02_01_03_20_22_21_31_08_10_09_11_28_30_29_31:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpshuflw {{.*#+}} ymm0 = ymm0[0,2,1,3,4,5,6,7,8,10,9,11,12,13,14,15]
 ; AVX2-NEXT:    vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,4,6,5,7,8,9,10,11,12,14,13,15]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_02_01_03_20_22_21_31_08_10_09_11_28_30_29_31:
 ; AVX512VL:       # BB#0:

 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm2[0,1,2,3],xmm1[4,5,6,7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[0,1,2,3,4,5,10,11,8,9,10,11,12,13,6,7]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_00_01_02_21_20_21_22_11_08_09_10_29_28_29_30_11:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3],ymm0[4,5],ymm1[6,7]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <255,255,255,255,255,255,0,0,255,255,255,255,255,255,u,u,255,255,255,255,255,255,255,255,255,255,255,255,255,255,u,u>
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5,10,11,8,9,10,11,12,13,6,7,16,17,18,19,20,21,26,27,24,25,26,27,28,29,22,23]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_00_01_02_21_20_21_22_11_08_09_10_29_28_29_30_11:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,1,2,21,20,21,22,11,8,9,10,29,28,29,30,11]

 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm3[4,5,6,7]
 ; AVX1-NEXT:    vpalignr {{.*#+}} xmm0 = xmm1[10,11,12,13,14,15],xmm0[0,1,2,3,4,5,6,7,8,9]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_21_22_23_00_01_02_03_12_29_30_31_08_09_10_11_12:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpblendw {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm1[5,6,7],ymm0[8,9,10,11,12],ymm1[13,14,15]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpalignr {{.*#+}} ymm0 = ymm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9,26,27,28,29,30,31,16,17,18,19,20,21,22,23,24,25]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_21_22_23_00_01_02_03_12_29_30_31_08_09_10_11_12:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm2 = [21,22,23,0,1,2,3,12,29,30,31,8,9,10,11,12]

 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3],xmm1[4],xmm0[5,6,7]
 ; AVX1-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9]
 ; AVX1-NEXT:    vpalignr {{.*#+}} xmm1 = xmm1[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_05_06_07_00_01_02_03_12_13_14_15_08_09_10_11_12:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpalignr {{.*#+}} ymm0 = ymm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9,26,27,28,29,30,31,16,17,18,19,20,21,22,23,24,25]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_05_06_07_00_01_02_03_12_13_14_15_08_09_10_11_12:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [5,6,7,0,1,2,3,12,13,14,15,8,9,10,11,12]

 ; AVX1-NEXT:    vpsrldq {{.*#+}} xmm1 = xmm1[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1,2,3,4],xmm0[5,6,7]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_19_20_21_22_23_00_01_10_27_28_29_30_31_08_09_10:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpblendw {{.*#+}} ymm0 = ymm0[0,1,2],ymm1[3,4,5,6,7],ymm0[8,9,10],ymm1[11,12,13,14,15]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpalignr {{.*#+}} ymm0 = ymm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5,22,23,24,25,26,27,28,29,30,31,16,17,18,19,20,21]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_19_20_21_22_23_00_01_10_27_28_29_30_31_08_09_10:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm2 = [3,4,5,6,7,16,17,26,11,12,13,14,15,24,25,26]

 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2],xmm0[3,4,5,6,7]
 ; AVX1-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5]
 ; AVX1-NEXT:    vpalignr {{.*#+}} xmm1 = xmm1[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_03_04_05_06_07_00_01_10_11_12_13_14_15_08_09_10:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpalignr {{.*#+}} ymm0 = ymm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5,22,23,24,25,26,27,28,29,30,31,16,17,18,19,20,21]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_03_04_05_06_07_00_01_10_11_12_13_14_15_08_09_10:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [3,4,5,6,7,0,1,10,11,12,13,14,15,8,9,10]

 ; AVX1-NEXT:    vpsrldq {{.*#+}} xmm0 = xmm0[6,7,8,9,10,11,12,13,14,15],zero,zero,zero,zero,zero,zero
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1,2,3,4],xmm1[5,6,7]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_03_04_05_06_07_16_17_26_11_12_13_14_15_24_25_26:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpblendw {{.*#+}} ymm0 = ymm1[0,1,2],ymm0[3,4,5,6,7],ymm1[8,9,10],ymm0[11,12,13,14,15]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpalignr {{.*#+}} ymm0 = ymm0[6,7,8,9,10,11,12,13,14,15,0,1,2,3,4,5,22,23,24,25,26,27,28,29,30,31,16,17,18,19,20,21]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_03_04_05_06_07_16_17_26_11_12_13_14_15_24_25_26:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm2 = [3,4,5,6,7,16,17,26,11,12,13,14,15,24,25,26]

 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0,1,2,3],xmm3[4,5,6,7]
 ; AVX1-NEXT:    vpalignr {{.*#+}} xmm0 = xmm0[10,11,12,13,14,15],xmm1[0,1,2,3,4,5,6,7,8,9]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v16i16_05_06_07_16_17_18_19_28_13_14_15_24_25_26_27_28:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpblendw {{.*#+}} ymm0 = ymm1[0,1,2,3,4],ymm0[5,6,7],ymm1[8,9,10,11,12],ymm0[13,14,15]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpalignr {{.*#+}} ymm0 = ymm0[10,11,12,13,14,15,0,1,2,3,4,5,6,7,8,9,26,27,28,29,30,31,16,17,18,19,20,21,22,23,24,25]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v16i16_05_06_07_16_17_18_19_28_13_14_15_24_25_26_27_28:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm2 = [21,22,23,0,1,2,3,12,29,30,31,8,9,10,11,12]

 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm1 = xmm1[0,1],xmm3[2,3,4,5],xmm1[6,7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[6,7,4,5,4,5,10,11,4,5,14,15,12,13,0,1]
 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm1[0,1],xmm0[2],xmm1[3,4,5],xmm0[6],xmm1[7]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: PR24935:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
-; AVX2-NEXT:    vpshufb {{.*#+}} ymm2 = ymm2[6,7,4,5,0,1,10,11,4,5,10,11,4,5,6,7,22,23,20,21,16,17,26,27,20,21,26,27,20,21,22,23]
-; AVX2-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[8,9,10,11,4,5,8,9,0,1,14,15,12,13,0,1,24,25,26,27,20,21,24,25,16,17,30,31,28,29,16,17]
+; AVX2-NEXT:    vpshufb {{.*#+}} ymm2 = ymm1[8,9,10,11,4,5,8,9,0,1,14,15,12,13,0,1,24,25,26,27,20,21,24,25,16,17,30,31,28,29,16,17]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[2,3,0,1]
+; AVX2-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[6,7,4,5,0,1,10,11,4,5,10,11,4,5,6,7,22,23,20,21,16,17,26,27,20,21,26,27,20,21,22,23]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = <255,255,255,255,u,u,255,255,255,255,0,0,u,u,0,0,u,u,u,u,255,255,0,0,u,u,u,u,u,u,0,0>
-; AVX2-NEXT:    vpblendvb %ymm3, %ymm2, %ymm1, %ymm1
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
-; AVX2-NEXT:    vpshufb {{.*#+}} ymm2 = ymm2[u,u,u,u,u,u,u,u,u,u,u,u,6,7,u,u,18,19,u,u,u,u,u,u,u,u,24,25,16,17,u,u]
-; AVX2-NEXT:    vpshuflw {{.*#+}} ymm0 = ymm0[0,1,1,3,4,5,6,7,8,9,9,11,12,13,14,15]
-; AVX2-NEXT:    vpshufhw {{.*#+}} ymm0 = ymm0[0,1,2,3,5,5,6,7,8,9,10,11,13,13,14,15]
-; AVX2-NEXT:    vpblendw {{.*#+}} ymm0 = ymm2[0],ymm0[1,2],ymm2[3],ymm0[4],ymm2[5,6,7,8],ymm0[9,10],ymm2[11],ymm0[12],ymm2[13,14,15]
+; AVX2-NEXT:    vpblendvb %ymm3, %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vpshuflw {{.*#+}} ymm2 = ymm0[0,1,1,3,4,5,6,7,8,9,9,11,12,13,14,15]
+; AVX2-NEXT:    vpshufhw {{.*#+}} ymm2 = ymm2[0,1,2,3,5,5,6,7,8,9,10,11,13,13,14,15]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[u,u,u,u,u,u,u,u,u,u,u,u,6,7,u,u,18,19,u,u,u,u,u,u,u,u,24,25,16,17,u,u]
+; AVX2-NEXT:    vpblendw {{.*#+}} ymm0 = ymm0[0],ymm2[1,2],ymm0[3],ymm2[4],ymm0[5,6,7,8],ymm2[9,10],ymm0[11],ymm2[12],ymm0[13,14,15]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,0,0,255,255,255,255,255,255,0,0,255,255,0,0,0,0,255,255,255,255,0,0,0,0,0,0,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: PR24935:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm2 = [11,10,17,13,10,7,27,0,17,25,0,12,29,20,16,8]
 ; AVX512VL-NEXT:    vpermi2w %ymm0, %ymm1, %ymm2

 ; AVX1-NEXT:    vpblendw {{.*#+}} xmm0 = xmm0[0,1],xmm1[2],xmm0[3,4,5,6,7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[6,7,0,1,0,1],zero,zero,xmm0[10,11],zero,zero,zero,zero,xmm0[4,5]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm1 = xmm1[14,15,0,1],zero,zero,xmm1[0,1,2,3,4,5,8,9,8,9]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: PR34369:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <255,255,u,u,0,0,255,255,u,u,255,255,u,u,u,u,255,255,255,255,255,255,u,u,255,255,u,u,u,u,255,255>
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[6,7,0,1,0,1],zero,zero,ymm0[10,11],zero,zero,zero,zero,ymm0[4,5,30,31,16,17],zero,zero,ymm0[16,17,18,19,20,21,24,25,24,25]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: PR34369:
 ; AVX512VL:       # BB#0:
 ; AVX512VL-NEXT:    vmovdqa {{.*#+}} ymm1 = [3,0,0,13,5,2,2,10,15,8,14,8,9,10,12,12]

llvm/trunk/test/CodeGen/X86/vector-shuffle-256-v32.ll

 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
 ; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_16_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; AVX2-NEXT:    vpshufb %ymm2, %ymm1, %ymm1
 ; AVX2-NEXT:    vpbroadcastb %xmm0, %xmm0
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [0,0,0,0,0,0,0,0,0,0,0,0,0,0,0,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_16_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512VL-NEXT:    vpxor %xmm2, %xmm2, %xmm2
 ; AVX512VL-NEXT:    vpshufb %ymm2, %ymm1, %ymm1
 ; AVX512VL-NEXT:    vpbroadcastb %xmm0, %xmm0
 ; AVX512VL-NEXT:    movl $32767, %eax # imm = 0x7FFF
 ; AVX512VL-NEXT:    kmovd %eax, %k1
 ; AVX512VL-NEXT:    vmovdqu8 %ymm0, %ymm1 {%k1}
 ; AVX512VL-NEXT:    vmovdqa %ymm1, %ymm0
 ; AVX512VL-NEXT:    retq

 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
 ; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,0,0,0,0,0,0,3,0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_17_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,255,u,u,u,u,u,u,u,u,u,u,u,u,u,u,255,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_17_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512VL-NEXT:    movl $1, %eax
 ; AVX512VL-NEXT:    kmovd %eax, %k1
 ; AVX512VL-NEXT:    vmovdqu8 %ymm0, %ymm1 {%k1}
 ; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm1[0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 17, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_18_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {
 ; AVX1-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_18_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm1
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
 ; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,0,0,0,0,0,5,0,0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_18_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,0,255,255,u,u,u,u,u,u,u,u,u,u,u,u,255,255,u,u,u,u,u,u,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,0,0,0,0,0,0,0,0,2,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_18_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512VL-NEXT:    movw $1, %ax
 ; AVX512VL-NEXT:    kmovd %eax, %k1
 ; AVX512VL-NEXT:    vmovdqu16 %ymm0, %ymm1 {%k1}
 ; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm1[0,0,0,0,0,0,0,0,0,0,0,0,0,2,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 18, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_19_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {
 ; AVX1-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_19_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm1
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
 ; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,0,0,0,0,7,0,0,0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_19_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = <0,0,255,255,u,u,u,u,u,u,u,u,u,u,u,u,255,255,u,u,u,u,u,u,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,0,0,0,0,0,0,0,3,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_19_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512VL-NEXT:    movw $1, %ax
 ; AVX512VL-NEXT:    kmovd %eax, %k1
 ; AVX512VL-NEXT:    vmovdqu16 %ymm0, %ymm1 {%k1}
 ; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm1[0,0,0,0,0,0,0,0,0,0,0,0,3,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 19, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_20_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {
 ; AVX1-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_20_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm1
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
 ; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,0,0,0,9,0,0,0,0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2OR512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_20_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2OR512VL:       # BB#0:
-; AVX2OR512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2OR512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3,4,5,6,7]
 ; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,0,0,0,0,0,0,4,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 20, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_21_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {
 ; AVX1-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_21_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm1
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
 ; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,0,0,11,0,0,0,0,0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2OR512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_21_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2OR512VL:       # BB#0:
-; AVX2OR512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2OR512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3,4,5,6,7]
 ; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,0,0,0,0,0,5,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 21, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_00_00_22_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {
 ; AVX1-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_22_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm1
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
 ; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,0,13,0,0,0,0,0,0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2OR512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_22_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2OR512VL:       # BB#0:
-; AVX2OR512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2OR512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3,4,5,6,7]
 ; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,0,0,0,0,6,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 22, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_00_23_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {
 ; AVX1-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_23_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm1
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
 ; AVX1-NEXT:    vpunpcklbw {{.*#+}} xmm0 = xmm0[0],xmm2[0],xmm0[1],xmm2[1],xmm0[2],xmm2[2],xmm0[3],xmm2[3],xmm0[4],xmm2[4],xmm0[5],xmm2[5],xmm0[6],xmm2[6],xmm0[7],xmm2[7]
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,15,0,0,0,0,0,0,0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2OR512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_23_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2OR512VL:       # BB#0:
-; AVX2OR512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX2OR512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0],ymm1[1,2,3,4,5,6,7]
 ; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,0,0,0,7,0,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 23, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_24_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {

