Forking this article for proprietary
Cleaned up some info today and noticed this article is really getting unruly. I propose that we fork the proprietary portion to its own article, such as "ATI Catalyst". Considering that the fglrx is now absolutely unsupported by Arch Linux, it would help make this article much less daunting to newcomers.
-- Thayer Mon Jan 4 10:53:18 PST 2010
- I wholly agree.
EXA and Adobe Flash performance
I noticed that enabling
Option "AccelMethod" "EXA"
as mentioned under "Performance tuning" lead to significantly worse Flash performance. This was using extra/flashplugin 10.0.22.87-1, aur/firefox-beta 3.5rc2-1 and extra/xf86-video-ati 6.12.2-2. My card (ATI Radeon 9550) uses the RV350 chipset, hence it is a bit dated.
If this isn't just a local phenomena, perhaps it should be mentioned?
I don't know. I disabled EXA, and got even worse Flash performance, according to a flash benchmark test I found at Snails Animation . (flashplugin 10.0.32.18-1.1, firefox 3.5.3-1, xf86-video-ati 6.12.2-2.) Same card and chipset, though. --Rataxes 16:38, 15 September 2009 (EDT)
KMS on r5xx not works with linux 18.104.22.168. only if libdrm, libgl, mesa, xf86-video-ati - all installed from git.
KMS seems to be enabled by default on r5xx with latest packages from arch. Should all those messages about it being unsupported be removed? TheCycoONE
If KMS is enabled, dri-driver (r600_dri.so) does not load. X switch to software rasterization (swrast_dri.so) if available. Using 2.6.34-zen1, xorg 22.214.171.1242, mesa and xf86-video-ati from git master. approve plz. --ReMiND 05:47, 27 June 2010 (EDT)
- Problem is gone after reading http://www.x.org/wiki/radeonBuildHowTo#Troubleshooting and including new radeon_ucode firmware R600_rlc.bin (provides with linux-firmware package) ReMiND 14:21, 10 July 2010 (EDT)
I don't know how others feel about this, but my preferred setup is to put options radeon modeset=1 in modprobe.conf and then to include modprobe.con along with the radeon module in mkinitcpio.conf. This way KMS is enabled once for both early and late start. Also, in the event of problems, radeon.modeset=1 can still be added as a kernel option to the bootloader. --piezoelectric
Driver ati or radeon
I just did the tutorial installing:
pacman -S xf86-video-ati libgl ati-dri
I could only get it working by changing in xorg.conf:
as the tutorial stated. I have no idea why/how, but maybe it'll help people reading it here.
--Overwave 16:02, 7 November 2010 (EST) The same happened for me, thanks for your tip.
radeon-initrd does not exist in AUR
Re: http://wiki.archlinux.org/index.php/Ati#Early_KMS_start I think the radeon-initrd package has been removed since it's no longer needed, should the "tip" just say "add radeon to HOOKS" and not install radeon-initrd? --Unhammer 03:14, 23 August 2010 (EDT)
- actually, I think the tip should probably just be removed... I tried adding radeon to HOOKS without having radeon-initrd, and got an error on making the image (I used to need radeon-initrd for 3d to work, now it seems to be enough to have the newest linux-firmware, no special HOOKS needed). --Unhammer 07:29, 23 August 2010 (EDT)
- Yes, it should be removed. Please do so. Rransom 19:01, 24 August 2010 (EDT)
required virtual size does not fit available size ...
I was getting this error when trying to setup two screens with different resolutions using 'gnome-display-properties' (which in turn uses xrandr) : required virtual size does not fit available size: requested=(3200, 1080), minimum=(320, 200), maximum=(1920, 1920)
Removing 'nomodeset' from my kernel parameters fixed it. Removing the 'nomodeset' paramter should be discussed somewhere. I don't know enough about what it actually does to do so intelligently. --Gdw2 13:52, 21 March 2011 (EDT)