[x86] Teach the new vector shuffle lowering to use AVX2 instructions for

v4f64 and v8f32 shuffles when they are lane-crossing. We have fully
general lane-crossing permutation functions in AVX2 that make this easy.

Part of this also changes exactly when and how these vectors are split
up when we don't have AVX2. This isn't always a win but it usually is
a win, so on the balance I think its better. The primary regressions are
all things that just need to be fixed anyways such as modeling when
a blend can be completely accomplished via VINSERTF128, etc.

Also, this highlights one of the few remaining big features: we do
a really poor job of inserting elements into AVX registers efficiently.

This completes almost all of the big tricks I have in mind for AVX2. The
only things left that I plan to add:

1) element insertion smarts
2) palignr and other fairly specialized lowerings when they happen to
   apply

git-svn-id: https://llvm.org/svn/llvm-project/llvm/trunk@218449 91177308-0d34-0410-b5e6-96231b3b80d8

diff --git a/lib/Target/X86/X86ISelLowering.cpp b/lib/Target/X86/X86ISelLowering.cpp
index 16ebe684eaa87c6c9325ebea7a87471d58c6aa93..44cd07b72d14d75f2830885df372431406dfffa5 100644 (file)
--- a/lib/Target/X86/X86ISelLowering.cpp
+++ b/lib/Target/X86/X86ISelLowering.cpp
@@ -9428,17 +9428,24 @@ static SDValue lowerV4F64VectorShuffle(SDValue Op, SDValue V1, SDValue V2,
   ArrayRef<int> Mask = SVOp->getMask();
   assert(Mask.size() == 4 && "Unexpected mask size for v4 shuffle!");
 
-  if (is128BitLaneCrossingShuffleMask(MVT::v4f64, Mask))
+  if (isSingleInputShuffleMask(Mask)) {
+    if (!is128BitLaneCrossingShuffleMask(MVT::v4f64, Mask)) {
+      // Non-half-crossing single input shuffles can be lowerid with an
+      // interleaved permutation.
+      unsigned VPERMILPMask = (Mask[0] == 1) | ((Mask[1] == 1) << 1) |
+                              ((Mask[2] == 3) << 2) | ((Mask[3] == 3) << 3);
+      return DAG.getNode(X86ISD::VPERMILPI, DL, MVT::v4f64, V1,
+                         DAG.getConstant(VPERMILPMask, MVT::i8));
+    }
+
+    // With AVX2 we have direct support for this permutation.
+    if (Subtarget->hasAVX2())
+      return DAG.getNode(X86ISD::VPERMI, DL, MVT::v4f64, V1,
+                         getV4X86ShuffleImm8ForMask(Mask, DAG));
+
+    // Otherwise, fall back.
     return lowerVectorShuffleAsLanePermuteAndBlend(DL, MVT::v4f64, V1, V2, Mask,
                                                    DAG);
-
-  if (isSingleInputShuffleMask(Mask)) {
-    // Non-half-crossing single input shuffles can be lowerid with an
-    // interleaved permutation.
-    unsigned VPERMILPMask = (Mask[0] == 1) | ((Mask[1] == 1) << 1) |
-                            ((Mask[2] == 3) << 2) | ((Mask[3] == 3) << 3);
-    return DAG.getNode(X86ISD::VPERMILPI, DL, MVT::v4f64, V1,
-                       DAG.getConstant(VPERMILPMask, MVT::i8));
   }
 
   // X86 has dedicated unpack instructions that can handle specific blend
@@ -9551,10 +9558,6 @@ static SDValue lowerV8F32VectorShuffle(SDValue Op, SDValue V1, SDValue V2,
   ArrayRef<int> Mask = SVOp->getMask();
   assert(Mask.size() == 8 && "Unexpected mask size for v8 shuffle!");
 
-  if (is128BitLaneCrossingShuffleMask(MVT::v8f32, Mask))
-    return lowerVectorShuffleAsLanePermuteAndBlend(DL, MVT::v8f32, V1, V2, Mask,
-                                                   DAG);
-
   if (SDValue Blend = lowerVectorShuffleAsBlend(DL, MVT::v8f32, V1, V2, Mask,
                                                 Subtarget, DAG))
     return Blend;
@@ -9589,9 +9592,21 @@ static SDValue lowerV8F32VectorShuffle(SDValue Op, SDValue V1, SDValue V2,
     for (int i = 0; i < 8; ++i)
       VPermMask[i] = Mask[i] < 0 ? DAG.getUNDEF(MVT::i32)
                                  : DAG.getConstant(Mask[i], MVT::i32);
-    return DAG.getNode(
-        X86ISD::VPERMILPV, DL, MVT::v8f32, V1,
-        DAG.getNode(ISD::BUILD_VECTOR, DL, MVT::v8i32, VPermMask));
+    if (!is128BitLaneCrossingShuffleMask(MVT::v8f32, Mask))
+      return DAG.getNode(
+          X86ISD::VPERMILPV, DL, MVT::v8f32, V1,
+          DAG.getNode(ISD::BUILD_VECTOR, DL, MVT::v8i32, VPermMask));
+
+    if (Subtarget->hasAVX2())
+      return DAG.getNode(X86ISD::VPERMV, DL, MVT::v8f32,
+                         DAG.getNode(ISD::BITCAST, DL, MVT::v8f32,
+                                     DAG.getNode(ISD::BUILD_VECTOR, DL,
+                                                 MVT::v8i32, VPermMask)),
+                         V1);
+
+    // Otherwise, fall back.
+    return lowerVectorShuffleAsLanePermuteAndBlend(DL, MVT::v8f32, V1, V2, Mask,
+                                                   DAG);
   }
 
