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

[X86][SSE] combineShuffleWithHorizOp - avoid unnecessary subtraction. NFCI. 

We can safely replace ((M - NumElts) % NumEltsPerLane) with (M % NumEltsPerLane) as the modulo result will be the same.
This commit is contained in:
Simon Pilgrim 2020-08-11 16:42:37 +01:00
parent c135744b1d
commit fe1f36986b
1 changed files with 1 additions and 1 deletions
llvm/lib/Target/X86
X86ISelLowering.cpp

2
llvm/lib/Target/X86/X86ISelLowering.cpp
View File

@ -35900,7 +35900,7 @@ static SDValue combineShuffleWithHorizOp(SDValue N, MVT VT, const SDLoc &DL,
(M % NumEltsPerLane) >= NumHalfEltsPerLane) (M % NumEltsPerLane) >= NumHalfEltsPerLane)
M -= NumHalfEltsPerLane; M -= NumHalfEltsPerLane;
if (NumElts <= M && BC1.getOperand(0) == BC1.getOperand(1) && if (NumElts <= M && BC1.getOperand(0) == BC1.getOperand(1) &&
((M - NumElts) % NumEltsPerLane) >= NumHalfEltsPerLane) (M % NumEltsPerLane) >= NumHalfEltsPerLane)
M -= NumHalfEltsPerLane; M -= NumHalfEltsPerLane;
} }
} }