 ; AVX2-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_24_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,3,0,1]
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,0,0,8,0,0,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_24_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
 ; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,0,0,8,0,0,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 24, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_25_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {

 ; AVX2-LABEL: shuffle_v32i8_00_00_00_00_00_00_25_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,3,0,1]
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,0,9,0,0,0,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_25_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
 ; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,0,9,0,0,0,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 25, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_00_00_00_00_26_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {

 ; AVX2-LABEL: shuffle_v32i8_00_00_00_00_00_26_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,3,0,1]
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,10,0,0,0,0,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_00_00_00_00_00_26_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
 ; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,0,10,0,0,0,0,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 26, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_00_00_00_27_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {

 ; AVX2-LABEL: shuffle_v32i8_00_00_00_00_27_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,3,0,1]
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,11,0,0,0,0,0,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_00_00_00_00_27_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
 ; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,0,11,0,0,0,0,0,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 27, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_00_00_28_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {

 ; AVX2-LABEL: shuffle_v32i8_00_00_00_28_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,3,0,1]
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,12,0,0,0,0,0,0,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_00_00_00_28_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
 ; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,0,12,0,0,0,0,0,0,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 28, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_00_29_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {

 ; AVX2-LABEL: shuffle_v32i8_00_00_29_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,3,0,1]
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,13,0,0,0,0,0,0,0,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_00_00_29_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
 ; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,0,13,0,0,0,0,0,0,0,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 29, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_30_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {

 ; AVX2-LABEL: shuffle_v32i8_00_30_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,3,0,1]
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,14,0,0,0,0,0,0,0,0,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_00_30_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
 ; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[0,14,0,0,0,0,0,0,0,0,0,0,0,0,0,0,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16,16]
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 30, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_31_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {

 ; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,3,0,1]
 ; AVX2-NEXT:    movl $15, %eax
 ; AVX2-NEXT:    vmovd %eax, %xmm1
 ; AVX2-NEXT:    vpshufb %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_31_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512VL-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm1[2,3,4,5,6,7]
 ; AVX512VL-NEXT:    movl $15, %eax
 ; AVX512VL-NEXT:    vmovd %eax, %xmm1
 ; AVX512VL-NEXT:    vpshufb %ymm1, %ymm0, %ymm0
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 31, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }

 
 define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_01_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_01_00(<32 x i8> %a, <32 x i8> %b) {
 ; AVX1-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_01_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_01_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_01_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_01_00:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX2-NEXT:    retq
-;
-; AVX512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_01_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_01_00:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0]
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_00_01_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_01_00:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,0,0,0,0,0,0,1,0]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_02_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_02_00_00(<32 x i8> %a, <32 x i8> %b) {
 ; AVX1-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_02_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_02_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,0,0,0,0,0,2,0,0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_02_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_02_00_00:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,0,0,0,0,0,2,0,0]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX2-NEXT:    retq
-;
-; AVX512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_02_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_02_00_00:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,0,0,0,0,0,2,0,0]
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_00_00_00_00_00_02_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_02_00_00:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,0,0,0,0,0,2,0,0]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 2, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 2, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_00_07_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_07_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {
 ; AVX1-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_07_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_07_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,7,0,0,0,0,0,0,0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_07_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_07_00_00_00_00_00_00_00:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,7,0,0,0,0,0,0,0]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX2-NEXT:    retq
-;
-; AVX512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_07_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_07_00_00_00_00_00_00_00:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,7,0,0,0,0,0,0,0]
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_00_07_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_07_00_00_00_00_00_00_00:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,0,7,0,0,0,0,0,0,0]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 7, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 7, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_00_00_00_00_00_00_08_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_08_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {
 ; AVX1-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_08_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_08_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,8,0,0,0,0,0,0,0,0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_08_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_08_00_00_00_00_00_00_00_00:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,8,0,0,0,0,0,0,0,0]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX2-NEXT:    retq
-;
-; AVX512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_08_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_08_00_00_00_00_00_00_00_00:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,8,0,0,0,0,0,0,0,0]
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v32i8_00_00_00_00_00_00_00_08_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_08_00_00_00_00_00_00_00_00:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,0,0,0,0,0,0,8,0,0,0,0,0,0,0,0]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 8, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 8, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_14_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_14_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {
 ; AVX1-LABEL: shuffle_v32i8_00_14_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_14_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,14,0,0,0,0,0,0,0,0,0,0,0,0,0,0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-LABEL: shuffle_v32i8_00_14_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_14_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,14,0,0,0,0,0,0,0,0,0,0,0,0,0,0]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX2-NEXT:    retq
-;
-; AVX512VL-LABEL: shuffle_v32i8_00_14_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_14_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,14,0,0,0,0,0,0,0,0,0,0,0,0,0,0]
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v32i8_00_14_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_14_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[0,14,0,0,0,0,0,0,0,0,0,0,0,0,0,0]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 0, i32 14, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 14, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_15_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_15_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00(<32 x i8> %a, <32 x i8> %b) {
 ; AVX1-LABEL: shuffle_v32i8_15_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_15_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    movl $15, %eax
 ; AVX1-NEXT:    vmovd %eax, %xmm1
 ; AVX1-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-LABEL: shuffle_v32i8_15_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_15_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    movl $15, %eax
-; AVX2-NEXT:    vmovd %eax, %xmm1
-; AVX2-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX2-NEXT:    retq
-;
-; AVX512VL-LABEL: shuffle_v32i8_15_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_15_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    movl $15, %eax
-; AVX512VL-NEXT:    vmovd %eax, %xmm1
-; AVX512VL-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v32i8_15_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00_15_00_00_00_00_00_00_00_00_00_00_00_00_00_00_00:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    movl $15, %eax
+; AVX2OR512VL-NEXT:    vmovd %eax, %xmm1
+; AVX2OR512VL-NEXT:    vpshufb %xmm1, %xmm0, %xmm0
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 15, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 15, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <32 x i8> %shuffle
 }
 
 define <32 x i8> @shuffle_v32i8_00_33_02_35_04_37_06_39_08_41_10_43_12_45_14_47_16_49_18_51_20_53_22_55_24_57_26_59_28_61_30_63(<32 x i8> %a, <32 x i8> %b) {
 ; AVX1-LABEL: shuffle_v32i8_00_33_02_35_04_37_06_39_08_41_10_43_12_45_14_47_16_49_18_51_20_53_22_55_24_57_26_59_28_61_30_63:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vmovaps {{.*#+}} ymm2 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]

 ; AVX1-NEXT:    vpor %xmm2, %xmm0, %xmm0
 ; AVX1-NEXT:    vmovdqa {{.*#+}} xmm2 = [255,255,0,0,255,255,255,255,0,0,0,255,255,0,255,255]
 ; AVX1-NEXT:    vpblendvb %xmm2, %xmm1, %xmm0, %xmm0
 ; AVX1-NEXT:    vinsertf128 $1, %xmm3, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: shuffle_v32i8_42_45_12_13_35_35_60_40_17_22_29_44_33_12_48_51_20_19_52_19_49_54_37_32_48_42_59_07_36_34_36_39:
 ; AVX2:       # BB#0:
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
-; AVX2-NEXT:    vpshufb {{.*#+}} ymm2 = ymm2[u,u,u,u,u,u,12,u,u,u,u,u,u,u,0,3,u,u,u,u,u,u,21,16,u,26,u,u,20,18,20,23]
-; AVX2-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[10,13,u,u,3,3,u,8,u,u,u,12,1,u,u,u,u,u,20,u,17,22,u,u,16,u,27,u,u,u,u,u]
+; AVX2-NEXT:    vpshufb {{.*#+}} ymm2 = ymm1[10,13,u,u,3,3,u,8,u,u,u,12,1,u,u,u,u,u,20,u,17,22,u,u,16,u,27,u,u,u,u,u]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[2,3,0,1]
+; AVX2-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[u,u,u,u,u,u,12,u,u,u,u,u,u,u,0,3,u,u,u,u,u,u,21,16,u,26,u,u,20,18,20,23]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm3 = <255,255,u,u,255,255,0,255,u,u,u,255,255,u,0,0,u,u,255,u,255,255,0,0,255,0,255,u,0,0,0,0>
-; AVX2-NEXT:    vpblendvb %ymm3, %ymm1, %ymm2, %ymm1
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
-; AVX2-NEXT:    vpshufb {{.*#+}} ymm2 = ymm2[u,u,u,u,u,u,u,u,1,6,13,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,23,u,u,u,u]
-; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[u,u,12,13,u,u,u,u,u,u,u,u,u,12,u,u,20,19,u,19,u,u,u,u,u,u,u,u,u,u,u,u]
-; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm2[2],ymm0[3,4,5],ymm2[6],ymm0[7]
+; AVX2-NEXT:    vpblendvb %ymm3, %ymm2, %ymm1, %ymm1
+; AVX2-NEXT:    vpshufb {{.*#+}} ymm2 = ymm0[u,u,12,13,u,u,u,u,u,u,u,u,u,12,u,u,20,19,u,19,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[u,u,u,u,u,u,u,u,1,6,13,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,23,u,u,u,u]
+; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm2[0,1],ymm0[2],ymm2[3,4,5],ymm0[6],ymm2[7]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,0,0,255,255,255,255,0,0,0,255,255,0,255,255,0,0,255,0,255,255,255,255,255,255,255,0,255,255,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm2, %ymm1, %ymm0, %ymm0
 ; AVX2-NEXT:    retq
 ;
 ; AVX512VL-LABEL: shuffle_v32i8_42_45_12_13_35_35_60_40_17_22_29_44_33_12_48_51_20_19_52_19_49_54_37_32_48_42_59_07_36_34_36_39:
 ; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
+; AVX512VL-NEXT:    vpermq {{.*#+}} ymm2 = ymm1[2,3,0,1]
 ; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[10,13,u,u,3,3,u,8,u,u,u,12,1,u,u,u,u,u,20,u,17,22,u,u,16,u,27,u,u,u,u,u]
 ; AVX512VL-NEXT:    movl $-222248896, %eax # imm = 0xF2C0C040
 ; AVX512VL-NEXT:    kmovd %eax, %k1
 ; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm1 {%k1} = ymm2[u,u,u,u,u,u,12,u,u,u,u,u,u,u,0,3,u,u,u,u,u,u,21,16,u,26,u,u,20,18,20,23]
-; AVX512VL-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
-; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm2 = ymm2[u,u,u,u,u,u,u,u,1,6,13,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,23,u,u,u,u]
-; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[u,u,12,13,u,u,u,u,u,u,u,u,u,12,u,u,20,19,u,19,u,u,u,u,u,u,u,u,u,u,u,u]
-; AVX512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1],ymm2[2],ymm0[3,4,5],ymm2[6],ymm0[7]
+; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm2 = ymm0[u,u,12,13,u,u,u,u,u,u,u,u,u,12,u,u,20,19,u,19,u,u,u,u,u,u,u,u,u,u,u,u]
+; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX512VL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[u,u,u,u,u,u,u,u,1,6,13,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,u,23,u,u,u,u]
+; AVX512VL-NEXT:    vpblendd {{.*#+}} ymm0 = ymm2[0,1],ymm0[2],ymm2[3,4,5],ymm0[6],ymm2[7]
 ; AVX512VL-NEXT:    movl $134948620, %eax # imm = 0x80B270C
 ; AVX512VL-NEXT:    kmovd %eax, %k1
 ; AVX512VL-NEXT:    vmovdqu8 %ymm0, %ymm1 {%k1}
 ; AVX512VL-NEXT:    vmovdqa %ymm1, %ymm0
 ; AVX512VL-NEXT:    retq
   %shuffle = shufflevector <32 x i8> %a, <32 x i8> %b, <32 x i32> <i32 42, i32 45, i32 12, i32 13, i32 35, i32 35, i32 60, i32 40, i32 17, i32 22, i32 29, i32 44, i32 33, i32 12, i32 48, i32 51, i32 20, i32 19, i32 52, i32 19, i32 49, i32 54, i32 37, i32 32, i32 48, i32 42, i32 59, i32 7, i32 36, i32 34, i32 36, i32 39>
   ret <32 x i8> %shuffle
 }