   // Otherwise fall back on generic blend lowering.

diff --git a/test/CodeGen/X86/vector-shuffle-256-v4.ll b/test/CodeGen/X86/vector-shuffle-256-v4.ll
index cca593912fb0ad35be4d14122afb28b417a42766..423400c00a6c74d3b11f3e28c6a93a817d8559b5 100644 (file)
--- a/test/CodeGen/X86/vector-shuffle-256-v4.ll
+++ b/test/CodeGen/X86/vector-shuffle-256-v4.ll
@@ -4,77 +4,112 @@
 target triple = "x86_64-unknown-unknown"
 
 define <4 x double> @shuffle_v4f64_0001(<4 x double> %a, <4 x double> %b) {
-; ALL-LABEL: @shuffle_v4f64_0001
-; ALL:       # BB#0:
-; ALL-NEXT:    vunpcklpd {{.*}} # xmm1 = xmm0[0,0]
-; ALL-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v4f64_0001
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vunpcklpd {{.*}} # xmm1 = xmm0[0,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v4f64_0001
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpermpd {{.*}} # ymm0 = ymm0[0,0,0,1]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 0, i32 1>
   ret <4 x double> %shuffle
 }
 
 define <4 x double> @shuffle_v4f64_0020(<4 x double> %a, <4 x double> %b) {
-; ALL-LABEL: @shuffle_v4f64_0020
-; ALL:       # BB#0:
-; ALL-NEXT:    vextractf128 $1, %ymm0, %xmm1
-; ALL-NEXT:    vunpcklpd {{.*}} # xmm1 = xmm1[0],xmm0[0]
-; ALL-NEXT:    vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
-; ALL-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v4f64_0020
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm1
+; AVX1-NEXT:    vunpcklpd {{.*}} # xmm1 = xmm1[0],xmm0[0]
+; AVX1-NEXT:    vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v4f64_0020
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpermpd {{.*}} # ymm0 = ymm0[0,0,2,0]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 0, i32 2, i32 0>
   ret <4 x double> %shuffle
 }
 
 define <4 x double> @shuffle_v4f64_0300(<4 x double> %a, <4 x double> %b) {
-; ALL-LABEL: @shuffle_v4f64_0300
-; ALL:       # BB#0:
-; ALL-NEXT:    vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
-; ALL-NEXT:    vpermilpd {{.*}} # ymm1 = ymm1[0,1,2,2]
-; ALL-NEXT:    vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v4f64_0300
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
+; AVX1-NEXT:    vpermilpd {{.*}} # ymm1 = ymm1[0,1,2,2]
+; AVX1-NEXT:    vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v4f64_0300
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpermpd {{.*}} # ymm0 = ymm0[0,3,0,0]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 3, i32 0, i32 0>
   ret <4 x double> %shuffle
 }
 
 define <4 x double> @shuffle_v4f64_1000(<4 x double> %a, <4 x double> %b) {
-; ALL-LABEL: @shuffle_v4f64_1000
-; ALL:       # BB#0:
-; ALL-NEXT:    vpermilpd {{.*}} # xmm1 = xmm0[1,0]
-; ALL-NEXT:    vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
-; ALL-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v4f64_1000
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilpd {{.*}} # xmm1 = xmm0[1,0]
+; AVX1-NEXT:    vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v4f64_1000
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpermpd {{.*}} # ymm0 = ymm0[1,0,0,0]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 1, i32 0, i32 0, i32 0>
   ret <4 x double> %shuffle
 }
 
 define <4 x double> @shuffle_v4f64_2200(<4 x double> %a, <4 x double> %b) {
-; ALL-LABEL: @shuffle_v4f64_2200
-; ALL:       # BB#0:
-; ALL-NEXT:    vperm2f128 {{.*}} # ymm0 = ymm0[2,3,0,1]
-; ALL-NEXT:    vpermilpd {{.*}} # ymm0 = ymm0[0,0,2,2]
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v4f64_2200
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vperm2f128 {{.*}} # ymm0 = ymm0[2,3,0,1]
+; AVX1-NEXT:    vpermilpd {{.*}} # ymm0 = ymm0[0,0,2,2]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v4f64_2200
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpermpd {{.*}} # ymm0 = ymm0[2,2,0,0]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 2, i32 2, i32 0, i32 0>
   ret <4 x double> %shuffle
 }
 
