You are not logged in.
I just jumped into the shark-infested pond that is the testing repository for the hell of it
But yes, 2D performance got an amazing boost, differing from when I used the xf86-video-intel driver in extra. In the older driver, I would experience random graphical glitches whenever my screen changed resolution, I had just logged in, or Wine programs start.
Just a quick question, is the kernel in the testing repository stock or modified by Arch devs? Just wondering if there's some random way to further increase performance, though kernel compilation scares me
Offline
The kernel is fairly vanilla. Here is the repo of the patches applied: http://projects.archlinux.org/?p=linux- … ;a=summary
I am also enjoying the speed boost. I can play games again instead of doing work. The last two weeks have been incredibly productive!
Offline
You can look at the PKGBUILD and see what patches are applied:
32-bit: http://repos.archlinux.org/viewvc.cgi/k … iew=markup
64-bit: http://repos.archlinux.org/viewvc.cgi/k … iew=markup
Last edited by Fackamato (2009-04-07 12:52:45)
Offline
Thanks! This is totally derailing my own topic, but if I randomly decided to downgrade back from Testing, say if 2.6.29 and xf86-video-intel-legacy were to be released, what would be the easiest way to do that?
Offline
You can look at the PKGBUILD and see what patches are applied:
32-bit: http://repos.archlinux.org/viewvc.cgi/k … iew=markup
32-bit: http://repos.archlinux.org/viewvc.cgi/k … iew=markup
Are you sure? I don't see the rt2870 patch in there but it's definitely in 2.6.29.1-2 (thanks for that tpowa, my wlan card finally works without custom drivers plus, arch is now probably the only distro to support this card out of the box! )
Last edited by schuay (2009-04-07 12:09:34)
Offline
I tried too the Xorg 1.6 + xf86-video-intel-legacy on my laptop (and libgl, mesa, etc.), that is equipped with a 945GM and it worked much better than Xorg 1.6 + xf86-video-intel. But the current Xorg 1.5.3 + xf86-video-intel from extra is still the best. Some benchmarks
Xorg 1.5.3 + xf86-video-intel 2.4.3
glxgears ~620 fps, compiz ~110 fps, assaultcube ~40-50 fps
Xorg 1.6 + xf86-video-intel-legacy
glxgears ~450 fps, compiz ~85 fps, assaultcube ~30-40 fps
Xorg 1.6 + xf86-video-intel
glxgears ~150 fps, compiz ~25-30 fps, assaultcube not playable.
Regards.
Giuseppe Borzi' - Registered Linux user #34028
Offline
Thanks! This is totally derailing my own topic, but if I randomly decided to downgrade back from Testing, say if 2.6.29 and xf86-video-intel-legacy were to be released, what would be the easiest way to do that?
Just remove the testing repo from your pacman.conf and all the testing packages you use will eventually go to core/extra.
Offline
Fackamato wrote:You can look at the PKGBUILD and see what patches are applied:
32-bit: http://repos.archlinux.org/viewvc.cgi/k … iew=markup
32-bit: http://repos.archlinux.org/viewvc.cgi/k … iew=markupAre you sure? I don't see the rt2870 patch in there but it's definitely in 2.6.29.1-2 (thanks for that tpowa, my wlan card finally works without custom drivers plus, arch is now probably the only distro to support this card out of the box! )
Look at the link I gave. It is probably in the arch patch conglomerate.
Offline
I just installed intel-legacy drivers... now I have more or less the same performance as before but without screen artifacts or crashes. Yeah!
By the way, when I run a 3D app (even glxgears) I get a message "Failed to initialize GEM. Falling back to classic.". Should I worry about it? I'm running a 2.6.29 kernel, so I thought I had GEM up and running...
Offline
Look at the link I gave. It is probably in the arch patch conglomerate.
I did. I was trying to point out that the PKGBUILD doesn't show which patches are being applied to the kernel - confused me too for a while
Offline
I just installed intel-legacy drivers... now I have more or less the same performance as before but without screen artifacts or crashes. Yeah!
By the way, when I run a 3D app (even glxgears) I get a message "Failed to initialize GEM. Falling back to classic.". Should I worry about it? I'm running a 2.6.29 kernel, so I thought I had GEM up and running...
I'd like to know this too, just ran that and had the same result.
Offline
The legacy driver is too old to support GEM. Not only your kernel has to support it, but also the 2D driver. The fallback to classic means that the old code path is taken instead of the new slow GEM codepath, which increases performance a lot this way.
Offline
The legacy driver is too old to support GEM. Not only your kernel has to support it, but also the 2D driver. The fallback to classic means that the old code path is taken instead of the new slow GEM codepath, which increases performance a lot this way.
Well, actually I made some tries, the legacy driver is much more stable (I should say that non-legacy is much more UNstable...), but it is actually slower than the 2.6.99.902 in [testing]... Damn Intel... as far as you know, is there hope for proper support of old graphic chips in future releases of the non-legacy driver? Also for KMS and DRI2 goodness...
Offline
I just tried out the non legacy in testing and... Surprisingly enough, my 3D performance is upped. I still get random artifacts when I change resolutions, a problem I had before I switched to the legacy driver in the first place, but its running quite smoothly.
Something's wrong here XD
Offline
The intel-legacy drivers conflict with xorg-server-1.5.3 (probably the entire 1.5 branch). They want to remove xorg-server on install but offer nothing to replace it.
I have an Inspiron 1100, whose graphics chip is an 82845G/GL ("Brookdale") Intel IGP. The xf86-video-intel driver glitches horribly on this machine, crashing X on attempted start and creating wide bars of corruption that persist even in console mode and when changing the driver.
What can be done about this? Is there an older xorg-server PKGBUILD anywhere? AUR doesn't have one.
Offline
The new intel non-legacy drivers in testing ar up to par with the legacy drivers in my intel 945.. YMMV
Offline
snif, I'm going to cry, after 6 months of having Option "NoAccel" "true" in my xorg.conf I can finaly delete it!!! this is a big day!!, I can run google-earth!!
And where were all the sportsmen who always pulled you though?
They're all resting down in Cornwall
writing up their memoirs for a paper-back edition
of the Boy Scout Manual.
Offline
How to switch from xf86-video-intel to this xf86-video-intel-legacy driver?
Offline
How to switch from xf86-video-intel to this xf86-video-intel-legacy driver?
Simple removing "pacman -Rd xf86-video-intel" and installing "pacman -Sy xf86-video-intel-legacy" should do it, I think. Settings should stay the same.
Offline
xodeus wrote:How to switch from xf86-video-intel to this xf86-video-intel-legacy driver?
Simple removing "pacman -Rd xf86-video-intel" and installing "pacman -Sy xf86-video-intel-legacy" should do it, I think. Settings should stay the same.
That was easy. And it's working...
But.
glxgears ~250 FPS assaultcube not playable
Never tried to play any games before. Should I mention that I'm running with kms enabled?
Offline
drag0nl0rd wrote:xodeus wrote:How to switch from xf86-video-intel to this xf86-video-intel-legacy driver?
Simple removing "pacman -Rd xf86-video-intel" and installing "pacman -Sy xf86-video-intel-legacy" should do it, I think. Settings should stay the same.
That was easy. And it's working...
But.
glxgears ~250 FPS assaultcube not playable
Never tried to play any games before. Should I mention that I'm running with kms enabled?
No, you should not, cause you aren't. KMS is only available with UXA on newer intel drivers. And I would recommend you to play with your xorg.conf, you _might_ get much better performance with e.g. xaa.
Offline