llvm/trunk/test/CodeGen/X86/vector-shuffle-256-v8.ll

 
 define <8 x float> @shuffle_v8f32_32103210(<8 x float> %a, <8 x float> %b) {
 ; AVX1-LABEL: shuffle_v8f32_32103210:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[3,2,1,0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-LABEL: shuffle_v8f32_32103210:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[3,2,1,0]
-; AVX2-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX2-NEXT:    retq
-;
-; AVX512VL-LABEL: shuffle_v8f32_32103210:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[3,2,1,0]
-; AVX512VL-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v8f32_32103210:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[3,2,1,0]
+; AVX2OR512VL-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 3, i32 2, i32 1, i32 0>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_76547654(<8 x float> %a, <8 x float> %b) {
-; AVX1OR2-LABEL: shuffle_v8f32_76547654:
-; AVX1OR2:       # BB#0:
-; AVX1OR2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX1OR2-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
-; AVX1OR2-NEXT:    retq
+; AVX1-LABEL: shuffle_v8f32_76547654:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX1-NEXT:    retq
 ;
-; AVX512VL-LABEL: shuffle_v8f32_76547654:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX512VL-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[2,3,2,3]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v8f32_76547654:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX2OR512VL-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 7, i32 6, i32 5, i32 4>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_76543210(<8 x float> %a, <8 x float> %b) {
-; AVX1OR2-LABEL: shuffle_v8f32_76543210:
-; AVX1OR2:       # BB#0:
-; AVX1OR2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX1OR2-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
-; AVX1OR2-NEXT:    retq
+; AVX1-LABEL: shuffle_v8f32_76543210:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX1-NEXT:    retq
 ;
-; AVX512VL-LABEL: shuffle_v8f32_76543210:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX512VL-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[2,3,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v8f32_76543210:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX2OR512VL-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_3210ba98(<8 x float> %a, <8 x float> %b) {
 ; ALL-LABEL: shuffle_v8f32_3210ba98:
 ; ALL:       # BB#0:
 ; ALL-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0

 
 define <8 x i32> @shuffle_v8i32_32103210(<8 x i32> %a, <8 x i32> %b) {
 ; AVX1-LABEL: shuffle_v8i32_32103210:
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[3,2,1,0]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
 ; AVX1-NEXT:    retq
 ;
-; AVX2-LABEL: shuffle_v8i32_32103210:
-; AVX2:       # BB#0:
-; AVX2-NEXT:    vpermilps {{.*#+}} xmm0 = xmm0[3,2,1,0]
-; AVX2-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX2-NEXT:    retq
-;
-; AVX512VL-LABEL: shuffle_v8i32_32103210:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,2,1,0]
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v8i32_32103210:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpshufd {{.*#+}} xmm0 = xmm0[3,2,1,0]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 3, i32 2, i32 1, i32 0>
   ret <8 x i32> %shuffle
 }
 
 define <8 x i32> @shuffle_v8i32_76547654(<8 x i32> %a, <8 x i32> %b) {
-; AVX1OR2-LABEL: shuffle_v8i32_76547654:
-; AVX1OR2:       # BB#0:
-; AVX1OR2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX1OR2-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
-; AVX1OR2-NEXT:    retq
+; AVX1-LABEL: shuffle_v8i32_76547654:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX1-NEXT:    retq
 ;
-; AVX512VL-LABEL: shuffle_v8i32_76547654:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,2,3]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v8i32_76547654:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 7, i32 6, i32 5, i32 4>
   ret <8 x i32> %shuffle
 }
 
 define <8 x i32> @shuffle_v8i32_76543210(<8 x i32> %a, <8 x i32> %b) {
-; AVX1OR2-LABEL: shuffle_v8i32_76543210:
-; AVX1OR2:       # BB#0:
-; AVX1OR2-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX1OR2-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
-; AVX1OR2-NEXT:    retq
+; AVX1-LABEL: shuffle_v8i32_76543210:
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX1-NEXT:    retq
 ;
-; AVX512VL-LABEL: shuffle_v8i32_76543210:
-; AVX512VL:       # BB#0:
-; AVX512VL-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
-; AVX512VL-NEXT:    retq
+; AVX2OR512VL-LABEL: shuffle_v8i32_76543210:
+; AVX2OR512VL:       # BB#0:
+; AVX2OR512VL-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
+; AVX2OR512VL-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX2OR512VL-NEXT:    retq
   %shuffle = shufflevector <8 x i32> %a, <8 x i32> %b, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
   ret <8 x i32> %shuffle
 }
 
 define <8 x i32> @shuffle_v8i32_3210ba98(<8 x i32> %a, <8 x i32> %b) {
 ; ALL-LABEL: shuffle_v8i32_3210ba98:
 ; ALL:       # BB#0:
 ; ALL-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0

llvm/trunk/test/CodeGen/X86/vector-shuffle-512-v32.ll

 ; SKX-NEXT:    retq
   %c = shufflevector <32 x i16> %a, <32 x i16> undef, <32 x i32> <i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8, i32 8>
   ret <32 x i16> %c
 }
 
 define <32 x i16> @shuffle_v32i16_02_05_u_u_07_u_0a_01_00_05_u_04_07_u_0a_01_02_05_u_u_07_u_0a_01_00_05_u_04_07_u_0a_1f(<32 x i16> %a)  {
 ; KNL-LABEL: shuffle_v32i16_02_05_u_u_07_u_0a_01_00_05_u_04_07_u_0a_01_02_05_u_u_07_u_0a_01_00_05_u_04_07_u_0a_1f:
 ; KNL:       ## BB#0:
-; KNL-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm0[2,3,0,1]
-; KNL-NEXT:    vpshufb {{.*#+}} ymm3 = ymm2[0,1,10,11,8,9,8,9,14,15,2,3,4,5,2,3,16,17,26,27,24,25,24,25,30,31,18,19,20,21,18,19]
-; KNL-NEXT:    vpshufb {{.*#+}} ymm4 = ymm0[4,5,10,11,4,5,6,7,14,15,2,3,4,5,2,3,20,21,26,27,20,21,22,23,30,31,18,19,20,21,18,19]
-; KNL-NEXT:    vmovdqa {{.*#+}} ymm0 = <0,0,0,0,u,u,u,u,0,0,u,u,255,255,0,0,255,255,255,255,u,u,255,255,255,255,u,u,0,0,255,255>
-; KNL-NEXT:    vpblendvb %ymm0, %ymm3, %ymm4, %ymm0
-; KNL-NEXT:    vpshufb {{.*#+}} ymm2 = ymm2[0,1,10,11,8,9,8,9,14,15,6,7,4,5,14,15,16,17,26,27,24,25,24,25,30,31,22,23,20,21,30,31]
-; KNL-NEXT:    vmovdqa {{.*#+}} ymm3 = <255,255,255,255,u,u,u,u,255,255,u,u,0,0,255,255,0,0,0,0,u,u,0,0,0,0,u,u,255,255,u,u>
-; KNL-NEXT:    vpblendvb %ymm3, %ymm4, %ymm2, %ymm2
+; KNL-NEXT:    vpshufb {{.*#+}} ymm2 = ymm0[4,5,10,11,4,5,6,7,14,15,2,3,4,5,2,3,20,21,26,27,20,21,22,23,30,31,18,19,20,21,18,19]
+; KNL-NEXT:    vpermq {{.*#+}} ymm3 = ymm0[2,3,0,1]
+; KNL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm3[0,1,10,11,8,9,8,9,14,15,2,3,4,5,2,3,16,17,26,27,24,25,24,25,30,31,18,19,20,21,18,19]
+; KNL-NEXT:    vmovdqa {{.*#+}} ymm4 = <0,0,0,0,u,u,u,u,0,0,u,u,255,255,0,0,255,255,255,255,u,u,255,255,255,255,u,u,0,0,255,255>
+; KNL-NEXT:    vpblendvb %ymm4, %ymm0, %ymm2, %ymm0
+; KNL-NEXT:    vpshufb {{.*#+}} ymm3 = ymm3[0,1,10,11,8,9,8,9,14,15,6,7,4,5,14,15,16,17,26,27,24,25,24,25,30,31,22,23,20,21,30,31]
+; KNL-NEXT:    vmovdqa {{.*#+}} ymm4 = <255,255,255,255,u,u,u,u,255,255,u,u,0,0,255,255,0,0,0,0,u,u,0,0,0,0,u,u,255,255,u,u>
+; KNL-NEXT:    vpblendvb %ymm4, %ymm2, %ymm3, %ymm2
 ; KNL-NEXT:    vmovdqa {{.*#+}} ymm3 = <255,255,255,255,u,u,u,u,255,255,u,u,255,255,255,255,255,255,255,255,u,u,255,255,255,255,u,u,255,255,0,0>
 ; KNL-NEXT:    vpblendvb %ymm3, %ymm2, %ymm1, %ymm1
 ; KNL-NEXT:    retq
 ;
 ; SKX-LABEL: shuffle_v32i16_02_05_u_u_07_u_0a_01_00_05_u_04_07_u_0a_01_02_05_u_u_07_u_0a_01_00_05_u_04_07_u_0a_1f:
 ; SKX:       ## BB#0:
 ; SKX-NEXT:    vmovdqa64 {{.*#+}} zmm1 = <2,5,u,u,7,u,10,1,0,5,u,4,7,u,10,1,2,5,u,u,7,u,10,1,0,5,u,4,7,u,10,31>
 ; SKX-NEXT:    vpermw %zmm0, %zmm1, %zmm0
 ; SKX-NEXT:    retq
   %c = shufflevector <32 x i16> %a, <32 x i16> undef, <32 x i32> <i32 2, i32 5, i32 undef, i32 undef, i32 7, i32 undef, i32 10, i32 1,  i32 0, i32 5, i32 undef, i32 4, i32 7, i32 undef, i32 10, i32 1, i32 2, i32 5, i32 undef, i32 undef, i32 7, i32 undef, i32 10, i32 1,  i32 0, i32 5, i32 undef, i32 4, i32 7, i32 undef, i32 10, i32 31>
   ret <32 x i16> %c
 }
 
 define <32 x i16> @shuffle_v32i16_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_38(<32 x i16> %a, <32 x i16> %b)  {
 ; KNL-LABEL: shuffle_v32i16_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_18_0f_1f_0e_16_0d_1d_04_1e_0b_1b_0a_1a_09_19_08_38:
 ; KNL:       ## BB#0:
-; KNL-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
+; KNL-NEXT:    vpermq {{.*#+}} ymm2 = ymm1[2,3,0,1]
 ; KNL-NEXT:    vpblendw {{.*#+}} ymm2 = ymm1[0,1,2,3],ymm2[4,5],ymm1[6],ymm2[7],ymm1[8,9,10,11],ymm2[12,13],ymm1[14],ymm2[15]
 ; KNL-NEXT:    vpshufb {{.*#+}} ymm1 = ymm2[u,u,14,15,u,u,12,13,u,u,10,11,u,u,8,9,u,u,22,23,u,u,20,21,u,u,18,19,u,u,u,u]
-; KNL-NEXT:    vperm2i128 {{.*#+}} ymm4 = ymm0[2,3,0,1]
+; KNL-NEXT:    vpermq {{.*#+}} ymm4 = ymm0[2,3,0,1]
 ; KNL-NEXT:    vpblendw {{.*#+}} ymm0 = ymm0[0,1,2,3,4],ymm4[5,6,7],ymm0[8,9,10,11,12],ymm4[13,14,15]
 ; KNL-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[14,15,u,u,12,13,u,u,10,11,u,u,8,9,u,u,22,23,u,u,20,21,u,u,18,19,u,u,16,17,u,u]
 ; KNL-NEXT:    vpblendw {{.*#+}} ymm1 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7],ymm0[8],ymm1[9],ymm0[10],ymm1[11],ymm0[12],ymm1[13],ymm0[14],ymm1[15]
 ; KNL-NEXT:    vextracti128 $1, %ymm3, %xmm3
 ; KNL-NEXT:    vpbroadcastw %xmm3, %ymm3
 ; KNL-NEXT:    vmovdqa {{.*#+}} ymm4 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0]
 ; KNL-NEXT:    vpblendvb %ymm4, %ymm1, %ymm3, %ymm1
 ; KNL-NEXT:    vpshufb {{.*#+}} ymm2 = ymm2[u,u,14,15,u,u,12,13,u,u,10,11,u,u,8,9,u,u,22,23,u,u,20,21,u,u,18,19,u,u,16,17]