 define <4 x double> @shuffle_v4f64_3330(<4 x double> %a, <4 x double> %b) {
-; ALL-LABEL: @shuffle_v4f64_3330
-; ALL:       # BB#0:
-; ALL-NEXT:    vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
-; ALL-NEXT:    vpermilpd {{.*}} # ymm1 = ymm1[1,1,2,2]
-; ALL-NEXT:    vpermilpd {{.*}} # ymm0 = ymm0[0,0,3,2]
-; ALL-NEXT:    vblendpd {{.*}} # ymm0 = ymm1[0,1],ymm0[2],ymm1[3]
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v4f64_3330
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
+; AVX1-NEXT:    vpermilpd {{.*}} # ymm1 = ymm1[1,1,2,2]
+; AVX1-NEXT:    vpermilpd {{.*}} # ymm0 = ymm0[0,0,3,2]
+; AVX1-NEXT:    vblendpd {{.*}} # ymm0 = ymm1[0,1],ymm0[2],ymm1[3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v4f64_3330
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpermpd {{.*}} # ymm0 = ymm0[3,3,3,0]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 3, i32 3, i32 0>
   ret <4 x double> %shuffle
 }
 
 define <4 x double> @shuffle_v4f64_3210(<4 x double> %a, <4 x double> %b) {
-; ALL-LABEL: @shuffle_v4f64_3210
-; ALL:       # BB#0:
-; ALL-NEXT:    vperm2f128 {{.*}} # ymm0 = ymm0[2,3,0,1]
-; ALL-NEXT:    vpermilpd {{.*}} # ymm0 = ymm0[1,0,3,2]
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v4f64_3210
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vperm2f128 {{.*}} # ymm0 = ymm0[2,3,0,1]
+; AVX1-NEXT:    vpermilpd {{.*}} # ymm0 = ymm0[1,0,3,2]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v4f64_3210
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpermpd {{.*}} # ymm0 = ymm0[3,2,1,0]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 3, i32 2, i32 1, i32 0>
   ret <4 x double> %shuffle
 }
@@ -218,19 +253,33 @@ define <4 x double> @shuffle_v4f64_4163(<4 x double> %a, <4 x double> %b) {
 }
 
 define <4 x double> @shuffle_v4f64_0145(<4 x double> %a, <4 x double> %b) {
-; ALL-LABEL: @shuffle_v4f64_0145
-; ALL:       # BB#0:
-; ALL-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v4f64_0145
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm1
+; AVX1-NEXT:    vblendpd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v4f64_0145
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpermpd {{.*}} # ymm1 = ymm1[0,1,0,1]
+; AVX2-NEXT:    vblendpd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 0, i32 1, i32 4, i32 5>
   ret <4 x double> %shuffle
 }
 
