Radeon Gallium3D: A Half-Decade Behind Catalyst?

[phoronix]

Phoronix: Radeon Gallium3D: A Half-Decade Behind Catalyst?

What happens when you pull out some vintage computer hardware and run the latest Linux software as well as go back and run some of the oldest software available? Well, in the case of systems with antiquated R300-era ATI Radeon graphics, you are left with a downward slope in performance. Not only is the latest open-source Radeon graphics driver not always performing as well as an ancient Catalyst driver, but also the power consumption of the latest Linux code remains on an incline.

http://www.phoronix.com/vr.php?view=16958

[fernandoc1]

Try making these same tests, on a desktop environment other than Unity, to compare newer versions of drivers with the older.
In my case, when I switched from Unity 3D to XFCE, I got a significant performance improvement when running Skyrim, through Wine.
I can say that the game that was unplayble on Unity, become playable on XFCE, something comparable to the Windows performance.
My system is:
AMD Phenon X4 2.4GHz, 4GB DDR2 800MHz and Nvidia GeForce 8600 GT (Driver 280.13).

[dsmithhfx]

My system is:
AMD Phenon X4 2.4GHz, 4GB DDR2 800MHz and Nvidia GeForce 8600 GT (Driver 280.13).

There is tremendous variation among older and newer ATI cards in terms of how they perform using free and binary ATI drivers... what makes you suppose your experience using Nvidia is in any way relevant?

[HokTar]

AFAIK Hyper-Z is not enabled by default on r300g. Marek wrote onto the mailing list that there are still some bugs he could not figure out, hence it is disabled.

[fernandoc1]

What I'm suggesting is that Unity is a heavy weight Desktop and it can compromise the performance of running applications.
If you see the chart on the benchmark article, you will find out that along side you are comparing different device drivers, there is also a difference in the desktop environment.
In my opinion it is not fair compare software running on different environments.

I'm not saying that it will work for Radeon hardware, but in my experience, I always found that one of the huge performance bottle neck in Ubuntu, was the Unity interface.
That is why I've switched to Xubuntu, in this last release cycle of Ubuntu.

[Reloaded211]

Too long, didn't read.

[Tgui]

Too long, didn't read.

With your obviously busy schedule, it's great you took the time to bless us with such an insightful post.

[Reloaded211]

With your obviously busy schedule, it's great you took the time to bless us with such an insightful post.

That was sarcasm. I actually did RTFA and wasn't mistaken - it has a huge label saying "OSS drivers suck" written all over it. I'm not saying that it's not true, but is it really necessary to repeat it again and again? Radeon/Noveau drivers will always be behind Catalyst/Nvidia because they are created by a handful of volunteers. We should appreciate that rather than complaining about it.

Catalyst driver may be faster, but it only works with certain kernel and X server releases and is rendered useless as soon as newer kernel/X server versions comes out. I actually got tired of trying to install Catalyst in Fedora, settled with Radeon, and that's it. The driver gets the job done, and that's all I need.

[Reloaded211]

These benchmarks should be done in single-user mode running applications in a dedicated X server. I'm running games myself that way using xgame script, makes them noticeably faster.

[marek]

r300g on AGP is known to be slow. I don't think I made any optimizations to make AGP any faster, as I don't have such a system and neither do I want one. When adding optimizations, I was mainly testing R500 with PCIe, where most of the user base is. Not sure if anybody else was trying to optimize for AGP...

Also, using Unity might have had a negative impact on the results.