[X86][AVX512] Use lowerShuffleAsRepeatedMaskAndLanePermute for non-VBMI v64i8 shuffles (PR31470)
llvm-svn: 291347
diff --git a/llvm/lib/Target/X86/X86ISelLowering.cpp b/llvm/lib/Target/X86/X86ISelLowering.cpp
index 7f72ab1..ad58f44 100644
--- a/llvm/lib/Target/X86/X86ISelLowering.cpp
+++ b/llvm/lib/Target/X86/X86ISelLowering.cpp
@@ -12965,6 +12965,12 @@
if (Subtarget.hasVBMI())
return lowerVectorShuffleWithPERMV(DL, MVT::v64i8, Mask, V1, V2, DAG);
+ // Try to create an in-lane repeating shuffle mask and then shuffle the
+ // the results into the target lanes.
+ if (SDValue V = lowerShuffleAsRepeatedMaskAndLanePermute(
+ DL, MVT::v64i8, V1, V2, Mask, Subtarget, DAG))
+ return V;
+
// FIXME: Implement direct support for this type!
return splitAndLowerVectorShuffle(DL, MVT::v64i8, V1, V2, Mask, DAG);
}
diff --git a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
index de48394..9167c09 100644
--- a/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
+++ b/llvm/lib/Target/X86/X86TargetTransformInfo.cpp
@@ -639,8 +639,7 @@
{ TTI::SK_Reverse, MVT::v32i16, 1 }, // vpermw
{ TTI::SK_Reverse, MVT::v16i16, 1 }, // vpermw
- { TTI::SK_Reverse, MVT::v64i8, 6 }, // vextracti64x4 + 2*vperm2i128
- // + 2*pshufb + vinserti64x4
+ { TTI::SK_Reverse, MVT::v64i8, 2 }, // pshufb + vshufi64x2
{ TTI::SK_PermuteSingleSrc, MVT::v32i16, 1 }, // vpermw
{ TTI::SK_PermuteSingleSrc, MVT::v16i16, 1 }, // vpermw