 define <4 x double> @shuffle_v4f64_4501(<4 x double> %a, <4 x double> %b) {
-; ALL-LABEL: @shuffle_v4f64_4501
-; ALL:       # BB#0:
-; ALL-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v4f64_4501
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; AVX1-NEXT:    vblendpd {{.*}} # ymm0 = ymm1[0,1],ymm0[2,3]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v4f64_4501
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpermpd {{.*}} # ymm0 = ymm0[0,1,0,1]
+; AVX2-NEXT:    vblendpd {{.*}} # ymm0 = ymm1[0,1],ymm0[2,3]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <4 x double> %a, <4 x double> %b, <4 x i32> <i32 4, i32 5, i32 0, i32 1>
   ret <4 x double> %shuffle
 }
@@ -374,10 +423,9 @@ define <4 x i64> @shuffle_v4i64_3210(<4 x i64> %a, <4 x i64> %b) {
 define <4 x i64> @shuffle_v4i64_0124(<4 x i64> %a, <4 x i64> %b) {
 ; AVX1-LABEL: @shuffle_v4i64_0124
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
 ; AVX1-NEXT:    vunpcklpd {{.*}} # xmm1 = xmm1[0,0]
-; AVX1-NEXT:    vblendpd {{.*}} # xmm1 = xmm2[0],xmm1[1]
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm1
+; AVX1-NEXT:    vblendpd {{.*}} # ymm0 = ymm0[0,1,2],ymm1[3]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: @shuffle_v4i64_0124
@@ -392,10 +440,9 @@ define <4 x i64> @shuffle_v4i64_0124(<4 x i64> %a, <4 x i64> %b) {
 define <4 x i64> @shuffle_v4i64_0142(<4 x i64> %a, <4 x i64> %b) {
 ; AVX1-LABEL: @shuffle_v4i64_0142
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
-; AVX1-NEXT:    vunpcklpd {{.*}} # xmm2 = xmm2[0,0]
-; AVX1-NEXT:    vblendpd {{.*}} # xmm1 = xmm1[0],xmm2[1]
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm1
+; AVX1-NEXT:    vpermilpd {{.*}} # ymm0 = ymm0[0,1,2,2]
+; AVX1-NEXT:    vblendpd {{.*}} # ymm0 = ymm0[0,1],ymm1[2],ymm0[3]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: @shuffle_v4i64_0142
@@ -413,9 +460,9 @@ define <4 x i64> @shuffle_v4i64_0412(<4 x i64> %a, <4 x i64> %b) {
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
 ; AVX1-NEXT:    vshufpd {{.*}} # xmm2 = xmm0[1],xmm2[0]
-; AVX1-NEXT:    vunpcklpd {{.*}} # xmm1 = xmm1[0,0]
-; AVX1-NEXT:    vblendpd {{.*}} # xmm0 = xmm0[0],xmm1[1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vpermilpd {{.*}} # ymm1 = ymm1[0,0,2,2]
+; AVX1-NEXT:    vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2,3]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: @shuffle_v4i64_0412
@@ -434,8 +481,8 @@ define <4 x i64> @shuffle_v4i64_4012(<4 x i64> %a, <4 x i64> %b) {
 ; AVX1-NEXT:    vextractf128 $1, %ymm0, %xmm2
 ; AVX1-NEXT:    vshufpd {{.*}} # xmm2 = xmm0[1],xmm2[0]
 ; AVX1-NEXT:    vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
-; AVX1-NEXT:    vblendpd {{.*}} # xmm0 = xmm1[0],xmm0[1]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vblendpd {{.*}} # ymm0 = ymm1[0],ymm0[1,2,3]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: @shuffle_v4i64_4012
@@ -450,7 +497,8 @@ define <4 x i64> @shuffle_v4i64_4012(<4 x i64> %a, <4 x i64> %b) {
 define <4 x i64> @shuffle_v4i64_0145(<4 x i64> %a, <4 x i64> %b) {
 ; AVX1-LABEL: @shuffle_v4i64_0145
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm1
+; AVX1-NEXT:    vblendpd {{.*}} # ymm0 = ymm0[0,1],ymm1[2,3]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: @shuffle_v4i64_0145
@@ -465,11 +513,11 @@ define <4 x i64> @shuffle_v4i64_0145(<4 x i64> %a, <4 x i64> %b) {
 define <4 x i64> @shuffle_v4i64_0451(<4 x i64> %a, <4 x i64> %b) {
 ; AVX1-LABEL: @shuffle_v4i64_0451
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vpermilpd {{.*}} # xmm2 = xmm1[1,0]
-; AVX1-NEXT:    vblendpd {{.*}} # xmm2 = xmm2[0],xmm0[1]
-; AVX1-NEXT:    vunpcklpd {{.*}} # xmm1 = xmm1[0,0]
-; AVX1-NEXT:    vblendpd {{.*}} # xmm0 = xmm0[0],xmm1[1]
-; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; AVX1-NEXT:    vunpcklpd {{.*}} # xmm2 = xmm1[0,0]
+; AVX1-NEXT:    vpermilpd {{.*}} # xmm1 = xmm1[1,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm1
+; AVX1-NEXT:    vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2],ymm0[3]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: @shuffle_v4i64_0451
@@ -485,7 +533,8 @@ define <4 x i64> @shuffle_v4i64_0451(<4 x i64> %a, <4 x i64> %b) {
 define <4 x i64> @shuffle_v4i64_4501(<4 x i64> %a, <4 x i64> %b) {
 ; AVX1-LABEL: @shuffle_v4i64_4501
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; AVX1-NEXT:    vblendpd {{.*}} # ymm0 = ymm1[0,1],ymm0[2,3]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: @shuffle_v4i64_4501
@@ -500,11 +549,11 @@ define <4 x i64> @shuffle_v4i64_4501(<4 x i64> %a, <4 x i64> %b) {
 define <4 x i64> @shuffle_v4i64_4015(<4 x i64> %a, <4 x i64> %b) {
 ; AVX1-LABEL: @shuffle_v4i64_4015
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vpermilpd {{.*}} # xmm2 = xmm0[1,0]
-; AVX1-NEXT:    vblendpd {{.*}} # xmm2 = xmm2[0],xmm1[1]
-; AVX1-NEXT:    vunpcklpd {{.*}} # xmm0 = xmm0[0,0]
-; AVX1-NEXT:    vblendpd {{.*}} # xmm0 = xmm1[0],xmm0[1]
-; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm1, %ymm1
+; AVX1-NEXT:    vunpcklpd {{.*}} # xmm2 = xmm0[0,0]
+; AVX1-NEXT:    vpermilpd {{.*}} # xmm0 = xmm0[1,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm2, %ymm0
+; AVX1-NEXT:    vblendpd {{.*}} # ymm0 = ymm1[0],ymm0[1,2],ymm1[3]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: @shuffle_v4i64_4015
@@ -526,10 +575,10 @@ define <4 x i64> @stress_test1(<4 x i64> %a, <4 x i64> %b) {
 ; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm1
 ; AVX1-NEXT:    vpermilpd {{.*}} # xmm1 = xmm1[1,0]
 ; AVX1-NEXT:    vpermilpd {{.*}} # ymm0 = ymm0[1,1,3,2]
-; AVX1-NEXT:    vpermilpd {{.*}} # xmm0 = xmm0[1,0]
-; AVX1-NEXT:    vblendpd {{.*}} # xmm0 = xmm0[0],xmm1[1]
-; AVX1-NEXT:    vpermilpd {{.*}} # xmm1 = xmm1[1,0]
-; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm0, %ymm0
+; AVX1-NEXT:    vpermilpd {{.*}} # xmm2 = xmm1[1,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm1, %ymm1
+; AVX1-NEXT:    vpermilpd {{.*}} # ymm0 = ymm0[1,0,2,2]
+; AVX1-NEXT:    vblendpd {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: @stress_test1

