Closed R-Goc closed 2 months ago
I'd appreciate if you could make the PR, so that I can then test it. I tried PGO in Willow and it lost 30 elo, so apparently it's not very easy for me!
This should probably also get closed. I don't know if you tested this as a gainer or non-reg. PGO might make more sense when the code gets bigger, but the gain will never exceed a few percent.
Adding PGO seems like an easy way to increase the speed of the binary, and is very easy as the makefile only targets the clang compiler. I have a naive implementation written, should I make a PR, or do you have a plan for it already?