llvm/trunk/test/CodeGen/X86/vector-shuffle-512-v64.ll

; AVX512VBMI-NEXT:    retq
   ret <64 x i8> %shuffle
 }
 
 define <64 x i8> @shuffle_v64i8_63_62_61_60_59_58_57_56_55_54_53_52_51_50_49_48_47_46_45_44_43_42_41_40_39_38_37_36_35_34_33_32_31_30_29_28_27_26_25_24_23_22_21_20_19_18_17_16_15_14_13_12_11_10_09_08_07_06_05_04_03_02_01_00(<64 x i8> %a) {
 ; AVX512F-LABEL: shuffle_v64i8_63_62_61_60_59_58_57_56_55_54_53_52_51_50_49_48_47_46_45_44_43_42_41_40_39_38_37_36_35_34_33_32_31_30_29_28_27_26_25_24_23_22_21_20_19_18_17_16_15_14_13_12_11_10_09_08_07_06_05_04_03_02_01_00:
 ; AVX512F:       # BB#0:
 ; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm3 = [15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0,15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0]
 ; AVX512F-NEXT:    vpshufb %ymm3, %ymm1, %ymm1
-; AVX512F-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
+; AVX512F-NEXT:    vpermq {{.*#+}} ymm2 = ymm1[2,3,0,1]
 ; AVX512F-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
-; AVX512F-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512F-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512F-NEXT:    vmovdqa %ymm2, %ymm0
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: shuffle_v64i8_63_62_61_60_59_58_57_56_55_54_53_52_51_50_49_48_47_46_45_44_43_42_41_40_39_38_37_36_35_34_33_32_31_30_29_28_27_26_25_24_23_22_21_20_19_18_17_16_15_14_13_12_11_10_09_08_07_06_05_04_03_02_01_00:
 ; AVX512BW:       # BB#0:
 ; AVX512BW-NEXT:    vpshufb {{.*#+}} zmm0 = zmm0[15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0,31,30,29,28,27,26,25,24,23,22,21,20,19,18,17,16,47,46,45,44,43,42,41,40,39,38,37,36,35,34,33,32,63,62,61,60,59,58,57,56,55,54,53,52,51,50,49,48]
 ; AVX512BW-NEXT:    vshufi64x2 {{.*#+}} zmm0 = zmm0[6,7,4,5,2,3,0,1]
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512DQ-LABEL: shuffle_v64i8_63_62_61_60_59_58_57_56_55_54_53_52_51_50_49_48_47_46_45_44_43_42_41_40_39_38_37_36_35_34_33_32_31_30_29_28_27_26_25_24_23_22_21_20_19_18_17_16_15_14_13_12_11_10_09_08_07_06_05_04_03_02_01_00:
 ; AVX512DQ:       # BB#0:
 ; AVX512DQ-NEXT:    vmovdqa {{.*#+}} ymm3 = [15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0,15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0]
 ; AVX512DQ-NEXT:    vpshufb %ymm3, %ymm1, %ymm1
-; AVX512DQ-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
+; AVX512DQ-NEXT:    vpermq {{.*#+}} ymm2 = ymm1[2,3,0,1]
 ; AVX512DQ-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
-; AVX512DQ-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512DQ-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512DQ-NEXT:    vmovdqa %ymm2, %ymm0
 ; AVX512DQ-NEXT:    retq
 ;
 ; AVX512VBMI-LABEL: shuffle_v64i8_63_62_61_60_59_58_57_56_55_54_53_52_51_50_49_48_47_46_45_44_43_42_41_40_39_38_37_36_35_34_33_32_31_30_29_28_27_26_25_24_23_22_21_20_19_18_17_16_15_14_13_12_11_10_09_08_07_06_05_04_03_02_01_00:
 ; AVX512VBMI:       # BB#0:
 ; AVX512VBMI-NEXT:    vmovdqa64 {{.*#+}} zmm1 = [63,62,61,60,59,58,57,56,55,54,53,52,51,50,49,48,47,46,45,44,43,42,41,40,39,38,37,36,35,34,33,32,31,30,29,28,27,26,25,24,23,22,21,20,19,18,17,16,15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0]
 ; AVX512VBMI-NEXT:    vpermb %zmm0, %zmm1, %zmm0
 ; AVX512VBMI-NEXT:    retq
; AVX512VBMI-NEXT:    retq
   ret <64 x i8> %shuffle
 }
 
 define <64 x i8> @shuffle_v64i8_63_zz_61_zz_59_zz_57_zz_55_zz_53_zz_51_zz_49_zz_47_zz_45_zz_43_zz_41_zz_39_zz_37_zz_35_zz_33_zz_31_zz_29_zz_27_zz_25_zz_23_zz_21_zz_19_zz_17_zz_15_zz_13_zz_11_zz_9_zz_7_zz_5_zz_3_zz_1_zz(<64 x i8> %a) {
 ; AVX512F-LABEL: shuffle_v64i8_63_zz_61_zz_59_zz_57_zz_55_zz_53_zz_51_zz_49_zz_47_zz_45_zz_43_zz_41_zz_39_zz_37_zz_35_zz_33_zz_31_zz_29_zz_27_zz_25_zz_23_zz_21_zz_19_zz_17_zz_15_zz_13_zz_11_zz_9_zz_7_zz_5_zz_3_zz_1_zz:
 ; AVX512F:       # BB#0:
 ; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm3 = <15,u,13,u,11,u,9,u,7,u,5,u,3,u,1,u,15,u,13,u,11,u,9,u,7,u,5,u,3,u,1,u>
 ; AVX512F-NEXT:    vpshufb %ymm3, %ymm1, %ymm1
-; AVX512F-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm1[2,3,0,1]
+; AVX512F-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[2,3,0,1]
 ; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm4 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; AVX512F-NEXT:    vpand %ymm4, %ymm1, %ymm2
 ; AVX512F-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
-; AVX512F-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX512F-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
 ; AVX512F-NEXT:    vpand %ymm4, %ymm0, %ymm1
 ; AVX512F-NEXT:    vmovdqa %ymm2, %ymm0
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: shuffle_v64i8_63_zz_61_zz_59_zz_57_zz_55_zz_53_zz_51_zz_49_zz_47_zz_45_zz_43_zz_41_zz_39_zz_37_zz_35_zz_33_zz_31_zz_29_zz_27_zz_25_zz_23_zz_21_zz_19_zz_17_zz_15_zz_13_zz_11_zz_9_zz_7_zz_5_zz_3_zz_1_zz:
 ; AVX512BW:       # BB#0:
 ; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm1 = <15,u,13,u,11,u,9,u,7,u,5,u,3,u,1,u,15,u,13,u,11,u,9,u,7,u,5,u,3,u,1,u>
 ; AVX512BW-NEXT:    vpshufb %ymm1, %ymm0, %ymm2
-; AVX512BW-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm2[2,3,0,1]
+; AVX512BW-NEXT:    vpermq {{.*#+}} ymm2 = ymm2[2,3,0,1]
 ; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm3 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; AVX512BW-NEXT:    vpand %ymm3, %ymm2, %ymm2
 ; AVX512BW-NEXT:    vextracti64x4 $1, %zmm0, %ymm0
 ; AVX512BW-NEXT:    vpshufb %ymm1, %ymm0, %ymm0
-; AVX512BW-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX512BW-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
 ; AVX512BW-NEXT:    vpand %ymm3, %ymm0, %ymm0
 ; AVX512BW-NEXT:    vinserti64x4 $1, %ymm2, %zmm0, %zmm0
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512DQ-LABEL: shuffle_v64i8_63_zz_61_zz_59_zz_57_zz_55_zz_53_zz_51_zz_49_zz_47_zz_45_zz_43_zz_41_zz_39_zz_37_zz_35_zz_33_zz_31_zz_29_zz_27_zz_25_zz_23_zz_21_zz_19_zz_17_zz_15_zz_13_zz_11_zz_9_zz_7_zz_5_zz_3_zz_1_zz:
 ; AVX512DQ:       # BB#0:
 ; AVX512DQ-NEXT:    vmovdqa {{.*#+}} ymm3 = <15,u,13,u,11,u,9,u,7,u,5,u,3,u,1,u,15,u,13,u,11,u,9,u,7,u,5,u,3,u,1,u>
 ; AVX512DQ-NEXT:    vpshufb %ymm3, %ymm1, %ymm1
-; AVX512DQ-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm1[2,3,0,1]
+; AVX512DQ-NEXT:    vpermq {{.*#+}} ymm1 = ymm1[2,3,0,1]
 ; AVX512DQ-NEXT:    vmovdqa {{.*#+}} ymm4 = [255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0,255,0]
 ; AVX512DQ-NEXT:    vpand %ymm4, %ymm1, %ymm2
 ; AVX512DQ-NEXT:    vpshufb %ymm3, %ymm0, %ymm0
-; AVX512DQ-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX512DQ-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
 ; AVX512DQ-NEXT:    vpand %ymm4, %ymm0, %ymm1
 ; AVX512DQ-NEXT:    vmovdqa %ymm2, %ymm0
 ; AVX512DQ-NEXT:    retq
 ;
 ; AVX512VBMI-LABEL: shuffle_v64i8_63_zz_61_zz_59_zz_57_zz_55_zz_53_zz_51_zz_49_zz_47_zz_45_zz_43_zz_41_zz_39_zz_37_zz_35_zz_33_zz_31_zz_29_zz_27_zz_25_zz_23_zz_21_zz_19_zz_17_zz_15_zz_13_zz_11_zz_9_zz_7_zz_5_zz_3_zz_1_zz:
 ; AVX512VBMI:       # BB#0:
 ; AVX512VBMI-NEXT:    vpxor %xmm1, %xmm1, %xmm1
 ; AVX512VBMI-NEXT:    vmovdqa64 {{.*#+}} zmm2 = [63,65,61,67,59,69,57,71,55,73,53,75,51,77,49,79,47,81,45,83,43,85,41,87,39,89,37,91,35,93,33,95,31,97,29,99,27,101,25,103,23,105,21,107,19,109,17,111,15,113,13,115,11,117,9,119,7,121,5,123,3,125,1,127]
 ; AVX512VBMI-NEXT:    vpermt2b %zmm1, %zmm2, %zmm0
 ; AVX512VBMI-NEXT:    retq
   %shuffle = shufflevector <64 x i8> %a, <64 x i8> zeroinitializer, <64 x i32> <i32 63, i32 64, i32 61, i32 64, i32 59, i32 64, i32 57, i32 64, i32 55, i32 64, i32 53, i32 64, i32 51, i32 64, i32 49, i32 64, i32 47, i32 64, i32 45, i32 64, i32 43, i32 64, i32 41, i32 64, i32 39, i32 64, i32 37, i32 64, i32 35, i32 64, i32 33, i32 64, i32 31, i32 64, i32 29, i32 64, i32 27, i32 64, i32 25, i32 64, i32 23, i32 64, i32 21, i32 64, i32 19, i32 64, i32 17, i32 64, i32 15, i32 64, i32 13, i32 64, i32 11, i32 64, i32 9, i32 64, i32 7, i32 64, i32 5, i32 64, i32 3, i32 64, i32 1, i32 64>
   ret <64 x i8> %shuffle
 }
 
