[PATCH] D82660: [RISCV] Optimize multiplication by specific immediates
Sam Elliott via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Fri Jun 26 08:44:44 PDT 2020
lenary added a comment.
Thanks for the patch!
This optimisation is done by DAGCombine if you instead implement `decomposeMulByConstant` in `RISCVTargetLowering`. Read the comment on the `TargetLoweringBase` class to understand how to use it. This is preferrable, because we don't want to maintain a target-specific copy of this optimisation if we can avoid it.
It would be sensible to base your implementation on the one in the x86 backend: `X86TargetLowering::decomposeMulByConstant`, which deals with some phase ordering issues around legalisation.
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D82660/new/
https://reviews.llvm.org/D82660
More information about the llvm-commits
mailing list