[PATCH] D89880: [AMDGPU] Reorder SIMemoryLegalizer functions to be consistent
Nicolai Hähnle via Phabricator via llvm-commits
llvm-commits at lists.llvm.org
Thu Oct 22 11:51:40 PDT 2020
nhaehnle added a comment.
If we're nitpicking over such details, I would point out that it would be better to structure the whole thing differently, with a single legalizer class in which the `insertAcquire` etc. methods have different paths based on the hardware generation.
The reasoning is that code of the form "gfx10 behaves like gfx7 except ABC, gfx7 behaves like gfx6 except DEF" becomes increasingly difficult to reason about or refactor the more hardware generations we add.
Repository:
rG LLVM Github Monorepo
CHANGES SINCE LAST ACTION
https://reviews.llvm.org/D89880/new/
https://reviews.llvm.org/D89880
More information about the llvm-commits
mailing list