RADV+Zink vs. RadeonSI OpenGL Performance On Mesa 23.2-devel

[Linuxxx]

I'm not able to reproduce the Zink score with Heaven. For me, RadeonSI beats it consistently. Below is an FPS chart of every Heaven frame at 1080p.
GPU: Radeon 7600, RadeonSI: 166 FPS, Zink: 150 FPS

Are you running the benchmark with the "schedutil" CPU governor on your end?

If so, then that explains why Zink is beating RadeonSI here, because Michael runs all gaming benchmarks with the performance governor, as it should be done.

[marek]

Are you running the benchmark with the "schedutil" CPU governor on your end?

If so, then that explains why Zink is beating RadeonSI here, because Michael runs all gaming benchmarks with the performance governor, as it should be done.

My kernel isn't built with any other governor.

[Linuxxx]

My kernel isn't built with any other governor.

Why though?

Are you compiling your own kernel or is it a distro-provided one?

And yeah, that pretty much explains why RadeonSI is beating Zink on your end, whereas in Michael's benchmarks it's the other way around.

Schedutil is simply a really badly designed CPU governor, unfortunately...

If you can, please re-test with the performance governor on your end, too.

Thanks!

[marek]

Why though?

Are you compiling your own kernel or is it a distro-provided one?

And yeah, that pretty much explains why RadeonSI is beating Zink on your end, whereas in Michael's benchmarks it's the other way around.

Schedutil is simply a really badly designed CPU governor, unfortunately...

If you can, please re-test with the performance governor on your end, too.

Thanks!

I only build my kernel with "performance" and "schedutil" and the default one is set to "schedutil". I thought it was performance.