OpenGL Threaded Optimizations Responsible For NVIDIA's Faster Performance?

[phoronix]

Phoronix: OpenGL Threaded Optimizations Responsible For NVIDIA's Faster Performance?

With the recent BioShock Infinite Linux benchmark results and the big Metro Redux graphics card comparison on Linux, a fair number of Linux gamers have been bringing it up in the forums with their hypothesis that NVIDIA's Linux gaming performance wins is due to their driver supporting OpenGL threaded optimizations. Well, that's not always the case, as shown in this article with a __GL_THREADED_OPTIMIZATIONS comparison in many Linux games with different GeForce hardware.

http://www.phoronix.com/vr.php?view=21568

[d2kx]

Very interesting. Actually I thought this would be the reason for the better performance vs. Catalyst aswell.

On a side note: fglrx 15.200 (Catalyst 15.3 Beta) only gave me a smaller FPS bump in Counter-Strike: Global Offensive, but it's much, much smoother now. There's no more stuttering and I did some tests and found that it was because of a much more efficient OpenGL binary shader cache. The stutterings I (and most/all other people) had (at least in CSGO, more likely in all demanding games) were due to slow shader compilation/caching (check ~/.AMD/GLCache/) and works much better with fglrx 15.200.

[dungeon]

Thanks for doing that comparison.

http://www.phoronix.com/forums/showthread.php?116073-BioShock-Infinite-Is-The-Latest-Game-Showing-Why-Linux-Gamers-Choose-NVIDIA&p=479464#post479464

Now waiting for Kano to comment, why other drivers needs to implement that hack When it might help only big chips, only few games, only few benchmarks or only few scenes... Even for those situations where it helps, people might even see drop in performance for some other particular scenes, etc...

[nightmarex]

Also Gw2 benefits from the threaded optimizations as well. It seems to be helpful in very CPU bound situations. As the test shows it really did help in BS:I.

I really don't know what's leading to the ~50fps cap for catalyst+BS:I (I should see if I can't break that in a bit) as other wine games can blitz past that with ease on lesser hardware not to mention the source games running at 300fps on catalyst. So something is screwy and I would love to find out what it is but honestly I am still betting on the CPU driver overhead being the base difference and the threaded making up the rest in this particular case but I could be wrong. Also anytime I see something sitting just under 60fps it makes me very suspicious that it is trying to frame sync or something similar (i.e. waiting for flips even if frame sync isn't set).

[dungeon]

I really don't know what's leading to the ~50fps cap for catalyst+BS:I

Not properly threaded game, eON wrapper, etc... Lowering down settings from Ultra, might help you to found an culprit. Some postprocessing effect might just eat whole bandwidth, too much bloom, too much particles, etc...

[Azrael5]

Mesa drivers

what about newst mesa drivers on ati & nvidia cards?

[Michael]

what about newst mesa drivers on ati & nvidia cards?

For what? They don't support BioShock Infinite / Metro Redux yet...

[Stellarwind]

I noticed threaded optimizations can actually reduce fps when I was trying to play Wildstar closed beta, so wasn't news to me. Game (with wine) was running like 20 fps without it and dropped to 7-8 fps with a setting enabled.

[przemoli]

I noticed threaded optimizations can actually reduce fps when I was trying to play Wildstar closed beta, so wasn't news to me. Game (with wine) was running like 20 fps without it and dropped to 7-8 fps with a setting enabled.

That was Mesa with or without CSTM?

[smitty3268]

That was Mesa with or without CSTM?

It couldn't be mesa if he was talking about the nvidia threaded optimizations - that's in the binary driver only.