[llvm] [SystemZ] Remove getInliningThresholdMultiplier() override (PR #94612)
Jonas Paulsson via llvm-commits
llvm-commits at lists.llvm.org
Tue Jul 23 06:12:48 PDT 2024
JonPsson1 wrote:
> Do I remember correctly that the change in [27e8c50](https://github.com/llvm/llvm-project/commit/27e8c50a4c34c6124622c17202ffb5cc4d8d1ebd) was also targeted specifically at increasing the imagick SPEC score?
Unfortunately, it seems the old discussion on Phabricator is lost so I am not sure which benchmark(s?) that relates to. This is a case where we actually do know for a specific type of function that inlining is highly beneficial, so that is certainly good to keep. How are you thinking that this would relate to the general multiplier? I could do some more benchmarking without this also if you think it would be relevant.
https://github.com/llvm/llvm-project/pull/94612
More information about the llvm-commits
mailing list