diff --git a/test/CodeGen/X86/vector-shuffle-256-v8.ll b/test/CodeGen/X86/vector-shuffle-256-v8.ll
index 38c75fe093cf86293cfe6bdd9cfcbc0a29f677a6..7e69ac5c537ea36208e1ca8e006234181df4ad75 100644 (file)
--- a/test/CodeGen/X86/vector-shuffle-256-v8.ll
+++ b/test/CodeGen/X86/vector-shuffle-256-v8.ll
@@ -4,92 +4,144 @@
 target triple = "x86_64-unknown-unknown"
 
 define <8 x float> @shuffle_v8f32_00000000(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: @shuffle_v8f32_00000000
-; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[0,0,0,0]
-; ALL-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v8f32_00000000
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[0,0,0,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v8f32_00000000
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vxorps %ymm1, %ymm1, %ymm1
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_00000010(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: @shuffle_v8f32_00000010
-; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # xmm1 = xmm0[0,0,0,0]
-; ALL-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[0,0,1,0]
-; ALL-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v8f32_00000010
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilps {{.*}} # xmm1 = xmm0[0,0,0,0]
+; AVX1-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[0,0,1,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v8f32_00000010
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovaps {{.*}} # ymm1 = [0,0,0,0,0,0,1,0]
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 1, i32 0>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_00000200(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: @shuffle_v8f32_00000200
-; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # xmm1 = xmm0[0,0,0,0]
-; ALL-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[0,2,0,0]
-; ALL-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v8f32_00000200
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilps {{.*}} # xmm1 = xmm0[0,0,0,0]
+; AVX1-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[0,2,0,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v8f32_00000200
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovaps {{.*}} # ymm1 = [0,0,0,0,0,2,0,0]
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 0, i32 2, i32 0, i32 0>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_00003000(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: @shuffle_v8f32_00003000
-; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # xmm1 = xmm0[0,0,0,0]
-; ALL-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[3,0,0,0]
-; ALL-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v8f32_00003000
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilps {{.*}} # xmm1 = xmm0[0,0,0,0]
+; AVX1-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[3,0,0,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v8f32_00003000
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovaps {{.*}} # ymm1 = [0,0,0,0,3,0,0,0]
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 3, i32 0, i32 0, i32 0>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_00040000(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: @shuffle_v8f32_00040000
-; ALL:       # BB#0:
-; ALL-NEXT:    vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm1[0,0,0,0,4,4,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,0,3,4,4,4,7]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2],ymm1[3,4,5,6,7]
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v8f32_00040000
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
+; AVX1-NEXT:    vpermilps {{.*}} # ymm1 = ymm1[0,0,0,0,4,4,4,4]
+; AVX1-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,0,3,4,4,4,7]
+; AVX1-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1,2],ymm1[3,4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v8f32_00040000
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovaps {{.*}} # ymm1 = [0,0,0,4,0,0,0,0]
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 4, i32 0, i32 0, i32 0, i32 0>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_00500000(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: @shuffle_v8f32_00500000
-; ALL:       # BB#0:
-; ALL-NEXT:    vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm1[u,u,1,u,4,4,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,0,4,4,6,4]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1],ymm1[2],ymm0[3],ymm1[4,5,6,7]
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v8f32_00500000
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
+; AVX1-NEXT:    vpermilps {{.*}} # ymm1 = ymm1[u,u,1,u,4,4,4,4]
+; AVX1-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,2,0,4,4,6,4]
+; AVX1-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0,1],ymm1[2],ymm0[3],ymm1[4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v8f32_00500000
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovaps {{.*}} # ymm1 = [0,0,5,0,0,0,0,0]
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 5, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_06000000(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: @shuffle_v8f32_06000000
-; ALL:       # BB#0:
-; ALL-NEXT:    vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm1[u,2,u,u,4,4,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,1,0,0,4,5,4,4]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2,3],ymm1[4,5,6,7]
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v8f32_06000000
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
+; AVX1-NEXT:    vpermilps {{.*}} # ymm1 = ymm1[u,2,u,u,4,4,4,4]
+; AVX1-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,1,0,0,4,5,4,4]
+; AVX1-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2,3],ymm1[4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v8f32_06000000
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovaps {{.*}} # ymm1 = [0,6,0,0,0,0,0,0]
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 6, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_70000000(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: @shuffle_v8f32_70000000
-; ALL:       # BB#0:
-; ALL-NEXT:    vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
-; ALL-NEXT:    vpermilps {{.*}} # ymm1 = ymm1[3,u,u,u,4,4,4,4]
-; ALL-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,0,0,4,4,4,4]
-; ALL-NEXT:    vblendps {{.*}} # ymm0 = ymm1[0],ymm0[1,2,3],ymm1[4,5,6,7]
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v8f32_70000000
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vperm2f128 {{.*}} # ymm1 = ymm0[2,3,0,1]
+; AVX1-NEXT:    vpermilps {{.*}} # ymm1 = ymm1[3,u,u,u,4,4,4,4]
+; AVX1-NEXT:    vpermilps {{.*}} # ymm0 = ymm0[0,0,0,0,4,4,4,4]
+; AVX1-NEXT:    vblendps {{.*}} # ymm0 = ymm1[0],ymm0[1,2,3],ymm1[4,5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v8f32_70000000
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vpxor %xmm1, %xmm1, %xmm1
+; AVX2-NEXT:    movl $7, %eax
+; AVX2-NEXT:    vpinsrd $0, %eax, %xmm1, %xmm1
+; AVX2-NEXT:    vpxor %ymm2, %ymm2, %ymm2
+; AVX2-NEXT:    vinserti128 $0, %xmm1, %ymm2, %ymm1
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 7, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0, i32 0>
   ret <8 x float> %shuffle
 }
@@ -104,23 +156,35 @@ define <8 x float> @shuffle_v8f32_01014545(<8 x float> %a, <8 x float> %b) {
 }
 