 define <64 x i8> @shuffle_v64i8_63_64_61_66_59_68_57_70_55_72_53_74_51_76_49_78_47_80_45_82_43_84_41_86_39_88_37_90_35_92_33_94_31_96_29_98_27_100_25_102_23_104_21_106_19_108_17_110_15_112_13_114_11_116_9_118_7_120_5_122_3_124_1_126(<64 x i8> %a, <64 x i8> %b) {
 ; AVX512F-LABEL: shuffle_v64i8_63_64_61_66_59_68_57_70_55_72_53_74_51_76_49_78_47_80_45_82_43_84_41_86_39_88_37_90_35_92_33_94_31_96_29_98_27_100_25_102_23_104_21_106_19_108_17_110_15_112_13_114_11_116_9_118_7_120_5_122_3_124_1_126:
 ; AVX512F:       # BB#0:
 ; AVX512F-NEXT:    vpbroadcastw {{.*#+}} ymm4 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX512F-NEXT:    vpblendvb %ymm4, %ymm2, %ymm1, %ymm1
-; AVX512F-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
+; AVX512F-NEXT:    vpermq {{.*#+}} ymm2 = ymm1[2,3,0,1]
 ; AVX512F-NEXT:    vpblendvb %ymm4, %ymm1, %ymm2, %ymm1
 ; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm5 = [15,0,13,2,11,4,9,6,7,8,5,10,3,12,1,14,15,0,13,2,11,4,9,6,7,8,5,10,3,12,1,14]
 ; AVX512F-NEXT:    vpshufb %ymm5, %ymm1, %ymm2
 ; AVX512F-NEXT:    vpblendvb %ymm4, %ymm3, %ymm0, %ymm0
-; AVX512F-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512F-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512F-NEXT:    vpblendvb %ymm4, %ymm0, %ymm1, %ymm0
 ; AVX512F-NEXT:    vpshufb %ymm5, %ymm0, %ymm1
 ; AVX512F-NEXT:    vmovdqa %ymm2, %ymm0
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512BW-LABEL: shuffle_v64i8_63_64_61_66_59_68_57_70_55_72_53_74_51_76_49_78_47_80_45_82_43_84_41_86_39_88_37_90_35_92_33_94_31_96_29_98_27_100_25_102_23_104_21_106_19_108_17_110_15_112_13_114_11_116_9_118_7_120_5_122_3_124_1_126:
 ; AVX512BW:       # BB#0:
 ; AVX512BW-NEXT:    vextracti64x4 $1, %zmm1, %ymm2
 ; AVX512BW-NEXT:    vpbroadcastw {{.*#+}} ymm3 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX512BW-NEXT:    vpblendvb %ymm3, %ymm2, %ymm0, %ymm2
-; AVX512BW-NEXT:    vperm2i128 {{.*#+}} ymm4 = ymm2[2,3,0,1]
+; AVX512BW-NEXT:    vpermq {{.*#+}} ymm4 = ymm2[2,3,0,1]
 ; AVX512BW-NEXT:    vpblendvb %ymm3, %ymm2, %ymm4, %ymm2
 ; AVX512BW-NEXT:    vmovdqa {{.*#+}} ymm4 = [15,0,13,2,11,4,9,6,7,8,5,10,3,12,1,14,15,0,13,2,11,4,9,6,7,8,5,10,3,12,1,14]
 ; AVX512BW-NEXT:    vpshufb %ymm4, %ymm2, %ymm2
 ; AVX512BW-NEXT:    vextracti64x4 $1, %zmm0, %ymm0
 ; AVX512BW-NEXT:    vpblendvb %ymm3, %ymm1, %ymm0, %ymm0
-; AVX512BW-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512BW-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512BW-NEXT:    vpblendvb %ymm3, %ymm0, %ymm1, %ymm0
 ; AVX512BW-NEXT:    vpshufb %ymm4, %ymm0, %ymm0
 ; AVX512BW-NEXT:    vinserti64x4 $1, %ymm2, %zmm0, %zmm0
 ; AVX512BW-NEXT:    retq
 ;
 ; AVX512DQ-LABEL: shuffle_v64i8_63_64_61_66_59_68_57_70_55_72_53_74_51_76_49_78_47_80_45_82_43_84_41_86_39_88_37_90_35_92_33_94_31_96_29_98_27_100_25_102_23_104_21_106_19_108_17_110_15_112_13_114_11_116_9_118_7_120_5_122_3_124_1_126:
 ; AVX512DQ:       # BB#0:
 ; AVX512DQ-NEXT:    vpbroadcastw {{.*#+}} ymm4 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255]
 ; AVX512DQ-NEXT:    vpblendvb %ymm4, %ymm2, %ymm1, %ymm1
-; AVX512DQ-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
+; AVX512DQ-NEXT:    vpermq {{.*#+}} ymm2 = ymm1[2,3,0,1]
 ; AVX512DQ-NEXT:    vpblendvb %ymm4, %ymm1, %ymm2, %ymm1
 ; AVX512DQ-NEXT:    vmovdqa {{.*#+}} ymm5 = [15,0,13,2,11,4,9,6,7,8,5,10,3,12,1,14,15,0,13,2,11,4,9,6,7,8,5,10,3,12,1,14]
 ; AVX512DQ-NEXT:    vpshufb %ymm5, %ymm1, %ymm2
 ; AVX512DQ-NEXT:    vpblendvb %ymm4, %ymm3, %ymm0, %ymm0
-; AVX512DQ-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
+; AVX512DQ-NEXT:    vpermq {{.*#+}} ymm1 = ymm0[2,3,0,1]
 ; AVX512DQ-NEXT:    vpblendvb %ymm4, %ymm0, %ymm1, %ymm0
 ; AVX512DQ-NEXT:    vpshufb %ymm5, %ymm0, %ymm1
 ; AVX512DQ-NEXT:    vmovdqa %ymm2, %ymm0
 ; AVX512DQ-NEXT:    retq
 ;
 ; AVX512VBMI-LABEL: shuffle_v64i8_63_64_61_66_59_68_57_70_55_72_53_74_51_76_49_78_47_80_45_82_43_84_41_86_39_88_37_90_35_92_33_94_31_96_29_98_27_100_25_102_23_104_21_106_19_108_17_110_15_112_13_114_11_116_9_118_7_120_5_122_3_124_1_126:
 ; AVX512VBMI:       # BB#0:
 ; AVX512VBMI-NEXT:    vmovdqa64 {{.*#+}} zmm2 = [63,64,61,66,59,68,57,70,55,72,53,74,51,76,49,78,47,80,45,82,43,84,41,86,39,88,37,90,35,92,33,94,31,96,29,98,27,100,25,102,23,104,21,106,19,108,17,110,15,112,13,114,11,116,9,118,7,120,5,122,3,124,1,126]
 ; AVX512VBMI-NEXT:    vpermt2b %zmm1, %zmm2, %zmm0
 ; AVX512VBMI-NEXT:    retq
   %shuffle = shufflevector <64 x i8> %a, <64 x i8> %b, <64 x i32> <i32 63, i32 64, i32 61, i32 66, i32 59, i32 68, i32 57, i32 70, i32 55, i32 72, i32 53, i32 74, i32 51, i32 76, i32 49, i32 78, i32 47, i32 80, i32 45, i32 82, i32 43, i32 84, i32 41, i32 86, i32 39, i32 88, i32 37, i32 90, i32 35, i32 92, i32 33, i32 94, i32 31, i32 96, i32 29, i32 98, i32 27, i32 100, i32 25, i32 102, i32 23, i32 104, i32 21, i32 106, i32 19, i32 108, i32 17, i32 110, i32 15, i32 112, i32 13, i32 114, i32 11, i32 116, i32 9, i32 118, i32 7, i32 120, i32 5, i32 122, i32 3, i32 124, i32 1, i32 126>
   ret <64 x i8> %shuffle
 }

llvm/trunk/test/CodeGen/X86/vector-shuffle-512-v8.ll

 ; AVX512F-32-NEXT:    retl
   %1 = shufflevector <8 x double> %a1, <8 x double> undef, <8 x i32> <i32 2, i32 3, i32 0, i32 1, i32 undef, i32 undef, i32 6, i32 7>
   ret <8 x double> %1
 }
 
 define <8 x double> @shuffle_v8f64_2301uuuu(<8 x double> %a0, <8 x double> %a1) {
 ; AVX512F-LABEL: shuffle_v8f64_2301uuuu:
 ; AVX512F:       # BB#0:
-; AVX512F-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm1[2,3,0,1]
+; AVX512F-NEXT:    vpermpd {{.*#+}} ymm0 = ymm1[2,3,0,1]
 ; AVX512F-NEXT:    retq
 ;
 ; AVX512F-32-LABEL: shuffle_v8f64_2301uuuu:
 ; AVX512F-32:       # BB#0:
-; AVX512F-32-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm1[2,3,0,1]
+; AVX512F-32-NEXT:    vpermpd {{.*#+}} ymm0 = ymm1[2,3,0,1]
 ; AVX512F-32-NEXT:    retl
   %1 = shufflevector <8 x double> %a1, <8 x double> undef, <8 x i32> <i32 2, i32 3, i32 0, i32 1, i32 undef, i32 undef, i32 undef, i32 undef>
   ret <8 x double> %1
 }
 
 define <8 x double> @shuffle_v8f64_uuu2301(<8 x double> %a0, <8 x double> %a1) {
 ; AVX512F-LABEL: shuffle_v8f64_uuu2301:
 ; AVX512F:       # BB#0:

llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-avx.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
-; RUN: llc < %s -mtriple=i686-unknown -mattr=+avx | FileCheck %s --check-prefix=X32 --check-prefix=X32-AVX
-; RUN: llc < %s -mtriple=i686-unknown -mattr=+avx2 | FileCheck %s --check-prefix=X32 --check-prefix=X32-AVX
+; RUN: llc < %s -mtriple=i686-unknown -mattr=+avx | FileCheck %s --check-prefix=X32 --check-prefix=X32-AVX --check-prefix=X32-AVX1
+; RUN: llc < %s -mtriple=i686-unknown -mattr=+avx2 | FileCheck %s --check-prefix=X32 --check-prefix=X32-AVX --check-prefix=X32-AVX2
 ; RUN: llc < %s -mtriple=i686-unknown -mattr=+avx512f | FileCheck %s --check-prefix=X32 --check-prefix=X32-AVX512
-; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx | FileCheck %s --check-prefix=X64 --check-prefix=X64-AVX
-; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx2 | FileCheck %s --check-prefix=X64 --check-prefix=X64-AVX
+; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx | FileCheck %s --check-prefix=X64 --check-prefix=X64-AVX --check-prefix=X64-AVX1
+; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx2 | FileCheck %s --check-prefix=X64 --check-prefix=X64-AVX --check-prefix=X64-AVX2
 ; RUN: llc < %s -mtriple=x86_64-unknown -mattr=+avx512f | FileCheck %s --check-prefix=X64 --check-prefix=X64-AVX512
 ;
 ; Combine tests involving AVX target shuffles
 
 declare <4 x float> @llvm.x86.avx.vpermil.ps(<4 x float>, i8)
 declare <8 x float> @llvm.x86.avx.vpermil.ps.256(<8 x float>, i8)
 declare <2 x double> @llvm.x86.avx.vpermil.pd(<2 x double>, i8)
 declare <4 x double> @llvm.x86.avx.vpermil.pd.256(<4 x double>, i8)

 ; X64-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[1,0,3,2,6,u,4,u]
 ; X64-NEXT:    retq
   %1 = tail call <8 x float> @llvm.x86.avx.vpermilvar.ps.256(<8 x float> %a0, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 0, i32 1, i32 2, i32 undef>)
   %2 = tail call <8 x float> @llvm.x86.avx.vpermilvar.ps.256(<8 x float>  %1, <8 x i32> <i32 2, i32 3, i32 0, i32 1, i32 2, i32 3, i32 0, i32 undef>)
   ret <8 x float> %2
 }
 
 define <8 x float> @combine_vpermilvar_vperm2f128_8f32(<8 x float> %a0) {
-; X32-LABEL: combine_vpermilvar_vperm2f128_8f32:
-; X32:       # BB#0:
-; X32-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
-; X32-NEXT:    retl
-;
-; X64-LABEL: combine_vpermilvar_vperm2f128_8f32:
-; X64:       # BB#0:
-; X64-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
-; X64-NEXT:    retq
+; X32-AVX1-LABEL: combine_vpermilvar_vperm2f128_8f32:
+; X32-AVX1:       # BB#0:
+; X32-AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; X32-AVX1-NEXT:    retl
+;
+; X32-AVX2-LABEL: combine_vpermilvar_vperm2f128_8f32:
+; X32-AVX2:       # BB#0:
+; X32-AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; X32-AVX2-NEXT:    retl
+;
+; X32-AVX512-LABEL: combine_vpermilvar_vperm2f128_8f32:
+; X32-AVX512:       # BB#0:
+; X32-AVX512-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; X32-AVX512-NEXT:    retl
+;
+; X64-AVX1-LABEL: combine_vpermilvar_vperm2f128_8f32:
+; X64-AVX1:       # BB#0:
+; X64-AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; X64-AVX1-NEXT:    retq
+;
+; X64-AVX2-LABEL: combine_vpermilvar_vperm2f128_8f32:
+; X64-AVX2:       # BB#0:
+; X64-AVX2-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; X64-AVX2-NEXT:    retq
+;
+; X64-AVX512-LABEL: combine_vpermilvar_vperm2f128_8f32:
+; X64-AVX512:       # BB#0:
+; X64-AVX512-NEXT:    vpermpd {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; X64-AVX512-NEXT:    retq
   %1 = tail call <8 x float> @llvm.x86.avx.vpermilvar.ps.256(<8 x float> %a0, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 3, i32 2, i32 1, i32 0>)
   %2 = shufflevector <8 x float> %1, <8 x float> undef, <8 x i32> <i32 4, i32 5, i32 6, i32 7, i32 0, i32 1, i32 2, i32 3>
   %3 = tail call <8 x float> @llvm.x86.avx.vpermilvar.ps.256(<8 x float>  %2, <8 x i32> <i32 3, i32 2, i32 1, i32 0, i32 3, i32 2, i32 1, i32 0>)
   ret <8 x float> %3
 }
 
 define <8 x float> @combine_vpermilvar_vperm2f128_zero_8f32(<8 x float> %a0) {
 ; X32-LABEL: combine_vpermilvar_vperm2f128_zero_8f32:

llvm/trunk/test/CodeGen/X86/vector-shuffle-combining-avx2.ll

; X64-NEXT:    retq
   %3 = bitcast <32 x i8> %2 to <4 x i64>
   ret <4 x i64> %3
 }
 
 define <4 x i64> @combine_pshufb_as_zext128(<32 x i8> %a0) {
 ; X32-LABEL: combine_pshufb_as_zext128:
 ; X32:       # BB#0:
 ; X32-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0]
-; X32-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; X32-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
 ; X32-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[15,14],zero,zero,zero,zero,zero,zero,ymm0[13,12],zero,zero,zero,zero,zero,zero,ymm0[31,30],zero,zero,zero,zero,zero,zero,ymm0[29,28],zero,zero,zero,zero,zero,zero
 ; X32-NEXT:    retl
 ;
 ; X64-LABEL: combine_pshufb_as_zext128:
 ; X64:       # BB#0:
 ; X64-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[15,14,13,12,11,10,9,8,7,6,5,4,3,2,1,0]
-; X64-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[0,1,0,1]
+; X64-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[0,1,0,1]
 ; X64-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[15,14],zero,zero,zero,zero,zero,zero,ymm0[13,12],zero,zero,zero,zero,zero,zero,ymm0[31,30],zero,zero,zero,zero,zero,zero,ymm0[29,28],zero,zero,zero,zero,zero,zero
 ; X64-NEXT:    retq
   %1 = shufflevector <32 x i8> %a0, <32 x i8> undef, <32 x i32> <i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0, i32 15, i32 14, i32 13, i32 12, i32 11, i32 10, i32 9, i32 8, i32 7, i32 6, i32 5, i32 4, i32 3, i32 2, i32 1, i32 0>
   %2 = call <32 x i8> @llvm.x86.avx2.pshuf.b(<32 x i8> %1, <32 x i8> <i8 15, i8 14, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 13, i8 12, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 15, i8 14, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 13, i8 12, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1, i8 -1>)
   %3 = bitcast <32 x i8> %2 to <4 x i64>
   ret <4 x i64> %3
 }
 

