maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity

[X86] lowerShuffleWithVPMOV - remove oneuse constraints on shuffle(trunc(x),undef) -> vpmov(x) lowering 

These were added in rG057bdd63 but shuffle combining has gotten a lot better at folding different vector widths since then.
This commit is contained in:
Simon Pilgrim 2022-08-11 14:04:09 +01:00
parent d09d4bd66c
commit 5dcf0c342b
1 changed files with 1 additions and 5 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
llvm/lib/Target/X86/X86ISelLowering.cpp
View File

@ -12412,11 +12412,7 @@ static SDValue lowerShuffleWithVPMOV(const SDLoc &DL, MVT VT, SDValue V1,
!Zeroable.extractBits(UpperElts, NumSrcElts).isAllOnes())
continue;
SDValue Src = V1;
if (!Src.hasOneUse())
return SDValue();
Src = peekThroughOneUseBitcasts(Src);
SDValue Src = peekThroughBitcasts(V1);
if (Src.getOpcode() != ISD::TRUNCATE ||
Src.getScalarValueSizeInBits() != (EltSizeInBits * Scale))
return SDValue();