Nouveau Fermi, GeForce GTX 580 Linux Benchmarks

[phoronix]

Phoronix: Nouveau Fermi, GeForce GTX 580 Linux Benchmarks

And so it begins. Less than 24 hours after presenting OpenBenchmarking.org on the weekend, dozens of results are beginning to pour in (the latest test results). In particular, there's already a few sets off results that may interest many Phoronix readers...

http://www.phoronix.com/vr.php?view=OTE0Mw

[aliendude5300]

My Results

I have a fairly high-end system with an EVGA GTX 580 Superclocked edition, an AMD Phenom II X6 1100T, and 16GB of RAM. I uploaded several benchmarking results. For those of you who are interested, here are a few of them:

Graphics: http://openbenchmarking.org/result/1...IV-EVGANVIDI98
OpenCL: http://openbenchmarking.org/result/1...IV-EVGANVIDI89
Xonotic: http://openbenchmarking.org/result/1...IV-XONOTICEV52

I will continue to benchmark and upload my results. My system is liquid cooled using a Corsair H50 closed-loop system, I have a Corsair AX1200 power supply and an MSI 890FXA-GD70 motherboard.

[aliendude5300]

Wow... shortly after posting my reply in this thread, I noticed the "Update" section that linked to two of my results. Thanks for that! I honestly didn't expect that at all... any requests?

[cl333r]

The question is why?

Why is G3D so fast with Fermi and so slow with AMD cards while the latter has a lot of open specs? And nvidia doesn't - it's all fishy..
Not pessimist, but it looks like these benchmarks are misleading.. how about more games besides Urban T. 2004, how about tessellation.. how bout 2D speed..

[ChrisXY]

EVGA GTX 580 Superclocked edition, an AMD Phenom II X6 1100T, and 16GB of RAM.
Graphics: http://openbenchmarking.org/result/1...IV-EVGANVIDI98

Tremulous: 28 FPS? Srsly?

[aliendude5300]

Tremulous: 28 FPS? Srsly?

I'm not sure what happened on that one. I'll reinstall the tremulous test, and run a tremulous specific benchmark to see if I get a better result. That fixed the world of padman test that failed previously. It worked fine after reinstalling it, when I ran the "mesa" test suite: http://openbenchmarking.org/result/1...IV-EVGAGEFOR48

[monraaf]

The question is why?

Why is G3D so fast with Fermi and so slow with AMD cards while the latter has a lot of open specs? And nvidia doesn't - it's all fishy..
Not pessimist, but it looks like these benchmarks are misleading.. how about more games besides Urban T. 2004, how about tessellation.. how bout 2D speed..

Indeed, if you look at this comparison

http://openbenchmarking.org/result/1102278-CALI-OANVIDI95,1102272-CALI-OANVC0164,1101019-IV-RADEON22575&compare

It seems opensource fermi can already reach of up to over 70 % of the blob performance, while opensource radeon 5770 is stuck below 20% of blob performance in this particular benchmark. So this makes my curious (and a bit jealous) myself.

[aliendude5300]

I'm not sure what happened on that one. I'll reinstall the tremulous test, and run a tremulous specific benchmark to see if I get a better result. That fixed the world of padman test that failed previously. It worked fine after reinstalling it, when I ran the "mesa" test suite: http://openbenchmarking.org/result/1...IV-EVGAGEFOR48

Here's what happened: The tremulous test starts, showing the loading screen, but as soon as the game is 100% loaded, the test exits, reporting 28 FPS instead of saying that the test failed to load. This is reproducible every time. Reinstalling the tremulous test did not fix the issue.

[aliendude5300]

I just installed tremulous using 'sudo apt-get install tremulous'. The game runs, although the resolution is poor and the FPS seem to be 'stuck' at exactly 90 FPS in-game. This probably means that there is a bug in the phoronix test suite tremulous test, as the actual game loads just fine.

[ChrisXY]

It seems opensource fermi can already reach of up to over 70 % of the blob performance, while opensource radeon 5770 is stuck below 20% of blob performance in this particular benchmark. So this makes my curious (and a bit jealous) myself.

But it doesn't show if everything is rendered correctly.