llvm/trunk/test/CodeGen/X86/vector-shuffle-combining.ll

 ; AVX1-NEXT:    vpermilps {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
 ; AVX1-NEXT:    vperm2f128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: combine_unneeded_subvector1:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    vpaddd {{.*}}(%rip), %ymm0, %ymm0
 ; AVX2-NEXT:    vpshufd {{.*#+}} ymm0 = ymm0[3,2,1,0,7,6,5,4]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm0 = ymm0[2,3,2,3]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,2,3]
 ; AVX2-NEXT:    retq
   %b = add <8 x i32> %a, <i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8>
   %c = shufflevector <8 x i32> %b, <8 x i32> undef, <8 x i32> <i32 7, i32 6, i32 5, i32 4, i32 7, i32 6, i32 5, i32 4>
   ret <8 x i32> %c
 }
 
 define <8 x i32> @combine_unneeded_subvector2(<8 x i32> %a, <8 x i32> %b) {
 ; SSE-LABEL: combine_unneeded_subvector2:

llvm/trunk/test/CodeGen/X86/vector-shuffle-v1.ll

; VL_BW_DQ-NEXT:    retq
   %b2 = icmp eq <16 x i32> %b, %b1
   %c = shufflevector <16 x i1> %a2, <16 x i1> %b2, <16 x i32> <i32 3, i32 6, i32 22, i32 12, i32 3, i32 7, i32 7, i32 0, i32 3, i32 6, i32 1, i32 13, i32 3, i32 21, i32 7, i32 0>
   ret <16 x i1> %c
 }
 
 define <32 x i1> @shuf32i1_3_6_22_12_3_7_7_0_3_6_1_13_3_21_7_0_3_6_22_12_3_7_7_0_3_6_1_13_3_21_7_0(<32 x i1> %a) {
 ; AVX512F-LABEL: shuf32i1_3_6_22_12_3_7_7_0_3_6_1_13_3_21_7_0_3_6_22_12_3_7_7_0_3_6_1_13_3_21_7_0:
 ; AVX512F:       # BB#0:
-; AVX512F-NEXT:    vperm2i128 {{.*#+}} ymm1 = ymm0[2,3,0,1]
-; AVX512F-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[u,u,6,u,u,u,u,u,u,u,u,u,u,5,u,u,19,22,u,28,19,23,23,16,19,22,17,29,19,u,23,16]
-; AVX512F-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[3,6,u,12,3,7,7,0,3,6,1,13,3,u,7,0,u,u,22,u,u,u,u,u,u,u,u,u,u,21,u,u]
+; AVX512F-NEXT:    vpshufb {{.*#+}} ymm1 = ymm0[3,6,u,12,3,7,7,0,3,6,1,13,3,u,7,0,u,u,22,u,u,u,u,u,u,u,u,u,u,21,u,u]
+; AVX512F-NEXT:    vpermq {{.*#+}} ymm0 = ymm0[2,3,0,1]
+; AVX512F-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[u,u,6,u,u,u,u,u,u,u,u,u,u,5,u,u,19,22,u,28,19,23,23,16,19,22,17,29,19,u,23,16]
 ; AVX512F-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,0,255,255,255,255,255,255,255,255,255,255,0,255,255,0,0,255,0,0,0,0,0,0,0,0,0,0,255,0,0]
-; AVX512F-NEXT:    vpblendvb %ymm2, %ymm0, %ymm1, %ymm0
+; AVX512F-NEXT:    vpblendvb %ymm2, %ymm1, %ymm0, %ymm0
 ; AVX512F-NEXT:    retq
 ;
 ; VL_BW_DQ-LABEL: shuf32i1_3_6_22_12_3_7_7_0_3_6_1_13_3_21_7_0_3_6_22_12_3_7_7_0_3_6_1_13_3_21_7_0:
 ; VL_BW_DQ:       # BB#0:
 ; VL_BW_DQ-NEXT:    vpsllw $7, %ymm0, %ymm0
 ; VL_BW_DQ-NEXT:    vpmovb2m %ymm0, %k0
 ; VL_BW_DQ-NEXT:    vpmovm2w %k0, %zmm0
 ; VL_BW_DQ-NEXT:    vmovdqa64 {{.*#+}} zmm1 = [3,6,22,12,3,7,7,0,3,6,1,13,3,21,7,0,3,6,22,12,3,7,7,0,3,6,1,13,3,21,7,0]

llvm/trunk/test/CodeGen/X86/vector-shuffle-v48.ll

 ; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
 ; RUN: llc -mtriple=x86_64-pc-linux  -mattr=+avx2 < %s | FileCheck %s
 define <32 x i8> @foo(<48 x i8>* %x0, <16 x i32> %x1, <16 x i32> %x2) {
 ; CHECK-LABEL: foo:
 ; CHECK:       # BB#0:
 ; CHECK-NEXT:    vmovdqu 32(%rdi), %xmm0
 ; CHECK-NEXT:    vmovdqu (%rdi), %ymm1
-; CHECK-NEXT:    vperm2i128 {{.*#+}} ymm2 = ymm1[2,3,0,1]
+; CHECK-NEXT:    vpermq {{.*#+}} ymm2 = ymm1[2,3,0,1]
 ; CHECK-NEXT:    vpshufb {{.*#+}} xmm2 = xmm2[u,u,u,u,u,u,u,u,u,u,u,0,2,3,5,6]
 ; CHECK-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[0,1,3,4,6,7,9,10,12,13,15,u,u,u,u,u,24,25,27,28,30,31,u,u,u,u,u,u,u,u,u,u]
 ; CHECK-NEXT:    vmovdqa {{.*#+}} ymm3 = <255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,255,255,255,255,255,255,u,u,u,u,u,u,u,u,u,u>
 ; CHECK-NEXT:    vpblendvb %ymm3, %ymm1, %ymm2, %ymm1
 ; CHECK-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[u,u,u,u,u,u,1,2,4,5,7,8,10,11,13,14]
 ; CHECK-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm0
 ; CHECK-NEXT:    vmovdqa {{.*#+}} ymm2 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0]
 ; CHECK-NEXT:    vpblendvb %ymm2, %ymm1, %ymm0, %ymm0
 ; CHECK-NEXT:    retq
   %1 = load <48 x i8>, <48 x i8>* %x0, align 1
   %2 = shufflevector <48 x i8> %1, <48 x i8> undef, <32 x i32> <i32 0, i32 1, i32 3, i32 4, i32 6, i32 7, i32 9, i32 10, i32 12, i32 13, i32 15, i32 16, i32 18, i32 19, i32 21, i32 22, i32 24, i32 25, i32 27, i32 28, i32 30, i32 31, i32 33, i32 34, i32 36, i32 37, i32 39, i32 40, i32 42, i32 43, i32 45, i32 46>
   ret <32 x i8> %2
 }

llvm/trunk/test/CodeGen/X86/x86-interleaved-access.ll

 ; AVX2-NEXT:    vextracti128 $1, %ymm11, %xmm10
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = <1,5,9,13,u,u,u,u,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    vpshufb %xmm2, %xmm10, %xmm3
 ; AVX2-NEXT:    vpshufb %xmm2, %xmm11, %xmm0
 ; AVX2-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm0[0],xmm3[0],xmm0[1],xmm3[1]
 ; AVX2-NEXT:    vpblendd {{.*#+}} xmm12 = xmm0[0,1],xmm5[2,3]
 ; AVX2-NEXT:    vextracti128 $1, %ymm6, %xmm13
 ; AVX2-NEXT:    vpshufb %xmm7, %xmm13, %xmm3
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm6 = ymm6[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm6 = ymm6[2,3,0,1]
 ; AVX2-NEXT:    vextracti128 $1, %ymm6, %xmm14
 ; AVX2-NEXT:    vpshufb %xmm7, %xmm14, %xmm7
 ; AVX2-NEXT:    vpunpckldq {{.*#+}} xmm3 = xmm7[0],xmm3[0],xmm7[1],xmm3[1]
 ; AVX2-NEXT:    vinserti128 $1, %xmm3, %ymm0, %ymm3
 ; AVX2-NEXT:    vextracti128 $1, %ymm4, %xmm7
 ; AVX2-NEXT:    vpshufb %xmm2, %xmm7, %xmm0
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm4 = ymm4[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm4 = ymm4[2,3,0,1]
 ; AVX2-NEXT:    vextracti128 $1, %ymm4, %xmm4
 ; AVX2-NEXT:    vpshufb %xmm2, %xmm4, %xmm2
 ; AVX2-NEXT:    vpunpckldq {{.*#+}} xmm0 = xmm2[0],xmm0[0],xmm2[1],xmm0[1]
 ; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm0
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm0[0,1,2,3,4,5],ymm3[6,7]
 ; AVX2-NEXT:    vpblendd {{.*#+}} ymm0 = ymm12[0,1,2,3],ymm0[4,5,6,7]
 ; AVX2-NEXT:    vpcmpeqb %ymm0, %ymm8, %ymm8
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm0 = <u,u,u,u,2,6,10,14,u,u,u,u,u,u,u,u>

 ; AVX512-NEXT:    vmovdqa {{.*#+}} xmm2 = <1,5,9,13,u,u,u,u,u,u,u,u,u,u,u,u>
 ; AVX512-NEXT:    vpshufb %xmm2, %xmm10, %xmm5
 ; AVX512-NEXT:    vpshufb %xmm2, %xmm0, %xmm3
 ; AVX512-NEXT:    vpunpckldq {{.*#+}} xmm3 = xmm3[0],xmm5[0],xmm3[1],xmm5[1]
 ; AVX512-NEXT:    vpblendd {{.*#+}} xmm11 = xmm3[0,1],xmm4[2,3]
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm7, %ymm5
 ; AVX512-NEXT:    vextracti128 $1, %ymm5, %xmm12
 ; AVX512-NEXT:    vpshufb %xmm6, %xmm12, %xmm3
-; AVX512-NEXT:    vperm2i128 {{.*#+}} ymm5 = ymm5[2,3,0,1]
+; AVX512-NEXT:    vpermq {{.*#+}} ymm5 = ymm5[2,3,0,1]
 ; AVX512-NEXT:    vextracti128 $1, %ymm5, %xmm13
 ; AVX512-NEXT:    vpshufb %xmm6, %xmm13, %xmm6
 ; AVX512-NEXT:    vpunpckldq {{.*#+}} xmm3 = xmm6[0],xmm3[0],xmm6[1],xmm3[1]
 ; AVX512-NEXT:    vinserti128 $1, %xmm3, %ymm0, %ymm3
 ; AVX512-NEXT:    vextracti128 $1, %ymm7, %xmm14
 ; AVX512-NEXT:    vpshufb %xmm2, %xmm14, %xmm4
-; AVX512-NEXT:    vperm2i128 {{.*#+}} ymm7 = ymm7[2,3,0,1]
+; AVX512-NEXT:    vpermq {{.*#+}} ymm7 = ymm7[2,3,0,1]
 ; AVX512-NEXT:    vextracti128 $1, %ymm7, %xmm7
 ; AVX512-NEXT:    vpshufb %xmm2, %xmm7, %xmm2
 ; AVX512-NEXT:    vpunpckldq {{.*#+}} xmm2 = xmm2[0],xmm4[0],xmm2[1],xmm4[1]
 ; AVX512-NEXT:    vinserti128 $1, %xmm2, %ymm0, %ymm2
 ; AVX512-NEXT:    vpblendd {{.*#+}} ymm2 = ymm2[0,1,2,3,4,5],ymm3[6,7]
 ; AVX512-NEXT:    vpblendd {{.*#+}} ymm2 = ymm11[0,1,2,3],ymm2[4,5,6,7]
 ; AVX512-NEXT:    vpcmpeqb %ymm2, %ymm8, %ymm8
 ; AVX512-NEXT:    vmovdqa {{.*#+}} xmm2 = <u,u,u,u,2,6,10,14,u,u,u,u,u,u,u,u>

