Cross thinLTO might result in performance loss?

Alex313031 / Mercury

Firefox fork with compiler optimizations and patches from Librewolf, Waterfox, and GNU IceCat.

https://thorium.rocks/mercury

Mozilla Public License 2.0

1.15k stars 25 forks source link

Cross thinLTO might result in performance loss? #156

Open K4sum1 opened 6 months ago

K4sum1 commented 6 months ago

I haven't tested myself and I kinda introduced another variable into it, but I built my Firefox fork with it, and got an issue about performance being worse. I went back to old LTO and performance is right back where it was according to the guy.

https://github.com/Eclipse-Community/r3dfox/issues/22

Alex313031 commented 6 months ago

@K4sum1 Thanks! I'll revert it then.

K4sum1 commented 6 months ago

@Alex313031 Can you hop on Discord btw?

Alex313031 commented 6 months ago

@K4sum1 OK im there now.

Girofox commented 6 months ago

Does this affect Speedometer 3.0 (Javascript) and Stylebench (CSS) benchmarks? So far i got the same performance as in Chrome in Speedometer 3.0 with SSE4 and AVX2 variants.