 define <8 x float> @shuffle_v8f32_00112233(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: @shuffle_v8f32_00112233
-; ALL:       # BB#0:
-; ALL-NEXT:    vunpcklps {{.*}} # xmm1 = xmm0[0,0,1,1]
-; ALL-NEXT:    vunpckhps {{.*}} # xmm0 = xmm0[2,2,3,3]
-; ALL-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v8f32_00112233
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vunpcklps {{.*}} # xmm1 = xmm0[0,0,1,1]
+; AVX1-NEXT:    vunpckhps {{.*}} # xmm0 = xmm0[2,2,3,3]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v8f32_00112233
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovaps {{.*}} # ymm1 = [0,0,1,1,2,2,3,3]
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 1, i32 1, i32 2, i32 2, i32 3, i32 3>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_00001111(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: @shuffle_v8f32_00001111
-; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # xmm1 = xmm0[0,0,0,0]
-; ALL-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[1,1,1,1]
-; ALL-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v8f32_00001111
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilps {{.*}} # xmm1 = xmm0[0,0,0,0]
+; AVX1-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[1,1,1,1]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm1, %ymm0
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v8f32_00001111
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovaps {{.*}} # ymm1 = [0,0,0,0,1,1,1,1]
+; AVX2-NEXT:    vpermps %ymm0, %ymm1, %ymm0
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 0, i32 0, i32 0, i32 1, i32 1, i32 1, i32 1>
   ret <8 x float> %shuffle
 }
@@ -135,13 +199,22 @@ define <8 x float> @shuffle_v8f32_81a3c5e7(<8 x float> %a, <8 x float> %b) {
 }
 
 define <8 x float> @shuffle_v8f32_08080808(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: @shuffle_v8f32_08080808
-; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # xmm1 = xmm1[0,0,2,0]
-; ALL-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[0,1,0,3]
-; ALL-NEXT:    vblendps {{.*}} # xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
-; ALL-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v8f32_08080808
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilps {{.*}} # xmm1 = xmm1[0,0,2,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm1, %ymm1
+; AVX1-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[0,1,0,3]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; AVX1-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v8f32_08080808
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vxorps %ymm2, %ymm2, %ymm2
+; AVX2-NEXT:    vpermps %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vpermps %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 8, i32 0, i32 8, i32 0, i32 8, i32 0, i32 8>
   ret <8 x float> %shuffle
 }
@@ -202,57 +275,85 @@ define <8 x float> @shuffle_v8f32_2a3b6e7f(<8 x float> %a, <8 x float> %b) {
 }
 
