Searched refs:commute (Results 1 – 12 of 12) sorted by relevance
169 // can be commuted. In many cases such commute transformation requires an opcode263 // the commute optimization is legal only if all users of FMA*_Int use only265 // may be not implemented yet we allow the routines doing the actual commute
602 // commute. We also allow fdiv[r]p/fsubrp even though they don't commute,
6194 // Pattern to commute if load is in first source.6980 // We can commute a load in the first operand by swapping the sources and7240 // Folding "To" vector - convert to perm2x128 and commute inputs.7568 // Pattern to commute if load is in first source.
52 // Zero mask does not add any restrictions to commute operands transformation.
1504 void ShuffleVectorInst::commute() { in commute() function in llvm::sandboxir::ShuffleVectorInst1508 cast<llvm::ShuffleVectorInst>(Val)->commute(); in commute()
2336 Shuf.commute(); in foldSelectShuffle()2942 SVI.commute(); in visitShuffleVectorInst()
544 LLVM_ABI void commute();
1964 LLVM_ABI void commute();
1769 void ShuffleVectorInst::commute() { in commute() function in ShuffleVectorInst
696 // TODO src0 contains the opsel bit for dst, so if we commute, need to mask and swap this
40196 commute