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

[iznogood]

Michael, on Bioshock, why are you showing the results before the optimisation and not after? The results you showed on your previous article had all cards performing above the 80 fps, which is not the case in this test

[pal666]

I think that day has already passed.

that day will never come because mesa is not going to get compat profile support

[eydee]

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...

This thing happens in pretty much all Source games. The issue may be in Valve's code.

[pal666]

what is the point of non free drivers?

paying customers

[pal666]

The question may become relevant in the future.

it can't, because you missed one feature

[pal666]

the only test which saddens me is bioshock. it greatly improved recently, but it needs more

[Azpegath]

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...

You should post a bug on this issue, with all the relevant information you've gathered. I think there are quite a few performance issues that never get reported.

[Adarion]

Looks nice, really nice. Though the factory OC'ed 470 comes often out on par with the 480. Also I guess the 460 occasionally suffers from being a 2 GB variant, at least it might on higher resolutions. But at all this is a very good progress, and one of the reasons why I support Team Red.

[Herem]

Michael the Bioshock results look like they may have an issue. Looking at the AMDGPU-PRO results the average frame rate of each card is almost exactly the same as the maximum, despite the minimum being in single digits in most cases. For instance the R9 Fury has a maximum frame rate of 82.38 which is only 0.27 fps higher than the average fps of 82.11, this is even though there is a minimum listed of 3.72 fps. The only card which does not have this behaviour is the RX460, which appears to me roughly half way between the min and max.

Xonotics is the only other benchmark which lists ave / min / max and in this test all cards have an average fps at roughly the mid point between the min and max fps. Obviously these are two different games, but it seems strange that the RX460 behaves in a similar manner in both games whereas all the other cards appear to have a fair degree of frame rate variance in Xonotics and almost 0 variance in Bioshock.

The Bioshock Mesa results are also inconsistent with the last set of results published in the BioShock Infinite Runs Much Faster For RadeonSI On Mesa Git: ~40% article. Both sets of results are listed as being run on 4.3 Mesa 12.1.0-devel- padoka PPA Gallium 0.4, however the New Mesa results appear to be approximately 20fps higher in the previous article.

Something else I noticed in the previous test results; the old Mesa graph shows the min / max to the left of each bar and the number to the right of each bar is clearly an average, whereas the New Mesa results only showed min. Is it possible the New Mesa results were displaying the maximum fps rather than the average fps? This would make the previous results more consistent with this article; however, it would also mean the New Mesa results from the previous article were only a small improvement, rather than the 40% performance improvement reported.

[dungeon]

Well previous test was on 16.10 and this one on 16.04... OS differences let say, alot improved and regressed in the meantime.