 define <8 x float> @shuffle_v8f32_08192a3b(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: @shuffle_v8f32_08192a3b
-; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # xmm2 = xmm1[0,2,2,3]
-; ALL-NEXT:    vpermilps {{.*}} # xmm3 = xmm0[2,1,3,3]
-; ALL-NEXT:    vblendps {{.*}} # xmm2 = xmm3[0],xmm2[1],xmm3[2],xmm2[3]
-; ALL-NEXT:    vpermilps {{.*}} # xmm1 = xmm1[0,0,2,1]
-; ALL-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[0,1,1,3]
-; ALL-NEXT:    vblendps {{.*}} # xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
-; ALL-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v8f32_08192a3b
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilps {{.*}} # xmm2 = xmm1[0,0,2,1]
+; AVX1-NEXT:    vpermilps {{.*}} # xmm1 = xmm1[0,2,2,3]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm1
+; AVX1-NEXT:    vpermilps {{.*}} # xmm2 = xmm0[0,1,1,3]
+; AVX1-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[2,1,3,3]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm2, %ymm0
+; AVX1-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v8f32_08192a3b
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovaps {{.*}} # ymm2 = <u,0,u,1,u,2,u,3>
+; AVX2-NEXT:    vpermps %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vmovaps {{.*}} # ymm2 = <0,u,1,u,2,u,3,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 8, i32 1, i32 9, i32 2, i32 10, i32 3, i32 11>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_08991abb(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: @shuffle_v8f32_08991abb
-; ALL:       # BB#0:
-; ALL-NEXT:    vpermilps {{.*}} # xmm2 = xmm0[1,1,2,3]
-; ALL-NEXT:    vpermilps {{.*}} # xmm3 = xmm1[0,2,3,3]
-; ALL-NEXT:    vblendps {{.*}} # xmm2 = xmm2[0],xmm3[1,2,3]
-; ALL-NEXT:    vunpcklps {{.*}} # xmm1 = xmm1[0,0,1,1]
-; ALL-NEXT:    vblendps {{.*}} # xmm0 = xmm0[0],xmm1[1,2,3]
-; ALL-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v8f32_08991abb
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilps {{.*}} # xmm2 = xmm0[1,1,2,3]
+; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vunpcklps {{.*}} # xmm2 = xmm1[0,0,1,1]
+; AVX1-NEXT:    vpermilps {{.*}} # xmm1 = xmm1[0,2,3,3]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm1
+; AVX1-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v8f32_08991abb
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovaps {{.*}} # ymm2 = <0,u,u,u,1,u,u,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vmovaps {{.*}} # ymm2 = <u,0,1,1,u,2,3,3>
+; AVX2-NEXT:    vpermps %ymm1, %ymm2, %ymm1
+; AVX2-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 8, i32 9, i32 9, i32 1, i32 10, i32 11, i32 11>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_091b2d3f(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: @shuffle_v8f32_091b2d3f
-; ALL:       # BB#0:
-; ALL-NEXT:    vextractf128 $1, %ymm1, %xmm2
-; ALL-NEXT:    vpermilps {{.*}} # xmm3 = xmm0[2,1,3,3]
-; ALL-NEXT:    vblendps {{.*}} # xmm2 = xmm3[0],xmm2[1],xmm3[2],xmm2[3]
-; ALL-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[0,1,1,3]
-; ALL-NEXT:    vblendps {{.*}} # xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
-; ALL-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v8f32_091b2d3f
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilps {{.*}} # xmm2 = xmm0[0,1,1,3]
+; AVX1-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[2,1,3,3]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm2, %ymm0
+; AVX1-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v8f32_091b2d3f
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovaps {{.*}} # ymm2 = <0,u,1,u,2,u,3,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 9, i32 1, i32 11, i32 2, i32 13, i32 3, i32 15>
   ret <8 x float> %shuffle
 }
 
