[llvm] AMDGPU: VALU data fast-forwarding needs no s_delay (PR #205481)
Reem Elkhouly via llvm-commits
llvm-commits at lists.llvm.org
Tue Sep 29 01:52:12 PDT 2026
amd-relkhoul wrote:
> Do you think it would be possible to split this into separate PRs?
I think it is possible. Most of the complexity was introduced by the implicit uses and defs. If we leave that as the last part, we shouldn't see regressions.
> Some points raised by an AI review of this PR
I'll go through them and fix the code where necessary.
Thanks for taking the time, Jay!
https://github.com/llvm/llvm-project/pull/205481
More information about the llvm-commits
mailing list