forked from OSchip/llvm-project
parent
c732c61b6d
commit
07984f2068
|
@ -11139,7 +11139,7 @@ static SDValue lowerV4F64VectorShuffle(SDValue Op, SDValue V1, SDValue V2,
|
|||
// we will be able to shuffle even across lanes the other input in a single
|
||||
// instruction so skip this pattern.
|
||||
if (!(Subtarget.hasAVX2() && (isShuffleMaskInputInPlace(0, Mask) ||
|
||||
isShuffleMaskInputInPlace(1, Mask))))
|
||||
isShuffleMaskInputInPlace(1, Mask))))
|
||||
if (SDValue Result = lowerVectorShuffleByMerging128BitLanes(
|
||||
DL, MVT::v4f64, V1, V2, Mask, Subtarget, DAG))
|
||||
return Result;
|
||||
|
|
Loading…
Reference in New Issue