AMDGPU-PRO vs. Open-Source Gallium3D OpenGL Performance On Polaris Is A Very Tight Race

[Nille_kungen]

I'm looking forward to the day when they're thinking they could probably just ditch Catalyst and make Mesa do their bidding.

I think that day has already passed.

[Azpegath]

This is a good sign, hopefully AMD will be able to put more and more effort into improving their Gallium drivers and unambiguously reach parity soon. I'm planning a new build in early 2017 and I'm seriously eying AMD since I'm fed up with proprietary drivers.

Have you done any Polaris Windows vs. Gallium/Pro benchmarks? That would be useful you're up for it.

I was thinking the same regarding Windows comparisons. And add the 290 (hopefully the regression is taken care of soon).

[Michael]

This is a good sign, hopefully AMD will be able to put more and more effort into improving their Gallium drivers and unambiguously reach parity soon. I'm planning a new build in early 2017 and I'm seriously eying AMD since I'm fed up with proprietary drivers.

Have you done any Polaris Windows vs. Gallium/Pro benchmarks? That would be useful you're up for it.

Yes, will be doing some next week or so now that I got these Linux numbers done, at least for tests where I have compatible windows test profile.

[bwyan]

I just tried duplicating these results using phoronix-test-suite. For some reason, the AMDGPU-PRO driver always performs much worse for me on my new AMD R9 Fury on Ubuntu 16.04 when comparing my results to the ones Michael are getting (around ~25-30 percent lower). Is there something that I may be missing that I should be aware of or is my GPU simply being bottlenecked by my i5-2500k CPU?

[dungeon]

Composition, governor... i can suggest to check as common diff . And to run some unigine benchmark, so if you still has there that much lower results then it is your CPU.

[bwyan]

Composition, governor... i can suggest to check as common diff . And to run some unigine benchmark, so if you still has there that much lower results then it is your CPU.

Thanks for the suggestion. I can't seem to find anything related to governor settings for the GPU though. Are You referring to the governor for the CPU?

[haagch]

Michael Can you make a frametime graph for csgo? Or maybe just try playing an arms race bot game or so? The average FPS seem okay, but ingame the FPS vary a lot and drop pretty low. For example: https://i.imgur.com/Jv6MF1n.png
Yes, that is a drop to 0 fps for ~0.5-1 second in there. They randomly happen. Makes it 100% unfit for playing competitively.

[Azpegath]

Michael Can you make a frametime graph for csgo? Or maybe just try playing an arms race bot game or so? The average FPS seem okay, but ingame the FPS vary a lot and drop pretty low. For example: https://i.imgur.com/Jv6MF1n.png
Yes, that is a drop to 0 fps for ~0.5-1 second in there. They randomly happen. Makes it 100% unfit for playing competitively.

I think it's caused by shader compiles. The same goes for Left 4 Dead and it's really annoying. If they do the shader cache, it will improve radically.

[haagch]

It's not. The HUD shows zero shaders created and zero compilations while it happens. But just now I noticed that the GTT usage drops a little bit, so I suspect something inefficient somewhere in the memory management. First thought is of course https://lists.freedesktop.org/archiv...st/001360.html but there can be no memory pressure on my 8 gigabyte rx 480...

[Michael]

Michael Can you make a frametime graph for csgo? Or maybe just try playing an arms race bot game or so? The average FPS seem okay, but ingame the FPS vary a lot and drop pretty low. For example: https://i.imgur.com/Jv6MF1n.png
Yes, that is a drop to 0 fps for ~0.5-1 second in there. They randomly happen. Makes it 100% unfit for playing competitively.

As far as I know, CS:GO doesn't expose the frame time information via a parseable way to PTS (e.g. output or file) so unless there are some options I am unaware of, not possible.