[PATCH] D111646: [AMDGPU] Enable load clustering in the post-RA scheduler

[Piotr Sobczak via Phabricator via llvm-commits]

piotr added a comment.

Sounds good to me. The runtime improvement from clustering is notoriously difficult to assess, but your static data shows some potential benefit.

Having said that, it probably makes sense to guard the mutation by the optlevel check, so we only enable it with -O2 or higher.

  OptLevel > CodeGenOpt::Less


Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D111646/new/

https://reviews.llvm.org/D111646