 ; AVX1-NEXT:    vzeroupper
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: interleaved_store_vf16_i8_stride3:
 ; AVX2:       # BB#0:
 ; AVX2-NEXT:    # kill: %XMM0<def> %XMM0<kill> %YMM0<def>
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm3
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm4 = ymm3[0,u,u,1,u,u,2,u,u,3,u,u,4,u,u,5,21,u,u,22,u,u,23,u,u,24,u,u,25,u,u,26]
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm3 = ymm3[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[2,3,0,1]
 ; AVX2-NEXT:    vpshufb {{.*#+}} ymm3 = ymm3[u,0,u,u,1,u,u,2,u,u,3,u,u,4,u,u,u,u,22,u,u,23,u,u,24,u,u,25,u,u,26,u]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm5 = <255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255>
 ; AVX2-NEXT:    vpblendvb %ymm5, %ymm4, %ymm3, %ymm3
 ; AVX2-NEXT:    vpshufb {{.*#+}} xmm4 = xmm2[0,1,0,1,0,1,6,7,2,3,2,3,4,5,4,5]
 ; AVX2-NEXT:    vpshuflw {{.*#+}} xmm5 = xmm2[2,1,3,3,4,5,6,7]
 ; AVX2-NEXT:    vpshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,4,4,4,7]
 ; AVX2-NEXT:    vinserti128 $1, %xmm5, %ymm4, %ymm4
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm5 = [255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255]
 ; AVX2-NEXT:    vpblendvb %ymm5, %ymm3, %ymm4, %ymm3
 ; AVX2-NEXT:    vpunpckhbw {{.*#+}} xmm0 = xmm0[8],xmm1[8],xmm0[9],xmm1[9],xmm0[10],xmm1[10],xmm0[11],xmm1[11],xmm0[12],xmm1[12],xmm0[13],xmm1[13],xmm0[14],xmm1[14],xmm0[15],xmm1[15]
 ; AVX2-NEXT:    vpshufb {{.*#+}} xmm0 = xmm0[u,6,7,u,8,9,u,10,11,u,12,13,u,14,15,u]
 ; AVX2-NEXT:    vpshufb {{.*#+}} xmm1 = xmm2[10,11,10,11,12,13,12,13,12,13,10,11,14,15,14,15]
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm2 = [0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0]
 ; AVX2-NEXT:    vpblendvb %xmm2, %xmm0, %xmm1, %xmm0
 ; AVX2-NEXT:    vmovdqu %xmm0, 32(%rdi)
 ; AVX2-NEXT:    vmovdqu %ymm3, (%rdi)
 ; AVX2-NEXT:    vzeroupper
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: interleaved_store_vf16_i8_stride3:
 ; AVX512:       # BB#0:
 ; AVX512-NEXT:    # kill: %XMM0<def> %XMM0<kill> %YMM0<def>
 ; AVX512-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm3
 ; AVX512-NEXT:    vpshufb {{.*#+}} ymm4 = ymm3[0,u,u,1,u,u,2,u,u,3,u,u,4,u,u,5,21,u,u,22,u,u,23,u,u,24,u,u,25,u,u,26]
-; AVX512-NEXT:    vperm2i128 {{.*#+}} ymm3 = ymm3[2,3,0,1]
+; AVX512-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[2,3,0,1]
 ; AVX512-NEXT:    vpshufb {{.*#+}} ymm3 = ymm3[u,0,u,u,1,u,u,2,u,u,3,u,u,4,u,u,u,u,22,u,u,23,u,u,24,u,u,25,u,u,26,u]
 ; AVX512-NEXT:    vmovdqa {{.*#+}} ymm5 = <255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255>
 ; AVX512-NEXT:    vpblendvb %ymm5, %ymm4, %ymm3, %ymm3
 ; AVX512-NEXT:    vpshufb {{.*#+}} xmm4 = xmm2[0,1,0,1,0,1,6,7,2,3,2,3,4,5,4,5]
 ; AVX512-NEXT:    vpshuflw {{.*#+}} xmm5 = xmm2[2,1,3,3,4,5,6,7]
 ; AVX512-NEXT:    vpshufhw {{.*#+}} xmm5 = xmm5[0,1,2,3,4,4,4,7]
 ; AVX512-NEXT:    vinserti128 $1, %xmm5, %ymm4, %ymm4
 ; AVX512-NEXT:    vmovdqa {{.*#+}} ymm5 = [255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255]

 ; AVX2-NEXT:    vmovdqu 160(%rdi), %ymm7
 ; AVX2-NEXT:    vmovdqu 128(%rdi), %ymm5
 ; AVX2-NEXT:    vmovdqu (%rdi), %ymm14
 ; AVX2-NEXT:    vmovdqu 32(%rdi), %ymm12
 ; AVX2-NEXT:    vmovdqu 64(%rdi), %ymm3
 ; AVX2-NEXT:    vmovdqu 96(%rdi), %ymm6
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = <255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0>
 ; AVX2-NEXT:    vpblendvb %ymm1, %ymm14, %ymm12, %ymm2
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm8 = ymm2[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm8 = ymm2[2,3,0,1]
 ; AVX2-NEXT:    vbroadcasti128 {{.*#+}} ymm9 = [255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255]
 ; AVX2-NEXT:    # ymm9 = mem[0,1,0,1]
 ; AVX2-NEXT:    vpblendvb %ymm9, %ymm2, %ymm8, %ymm2
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm10 = <0,3,6,9,12,15,2,5,8,11,14,1,4,7,10,13,0,3,6,9,12,15,u,u,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    vpshufb %ymm10, %ymm2, %ymm8
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm11 = <u,u,u,u,u,u,128,128,128,128,128,1,4,7,10,13>
 ; AVX2-NEXT:    vextracti128 $1, %ymm3, %xmm2
 ; AVX2-NEXT:    vpshufb %xmm11, %xmm2, %xmm4
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm13 = <u,u,u,u,u,u,2,5,8,11,14,128,128,128,128,128>
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm3 = ymm3[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm3 = ymm3[2,3,0,1]
 ; AVX2-NEXT:    vextracti128 $1, %ymm3, %xmm3
 ; AVX2-NEXT:    vpshufb %xmm13, %xmm3, %xmm0
 ; AVX2-NEXT:    vpor %xmm4, %xmm0, %xmm0
 ; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm0
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm15 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0]
 ; AVX2-NEXT:    vpblendvb %ymm15, %ymm8, %ymm0, %ymm0
 ; AVX2-NEXT:    vmovdqu %ymm0, -{{[0-9]+}}(%rsp) # 32-byte Spill
 ; AVX2-NEXT:    vpblendvb %ymm1, %ymm6, %ymm5, %ymm0
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm4 = ymm0[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm4 = ymm0[2,3,0,1]
 ; AVX2-NEXT:    vpblendvb %ymm9, %ymm0, %ymm4, %ymm0
 ; AVX2-NEXT:    vpshufb %ymm10, %ymm0, %ymm0
 ; AVX2-NEXT:    vextracti128 $1, %ymm7, %xmm10
 ; AVX2-NEXT:    vpshufb %xmm11, %xmm10, %xmm4
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm7 = ymm7[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm7 = ymm7[2,3,0,1]
 ; AVX2-NEXT:    vextracti128 $1, %ymm7, %xmm7
 ; AVX2-NEXT:    vpshufb %xmm13, %xmm7, %xmm1
 ; AVX2-NEXT:    vpor %xmm4, %xmm1, %xmm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm1, %ymm0, %ymm1
 ; AVX2-NEXT:    vpblendvb %ymm15, %ymm0, %ymm1, %ymm0
 ; AVX2-NEXT:    vmovdqu %ymm0, -{{[0-9]+}}(%rsp) # 32-byte Spill
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm13 = <u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255>
 ; AVX2-NEXT:    vpblendvb %ymm13, %ymm14, %ymm12, %ymm1
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm4 = ymm1[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm4 = ymm1[2,3,0,1]
 ; AVX2-NEXT:    vbroadcasti128 {{.*#+}} ymm11 = [0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0]
 ; AVX2-NEXT:    # ymm11 = mem[0,1,0,1]
 ; AVX2-NEXT:    vpblendvb %ymm11, %ymm1, %ymm4, %ymm1
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm8 = <u,u,u,u,u,128,128,128,128,128,128,2,5,8,11,14>
 ; AVX2-NEXT:    vpshufb %xmm8, %xmm2, %xmm0
 ; AVX2-NEXT:    vpblendvb %ymm13, %ymm6, %ymm5, %ymm13
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm15 = ymm13[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm15 = ymm13[2,3,0,1]
 ; AVX2-NEXT:    vpblendvb %ymm11, %ymm13, %ymm15, %ymm11
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm13 = <u,u,u,u,u,0,3,6,9,12,15,128,128,128,128,128>
 ; AVX2-NEXT:    vpshufb %xmm13, %xmm3, %xmm4
 ; AVX2-NEXT:    vpor %xmm0, %xmm4, %xmm0
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm4 = <1,4,7,10,13,0,3,6,9,12,15,2,5,8,11,14,1,4,7,10,13,u,u,u,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    vpshufb %ymm4, %ymm1, %ymm1
 ; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm0
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm15 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0]
 ; AVX2-NEXT:    vpblendvb %ymm15, %ymm1, %ymm0, %ymm9
 ; AVX2-NEXT:    vpshufb %ymm4, %ymm11, %ymm1
 ; AVX2-NEXT:    vpshufb %xmm8, %xmm10, %xmm4
 ; AVX2-NEXT:    vpshufb %xmm13, %xmm7, %xmm0
 ; AVX2-NEXT:    vpor %xmm4, %xmm0, %xmm0
 ; AVX2-NEXT:    vinserti128 $1, %xmm0, %ymm0, %ymm0
 ; AVX2-NEXT:    vpblendvb %ymm15, %ymm1, %ymm0, %ymm8
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm1 = <255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u>
 ; AVX2-NEXT:    vpblendvb %ymm1, %ymm5, %ymm6, %ymm4
 ; AVX2-NEXT:    vpblendvb %ymm1, %ymm12, %ymm14, %ymm1
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm5 = ymm4[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm5 = ymm4[2,3,0,1]
 ; AVX2-NEXT:    vbroadcasti128 {{.*#+}} ymm6 = [255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255]
 ; AVX2-NEXT:    # ymm6 = mem[0,1,0,1]
 ; AVX2-NEXT:    vpblendvb %ymm6, %ymm4, %ymm5, %ymm4
-; AVX2-NEXT:    vperm2i128 {{.*#+}} ymm5 = ymm1[2,3,0,1]
+; AVX2-NEXT:    vpermq {{.*#+}} ymm5 = ymm1[2,3,0,1]
 ; AVX2-NEXT:    vpblendvb %ymm6, %ymm1, %ymm5, %ymm1
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm5 = <u,u,u,u,u,128,128,128,128,128,0,3,6,9,12,15>
 ; AVX2-NEXT:    vpshufb %xmm5, %xmm10, %xmm6
 ; AVX2-NEXT:    vmovdqa {{.*#+}} xmm0 = <u,u,u,u,u,1,4,7,10,13,128,128,128,128,128,128>
 ; AVX2-NEXT:    vpshufb %xmm0, %xmm7, %xmm7
 ; AVX2-NEXT:    vpor %xmm7, %xmm6, %xmm6
 ; AVX2-NEXT:    vmovdqa {{.*#+}} ymm7 = <2,5,8,11,14,1,4,7,10,13,0,3,6,9,12,15,2,5,8,11,14,u,u,u,u,u,u,u,u,u,u,u>
 ; AVX2-NEXT:    vpshufb %ymm7, %ymm4, %ymm4

