AMD 2010 Catalyst Driver Year In Review

[phoronix]

Phoronix: AMD 2010 Catalyst Driver Year In Review

Earlier this month we delivered our annual performance look at NVIDIA's 2010 Linux graphics drivers and now the tables have turned to do our annual examination of the ATI/AMD Catalyst graphics drivers for the Radeon graphics processors. This was certainly an interesting year -- both good and bad -- for AMD with their Catalyst Linux driver.

http://www.phoronix.com/vr.php?view=15574

[Kano]

I don't think you benchmarked Heaven with tessellation as this was not possible with some drivers this year. Also the fps would be too high. I am not interested in OpenGL 3 benchmarks with an OpenGL 4 card, i usually compared DX11, OpenGL 4 Win/Linux with my 5670 with tesselation (normal) + 16x ani. Benchmarks where a medium range card reaches >60 fps is boring as the monitor would not need more anyway and many games usually do not render more fps when not in benchmark mode even without vsync.

[Azpegath]

Another request for the future, besides better video playback - would be Kernel Modesetting support! I now that it's been talked about over and over, but it is actually one of the things that makes me stick with the FOSS drivers. That, and the lack of support for newer Xorg-versions (or xorg-server that is). Since I'm using Gentoo with rolling releases, it's a bit of a pain to get support for something that Ubuntu hasn't started using yet.

[nevertell]

So eyefinity really does work ?

[Kano]

@Azpegath

kms maybe looks better when done right with a splash, but even with a u 11.04 alpah you see a garbled sceen after switching from plymouth to x. What's so much better there? Ok, when you only use vesa mode then it takes longer to switch but is that everything you do?

[Azpegath]

@Azpegath

kms maybe looks better when done right with a splash, but even with a u 11.04 alpah you see a garbled sceen after switching from plymouth to x. What's so much better there? Ok, when you only use vesa mode then it takes longer to switch but is that everything you do?

Currently it's more about the possibility of recovering from a kernel oops, since I can go back to the terminal (or it automatically does) and see the issue.
I don't have a garbled screen though, I just go from Grub to booting terminal, and then into X. I don't do "silent boots", and like the native high screen resolution of the terminal when using KMS.

I think I will have to re-evaluate my post though. The support for "the latest" xorg-server is probably more important for me than KMS.

[Kano]

Your system must be very unstable when you have got that many kernel oops. I usually reboot then, but it luckyly does not happen that often.

[Azpegath]

Your system must be very unstable when you have got that many kernel oops. I usually reboot then, but it luckyly does not happen that often.

Yes, sadly it is. The new 2.6.36 kernel is horrible, it actually crashes once every-other day. Part of them seem to be EXT4 issues (raw_spin etc) and some other seem related to sleep mode.
I've never had a kernel oops in many years, but now it's horrible.

[Apopas]

Gentoo here with kernel-2.6.36-r5 and ext4 fs as well.
I must say that it's rock solid and the only annoyance is some instability of Kwin because of the Intel drivers.

[Azpegath]

Gentoo here with kernel-2.6.36-r5 and ext4 fs as well.
I must say that it's rock solid and the only annoyance is some instability of Kwin because of the Intel drivers.

Weird.. Might be issues with the Radeon foss drivers as well then... But one of them it definitely EXT4-related. I saw a thread on the linux mailing list regarding it, and Alan Cox said it had been taken care of in .37, and that he should probably back port it into the .36 branch. I can't be sure it's the exact same issue though, but the backtrace seemed similar.

I probably have to go through my .config file to see if there's anything strange in it, I might have missed something when updating from .35.