 define <8 x float> @shuffle_v8f32_09ab1def(<8 x float> %a, <8 x float> %b) {
-; ALL-LABEL: @shuffle_v8f32_09ab1def
-; ALL:       # BB#0:
-; ALL-NEXT:    vextractf128 $1, %ymm1, %xmm2
-; ALL-NEXT:    vpermilps {{.*}} # xmm3 = xmm0[1,1,2,3]
-; ALL-NEXT:    vblendps {{.*}} # xmm2 = xmm3[0],xmm2[1,2,3]
-; ALL-NEXT:    vblendps {{.*}} # xmm0 = xmm0[0],xmm1[1,2,3]
-; ALL-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
-; ALL-NEXT:    retq
+; AVX1-LABEL: @shuffle_v8f32_09ab1def
+; AVX1:       # BB#0:
+; AVX1-NEXT:    vpermilps {{.*}} # xmm2 = xmm0[1,1,2,3]
+; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX1-NEXT:    retq
+;
+; AVX2-LABEL: @shuffle_v8f32_09ab1def
+; AVX2:       # BB#0:
+; AVX2-NEXT:    vmovaps {{.*}} # ymm2 = <0,u,u,u,1,u,u,u>
+; AVX2-NEXT:    vpermps %ymm0, %ymm2, %ymm0
+; AVX2-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
+; AVX2-NEXT:    retq
   %shuffle = shufflevector <8 x float> %a, <8 x float> %b, <8 x i32> <i32 0, i32 9, i32 10, i32 11, i32 1, i32 13, i32 14, i32 15>
   ret <8 x float> %shuffle
 }
@@ -754,9 +855,10 @@ define <8 x i32> @shuffle_v8i32_08080808(<8 x i32> %a, <8 x i32> %b) {
 ; AVX1-LABEL: @shuffle_v8i32_08080808
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpermilps {{.*}} # xmm1 = xmm1[0,0,2,0]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm1, %ymm1
 ; AVX1-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[0,1,0,3]
-; AVX1-NEXT:    vblendps {{.*}} # xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm0, %ymm0
+; AVX1-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: @shuffle_v8i32_08080808
@@ -865,13 +967,13 @@ define <8 x i32> @shuffle_v8i32_2a3b6e7f(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_08192a3b(<8 x i32> %a, <8 x i32> %b) {
 ; AVX1-LABEL: @shuffle_v8i32_08192a3b
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vpermilps {{.*}} # xmm2 = xmm1[0,2,2,3]
-; AVX1-NEXT:    vpermilps {{.*}} # xmm3 = xmm0[2,1,3,3]
-; AVX1-NEXT:    vblendps {{.*}} # xmm2 = xmm3[0],xmm2[1],xmm3[2],xmm2[3]
-; AVX1-NEXT:    vpermilps {{.*}} # xmm1 = xmm1[0,0,2,1]
-; AVX1-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[0,1,1,3]
-; AVX1-NEXT:    vblendps {{.*}} # xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
-; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vpermilps {{.*}} # xmm2 = xmm1[0,0,2,1]
+; AVX1-NEXT:    vpermilps {{.*}} # xmm1 = xmm1[0,2,2,3]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm1
+; AVX1-NEXT:    vpermilps {{.*}} # xmm2 = xmm0[0,1,1,3]
+; AVX1-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[2,1,3,3]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm2, %ymm0
+; AVX1-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: @shuffle_v8i32_08192a3b
@@ -890,11 +992,11 @@ define <8 x i32> @shuffle_v8i32_08991abb(<8 x i32> %a, <8 x i32> %b) {
 ; AVX1-LABEL: @shuffle_v8i32_08991abb
 ; AVX1:       # BB#0:
 ; AVX1-NEXT:    vpermilps {{.*}} # xmm2 = xmm0[1,1,2,3]
-; AVX1-NEXT:    vpermilps {{.*}} # xmm3 = xmm1[0,2,3,3]
-; AVX1-NEXT:    vblendps {{.*}} # xmm2 = xmm2[0],xmm3[1,2,3]
-; AVX1-NEXT:    vunpcklps {{.*}} # xmm1 = xmm1[0,0,1,1]
-; AVX1-NEXT:    vblendps {{.*}} # xmm0 = xmm0[0],xmm1[1,2,3]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vunpcklps {{.*}} # xmm2 = xmm1[0,0,1,1]
+; AVX1-NEXT:    vpermilps {{.*}} # xmm1 = xmm1[0,2,3,3]
+; AVX1-NEXT:    vinsertf128 $1, %xmm1, %ymm2, %ymm1
+; AVX1-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: @shuffle_v8i32_08991abb
@@ -912,12 +1014,10 @@ define <8 x i32> @shuffle_v8i32_08991abb(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_091b2d3f(<8 x i32> %a, <8 x i32> %b) {
 ; AVX1-LABEL: @shuffle_v8i32_091b2d3f
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
-; AVX1-NEXT:    vpermilps {{.*}} # xmm3 = xmm0[2,1,3,3]
-; AVX1-NEXT:    vblendps {{.*}} # xmm2 = xmm3[0],xmm2[1],xmm3[2],xmm2[3]
-; AVX1-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[0,1,1,3]
-; AVX1-NEXT:    vblendps {{.*}} # xmm0 = xmm0[0],xmm1[1],xmm0[2],xmm1[3]
-; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vpermilps {{.*}} # xmm2 = xmm0[0,1,1,3]
+; AVX1-NEXT:    vpermilps {{.*}} # xmm0 = xmm0[2,1,3,3]
+; AVX1-NEXT:    vinsertf128 $1, %xmm0, %ymm2, %ymm0
+; AVX1-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1],ymm0[2],ymm1[3],ymm0[4],ymm1[5],ymm0[6],ymm1[7]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: @shuffle_v8i32_091b2d3f
@@ -933,11 +1033,9 @@ define <8 x i32> @shuffle_v8i32_091b2d3f(<8 x i32> %a, <8 x i32> %b) {
 define <8 x i32> @shuffle_v8i32_09ab1def(<8 x i32> %a, <8 x i32> %b) {
 ; AVX1-LABEL: @shuffle_v8i32_09ab1def
 ; AVX1:       # BB#0:
-; AVX1-NEXT:    vextractf128 $1, %ymm1, %xmm2
-; AVX1-NEXT:    vpermilps {{.*}} # xmm3 = xmm0[1,1,2,3]
-; AVX1-NEXT:    vblendps {{.*}} # xmm2 = xmm3[0],xmm2[1,2,3]
-; AVX1-NEXT:    vblendps {{.*}} # xmm0 = xmm0[0],xmm1[1,2,3]
+; AVX1-NEXT:    vpermilps {{.*}} # xmm2 = xmm0[1,1,2,3]
 ; AVX1-NEXT:    vinsertf128 $1, %xmm2, %ymm0, %ymm0
+; AVX1-NEXT:    vblendps {{.*}} # ymm0 = ymm0[0],ymm1[1,2,3],ymm0[4],ymm1[5,6,7]
 ; AVX1-NEXT:    retq
 ;
 ; AVX2-LABEL: @shuffle_v8i32_09ab1def