 ; AVX2-NEXT:    vpaddb -{{[0-9]+}}(%rsp), %ymm1, %ymm1 # 32-byte Folded Reload
 ; AVX2-NEXT:    vpaddb %ymm0, %ymm9, %ymm0
 ; AVX2-NEXT:    vpaddb -{{[0-9]+}}(%rsp), %ymm0, %ymm0 # 32-byte Folded Reload
 ; AVX2-NEXT:    retq
 ;
 ; AVX512-LABEL: interleaved_load_vf64_i8_stride3:
 ; AVX512:       # BB#0:
 ; AVX512-NEXT:    vmovdqu64 (%rdi), %zmm0
-; AVX512-NEXT:    vmovdqu64 64(%rdi), %zmm8
+; AVX512-NEXT:    vmovdqu64 64(%rdi), %zmm9
 ; AVX512-NEXT:    vmovdqu64 128(%rdi), %zmm1
 ; AVX512-NEXT:    vmovdqa {{.*#+}} ymm10 = <u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255>
 ; AVX512-NEXT:    vextracti64x4 $1, %zmm1, %ymm14
 ; AVX512-NEXT:    vpblendvb %ymm10, %ymm1, %ymm14, %ymm3
-; AVX512-NEXT:    vperm2i128 {{.*#+}} ymm4 = ymm3[2,3,0,1]
+; AVX512-NEXT:    vpermq {{.*#+}} ymm4 = ymm3[2,3,0,1]
 ; AVX512-NEXT:    vmovdqa {{.*#+}} ymm5 = <u,255,u,u,255,u,u,255,u,u,255,u,u,255,u,u,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255>
 ; AVX512-NEXT:    vpblendvb %ymm5, %ymm3, %ymm4, %ymm3
 ; AVX512-NEXT:    vpshufb {{.*#+}} ymm3 = ymm3[u,u,u,u,u,u,u,u,u,u,u,1,4,7,10,13,16,19,22,25,28,31,18,21,24,27,30,17,20,23,26,29]
 ; AVX512-NEXT:    vinserti64x4 $1, %ymm3, %zmm0, %zmm11
 ; AVX512-NEXT:    vmovdqa {{.*#+}} ymm2 = <255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0>
-; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm9
-; AVX512-NEXT:    vpblendvb %ymm2, %ymm0, %ymm9, %ymm4
-; AVX512-NEXT:    vperm2i128 {{.*#+}} ymm5 = ymm4[2,3,0,1]
+; AVX512-NEXT:    vextracti64x4 $1, %zmm0, %ymm8
+; AVX512-NEXT:    vpblendvb %ymm2, %ymm0, %ymm8, %ymm4
+; AVX512-NEXT:    vpermq {{.*#+}} ymm5 = ymm4[2,3,0,1]
 ; AVX512-NEXT:    vmovdqa {{.*#+}} ymm6 = <255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,u,u,255,u,u,255,u,u,255,u,u,255,u,u,255>
 ; AVX512-NEXT:    vpblendvb %ymm6, %ymm4, %ymm5, %ymm4
 ; AVX512-NEXT:    vpshufb {{.*#+}} ymm5 = ymm4[0,3,6,9,12,15,2,5,8,11,14,1,4,7,10,13,16,19,22,25,28,31,u,u,u,u,u,u,u,u,u,u]
-; AVX512-NEXT:    vperm2i128 {{.*#+}} ymm4 = ymm8[2,3,0,1]
-; AVX512-NEXT:    vextracti128 $1, %ymm4, %xmm12
-; AVX512-NEXT:    vpshufb {{.*#+}} xmm7 = xmm12[u,u,u,u,u,u,2,5,8,11,14],zero,zero,zero,zero,zero
-; AVX512-NEXT:    vextracti128 $1, %ymm8, %xmm13
-; AVX512-NEXT:    vpshufb {{.*#+}} xmm4 = xmm13[u,u,u,u,u,u],zero,zero,zero,zero,zero,xmm13[1,4,7,10,13]
-; AVX512-NEXT:    vpor %xmm4, %xmm7, %xmm4
+; AVX512-NEXT:    vextracti128 $1, %ymm9, %xmm12
+; AVX512-NEXT:    vpshufb {{.*#+}} xmm7 = xmm12[u,u,u,u,u,u],zero,zero,zero,zero,zero,xmm12[1,4,7,10,13]
+; AVX512-NEXT:    vpermq {{.*#+}} ymm6 = ymm9[2,3,0,1]
+; AVX512-NEXT:    vextracti128 $1, %ymm6, %xmm13
+; AVX512-NEXT:    vpshufb {{.*#+}} xmm4 = xmm13[u,u,u,u,u,u,2,5,8,11,14],zero,zero,zero,zero,zero
+; AVX512-NEXT:    vpor %xmm7, %xmm4, %xmm4
 ; AVX512-NEXT:    vinserti128 $1, %xmm4, %ymm0, %ymm4
 ; AVX512-NEXT:    vmovdqa {{.*#+}} ymm7 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0]
 ; AVX512-NEXT:    vpblendvb %ymm7, %ymm5, %ymm4, %ymm4
-; AVX512-NEXT:    vextracti64x4 $1, %zmm8, %ymm7
+; AVX512-NEXT:    vextracti64x4 $1, %zmm9, %ymm7
 ; AVX512-NEXT:    vextracti128 $1, %ymm7, %xmm5
 ; AVX512-NEXT:    vpshufb {{.*#+}} xmm6 = zero,zero,zero,zero,zero,zero,xmm5[2,5,8,11,14,u,u,u,u,u]
 ; AVX512-NEXT:    vpshufb {{.*#+}} xmm2 = xmm7[0,3,6,9,12,15],zero,zero,zero,zero,zero,xmm7[u,u,u,u,u]
 ; AVX512-NEXT:    vpor %xmm6, %xmm2, %xmm2
 ; AVX512-NEXT:    vinserti64x4 $1, %ymm2, %zmm4, %zmm2
 ; AVX512-NEXT:    movabsq $-8796093022208, %rax # imm = 0xFFFFF80000000000
 ; AVX512-NEXT:    kmovq %rax, %k1
 ; AVX512-NEXT:    vmovdqu8 %zmm11, %zmm2 {%k1}
 ; AVX512-NEXT:    vmovdqa {{.*#+}} ymm11 = <255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u,0,255,u>
 ; AVX512-NEXT:    vpblendvb %ymm11, %ymm14, %ymm1, %ymm4
-; AVX512-NEXT:    vperm2i128 {{.*#+}} ymm6 = ymm4[2,3,0,1]
+; AVX512-NEXT:    vpermq {{.*#+}} ymm6 = ymm4[2,3,0,1]
 ; AVX512-NEXT:    vmovdqa {{.*#+}} ymm15 = <u,u,255,u,u,255,u,u,255,u,u,255,u,u,255,u,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0>
 ; AVX512-NEXT:    vpblendvb %ymm15, %ymm4, %ymm6, %ymm4
 ; AVX512-NEXT:    vpshufb {{.*#+}} ymm4 = ymm4[u,u,u,u,u,u,u,u,u,u,u,2,5,8,11,14,17,20,23,26,29,16,19,22,25,28,31,18,21,24,27,30]
 ; AVX512-NEXT:    vinserti64x4 $1, %ymm4, %zmm0, %zmm16
-; AVX512-NEXT:    vpblendvb %ymm10, %ymm0, %ymm9, %ymm6
-; AVX512-NEXT:    vperm2i128 {{.*#+}} ymm10 = ymm6[2,3,0,1]
+; AVX512-NEXT:    vpblendvb %ymm10, %ymm0, %ymm8, %ymm6
+; AVX512-NEXT:    vpermq {{.*#+}} ymm10 = ymm6[2,3,0,1]
 ; AVX512-NEXT:    vmovdqa {{.*#+}} ymm15 = <0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,u,255,u,u,255,u,u,255,u,u,255,u,u,255,u,u>
 ; AVX512-NEXT:    vpblendvb %ymm15, %ymm6, %ymm10, %ymm6
 ; AVX512-NEXT:    vpshufb {{.*#+}} ymm6 = ymm6[1,4,7,10,13,0,3,6,9,12,15,2,5,8,11,14,17,20,23,26,29,u,u,u,u,u,u,u,u,u,u,u]
-; AVX512-NEXT:    vperm2i128 {{.*#+}} ymm10 = ymm8[2,3,0,1]
-; AVX512-NEXT:    vextracti128 $1, %ymm10, %xmm4
-; AVX512-NEXT:    vpshufb {{.*#+}} xmm4 = xmm4[u,u,u,u,u,0,3,6,9,12,15],zero,zero,zero,zero,zero
-; AVX512-NEXT:    vextracti128 $1, %ymm8, %xmm3
-; AVX512-NEXT:    vpshufb {{.*#+}} xmm3 = xmm3[u,u,u,u,u],zero,zero,zero,zero,zero,zero,xmm3[2,5,8,11,14]
-; AVX512-NEXT:    vpor %xmm3, %xmm4, %xmm3
+; AVX512-NEXT:    vextracti128 $1, %ymm9, %xmm4
+; AVX512-NEXT:    vpshufb {{.*#+}} xmm4 = xmm4[u,u,u,u,u],zero,zero,zero,zero,zero,zero,xmm4[2,5,8,11,14]
+; AVX512-NEXT:    vpermq {{.*#+}} ymm9 = ymm9[2,3,0,1]
+; AVX512-NEXT:    vextracti128 $1, %ymm9, %xmm3
+; AVX512-NEXT:    vpshufb {{.*#+}} xmm3 = xmm3[u,u,u,u,u,0,3,6,9,12,15],zero,zero,zero,zero,zero
+; AVX512-NEXT:    vpor %xmm4, %xmm3, %xmm3
 ; AVX512-NEXT:    vinserti128 $1, %xmm3, %ymm0, %ymm3
-; AVX512-NEXT:    vmovdqa {{.*#+}} ymm8 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0]
-; AVX512-NEXT:    vpblendvb %ymm8, %ymm6, %ymm3, %ymm3
+; AVX512-NEXT:    vmovdqa {{.*#+}} ymm9 = [255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,255,0,0,0,0,0,0,0,0,0,0,0]
+; AVX512-NEXT:    vpblendvb %ymm9, %ymm6, %ymm3, %ymm3
 ; AVX512-NEXT:    vpshufb {{.*#+}} xmm6 = xmm7[1,4,7,10,13],zero,zero,zero,zero,zero,zero,xmm7[u,u,u,u,u]
 ; AVX512-NEXT:    vpshufb {{.*#+}} xmm4 = zero,zero,zero,zero,zero,xmm5[0,3,6,9,12,15,u,u,u,u,u]
 ; AVX512-NEXT:    vpor %xmm6, %xmm4, %xmm4
 ; AVX512-NEXT:    vinserti64x4 $1, %ymm4, %zmm3, %zmm3
 ; AVX512-NEXT:    vmovdqu8 %zmm16, %zmm3 {%k1}
 ; AVX512-NEXT:    vmovdqa {{.*#+}} ymm4 = <255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0,u,255,0>
 ; AVX512-NEXT:    vpblendvb %ymm4, %ymm1, %ymm14, %ymm1
-; AVX512-NEXT:    vperm2i128 {{.*#+}} ymm4 = ymm1[2,3,0,1]
+; AVX512-NEXT:    vpermq {{.*#+}} ymm4 = ymm1[2,3,0,1]
 ; AVX512-NEXT:    vmovdqa {{.*#+}} ymm6 = <255,u,u,255,u,u,255,u,u,255,u,u,255,u,u,255,255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255>
 ; AVX512-NEXT:    vpblendvb %ymm6, %ymm1, %ymm4, %ymm1
 ; AVX512-NEXT:    vpshufb {{.*#+}} ymm1 = ymm1[u,u,u,u,u,u,u,u,u,u,0,3,6,9,12,15,18,21,24,27,30,17,20,23,26,29,16,19,22,25,28,31]
 ; AVX512-NEXT:    vinserti64x4 $1, %ymm1, %zmm0, %zmm1
-; AVX512-NEXT:    vpblendvb %ymm11, %ymm9, %ymm0, %ymm0
-; AVX512-NEXT:    vperm2i128 {{.*#+}} ymm4 = ymm0[2,3,0,1]
+; AVX512-NEXT:    vpblendvb %ymm11, %ymm8, %ymm0, %ymm0
+; AVX512-NEXT:    vpermq {{.*#+}} ymm4 = ymm0[2,3,0,1]
 ; AVX512-NEXT:    vmovdqa {{.*#+}} ymm6 = <255,0,255,255,0,255,255,0,255,255,0,255,255,0,255,255,u,u,255,u,u,255,u,u,255,u,u,255,u,u,255,u>
 ; AVX512-NEXT:    vpblendvb %ymm6, %ymm0, %ymm4, %ymm0
 ; AVX512-NEXT:    vpshufb {{.*#+}} ymm0 = ymm0[2,5,8,11,14,1,4,7,10,13,0,3,6,9,12,15,18,21,24,27,30,u,u,u,u,u,u,u,u,u,u,u]
-; AVX512-NEXT:    vpshufb {{.*#+}} xmm4 = xmm12[u,u,u,u,u,1,4,7,10,13],zero,zero,zero,zero,zero,zero
-; AVX512-NEXT:    vpshufb {{.*#+}} xmm6 = xmm13[u,u,u,u,u],zero,zero,zero,zero,zero,xmm13[0,3,6,9,12,15]
-; AVX512-NEXT:    vpor %xmm4, %xmm6, %xmm4
+; AVX512-NEXT:    vpshufb {{.*#+}} xmm4 = xmm12[u,u,u,u,u],zero,zero,zero,zero,zero,xmm12[0,3,6,9,12,15]
+; AVX512-NEXT:    vpshufb {{.*#+}} xmm6 = xmm13[u,u,u,u,u,1,4,7,10,13],zero,zero,zero,zero,zero,zero
+; AVX512-NEXT:    vpor %xmm6, %xmm4, %xmm4
 ; AVX512-NEXT:    vinserti128 $1, %xmm4, %ymm0, %ymm4
-; AVX512-NEXT:    vpblendvb %ymm8, %ymm0, %ymm4, %ymm0
+; AVX512-NEXT:    vpblendvb %ymm9, %ymm0, %ymm4, %ymm0
 ; AVX512-NEXT:    vpshufb {{.*#+}} xmm4 = zero,zero,zero,zero,zero,xmm5[1,4,7,10,13,u,u,u,u,u,u]
 ; AVX512-NEXT:    vpshufb {{.*#+}} xmm5 = xmm7[2,5,8,11,14],zero,zero,zero,zero,zero,xmm7[u,u,u,u,u,u]
 ; AVX512-NEXT:    vpor %xmm4, %xmm5, %xmm4
 ; AVX512-NEXT:    vinserti64x4 $1, %ymm4, %zmm0, %zmm0
 ; AVX512-NEXT:    movl $-2097152, %eax # imm = 0xFFE00000
 ; AVX512-NEXT:    kmovd %eax, %k1
 ; AVX512-NEXT:    vmovdqu16 %zmm1, %zmm0 {%k1}
 ; AVX512-NEXT:    vpaddb %zmm0, %zmm3, %zmm0