You are not logged in.
Pages: 1
Is still needed to disable the page flipping functionality in X when using intel driver?
From my xorg.log:
[ 11680.004] (II) intel(0): Pageflipping enabled in kernel, but disabled in X
[ 11680.004] (II) intel(0): Don't panic: https://bugzilla.redhat.com/588421
Maybe i didn't understood well, but here:
https://admin.fedoraproject.org/updates … 1.0-4.fc13
...the bug seems to be solved (maybe is solved just by disabling pageflipping?)
Also, does pageflipping really boost performance?
-edit
I'm asking because i get very bad performance even in world of goo when going anything higher than 1024x600. (945GME on eeepc), and also glxgears acts strange at higher resolutions (i have to move the mouse around or just keep my finger on the touchpad to get right fps...), and maybe this is related.
Last edited by kokoko3k (2010-08-16 18:20:30)
Help me to improve ssh-rdp !
Retroarch User? Try my koko-aio shader !
Offline
(...) and also glxgears acts strange at higher resolutions (i have to move the mouse around or just keep my finger on the touchpad to get right fps...), and maybe this is related.
I have the same behaviour... don't worry with glxgears. But anyone knows why do we have to use the touchpad or have some activity on background to keep my/our systems alive?
I also have abnormal low fps with 3D games on a gma900.
Last edited by TigTex (2010-08-16 19:02:23)
.::. TigTex @ Portugal .::.
Offline
AFAIK the bug is solved, but xf86-video-intel is not compiled to use it. It should be straightforward to rebuild it with pageflipping enabled.
Offline
test 2.6.35 + intel drivers from upstream (skip the remove flipping patch in the PKGBUILD)
it works great here (even if glxgears cant sync to the 60fps it should the rest of the system works great)
Offline
test 2.6.35 + intel drivers from upstream (skip the remove flipping patch in the PKGBUILD)
it works great here (even if glxgears cant sync to the 60fps it should the rest of the system works great)
Is 2.6.35 a typo or did you mean that the new kernel is needed to safely remove the no-pageflipping patch from 2.12 intel driver?
-EDIT:
Tried the 2.12 driver from abs and removed the patch command.
Performance are more or less the same, and the system seems NOT stable (i've had two freeze playing with compiz indirect/direct rendering while worldofgoo and glxgears were running, but i guess it is normal, or at least to be expected, and another just by launching glxgears under compiz)
-EDIT2:
pageflipping unrelated:
A little offtopic, but switching to compiz with direct rendering turned on, makes worldofgoo (in window mode) smoother.
Unfortunately some garbage is left on screen with qt3 applications menu and evenin firefox.
Last edited by kokoko3k (2010-08-17 08:44:14)
Help me to improve ssh-rdp !
Retroarch User? Try my koko-aio shader !
Offline
2.6.35 is not a typo -- the kernel fix went in in 2.6.35.
Offline
Ok then, it was unclear to me where the bug came from and i will make another try when kernel35 will leave [testing].
But most important, did anyone experienced a performance boost with pageflipping?
Help me to improve ssh-rdp !
Retroarch User? Try my koko-aio shader !
Offline
Ok then, it was unclear to me where the bug came from and i will make another try when kernel35 will leave [testing].
But most important, did anyone experienced a performance boost with pageflipping?
You are just gonna wait to long for kernel35 to hit extra!
I only know that performance with xf86-video-intel is very bad and Kwin on KDE not working good.
I do not speak English, but I understand...
Offline
Ok then, it was unclear to me where the bug came from and i will make another try when kernel35 will leave [testing].
But most important, did anyone experienced a performance boost with pageflipping?
I'm not sure if it's pageflipping or the 2.6.35 kernel or both together, but I do see an improvement on my netbook. Duke Nukem 3D is now almost playable in polymost mode. Don't expect miracles though and accept that the 950GMA simply sucks (it really does) and that a netbook is not a gaming machine, not even for simple games.
Offline
kokoko3k wrote:Ok then, it was unclear to me where the bug came from and i will make another try when kernel35 will leave [testing].
But most important, did anyone experienced a performance boost with pageflipping?I'm not sure if it's pageflipping or the 2.6.35 kernel or both together, but I do see an improvement on my netbook. Duke Nukem 3D is now almost playable in polymost mode. Don't expect miracles though and accept that the 950GMA simply sucks (it really does) and that a netbook is not a gaming machine, not even for simple games.
To a large degree, yes, but it should be noted that performance for these drivers used to be quite good(relatively speaking) before Q2 2008 or so(I used to be able to get a solid 50fps or so in UrbanTerror) and has been sporadic at best ever since.
Offline
Today i tried 2.6.35 and i've to say that things seems to be smoother, but only if i don't use pageflipping (!?)
I've also disabled vsync by default in xorg.conf (Option "SwapbuffersWait" "False") which seems to help too.
However, 2.6.33 performs definitely better without vsync.
Last edited by kokoko3k (2010-08-20 11:17:56)
Help me to improve ssh-rdp !
Retroarch User? Try my koko-aio shader !
Offline
Also installed the new kernel today, the Problems remain. Tried several different configurations.
What I noticed is that when I start X, the performance is very good (as it was in 2008/2009) for about 5 seconds and then it degrades quickly and remains low. Btw, glxgears doesn't turn the gears. FPS get printed to the terminal, but the animation is not apparent. I tried vsync off, on, tiling on and off. Also, the PageFlip option doesn't work ([ 1741.789] (WW) intel(0): Option "PageFlip" is not used). Same for "TripleBuffer".
GMA945 here. I really miss the 2008/2009 times when everything was super-smooth. Why oh why doesn't this regression get fixed...
Offline
I now filed a bug at freedesktop about this: https://bugs.freedesktop.org/show_bug.cgi?id=29704
Offline
Pages: 1