You are not logged in.
I experienced another problem:
after compiling the latest xorg-server (with downgrading of libxfont, I'll try out the patch later) I got a white desktop after starting Compiz. I then downgraded the xorg-server and had really strange performance issues. I tracked down the problem and downgraded libdrm back to 2.4.3 (from 2.4.4) and everything was smooth again.
I try out the patch now and update all packages again (libxfont and libdrm) and report back if there are any problems
Akii
Edit: Problem is present event with the patch though I think this might be a problem with compiz itself not with the driver... we'll see...
Last edited by Akii (2009-02-25 18:13:38)
Offline
I have a desktop with the G45 chipset and I'm using the integrated X4500HD card. It runs fine and I've had no problems with it. I'm using the provided packages by Arch Linux (xf86-video-intel-2.4.3). When all the stack gets upgraded I will get higher performance (in Fedora 10 I do get it), but even in it's current state the performance is perfectly ok. gtkperf runs in about 5-6 seconds. Kwin effects are smooth (though I turn them off most of the time, i just don't like them). Tuxracer runs fine too.
I can't say about HDMI audio, since I haven't tried it, but I think it should work. Can't say about switching between displays either.
Thanks I was pretty scared with these threads about intel graphics card and having to take that, and that, and also that from git, using rc versions etc.
Offline
X Server 1.6 just got released, and according to the arch-dev-public the new stack will hit testing this week. 2.6.29 will be out probably in late March but I am gonna try a -rc and report back soon.
Offline
I never could compile xorg-server without being hit with some errors. What packages would this be related to:
rroutput.c:542: warning: no previous prototype for 'RRSetPrimaryOutput'
rroutput.c: In function 'ProcRRSetOutputPrimary':
rroutput.c:568: error: 'xRRSetOutputPrimaryReq' undeclared (first use in this function)
rroutput.c:568: error: (Each undeclared identifier is reported only once
rroutput.c:568: error: for each function it appears in.)
rroutput.c:568: error: 'stuff' undeclared (first use in this function)
rroutput.c:568: error: expected expression before ')' token
rroutput.c: In function 'ProcRRGetOutputPrimary':
rroutput.c:606: error: 'xRRGetOutputPrimaryReq' undeclared (first use in this function)
rroutput.c:606: error: 'stuff' undeclared (first use in this function)
rroutput.c:606: error: expected expression before ')' token
rroutput.c:609: error: 'xRRGetOutputPrimaryReply' undeclared (first use in this function)
rroutput.c:609: error: expected ';' before 'rep'
rroutput.c:626: error: 'rep' undeclared (first use in this function)
make[1]: *** [rroutput.lo] Error 1
make[1]: Leaving directory `/home/ronin-sage/builds/xorg-server/src/xorg-server-1.6.0/randr'
make: *** [all-recursive] Error 1
==> ERROR: Build Failed.
Aborting...
Offline
X Server 1.6 just got released, and according to the arch-dev-public the new stack will hit testing this week. 2.6.29 will be out probably in late March but I am gonna try a -rc and report back soon.
seeing the new intel driver has just been released, I tried the stack again with the new x server and intel driver.
Apart from realising the evdev driver also needed to be updated for hotplugging to work, it went without a hitch. I am now enjoying smooth [uxa] accelerated graphics. Very nice (and somewhat annoying I manage to make it work a week before it hits testing - but hell, I learned a lot).
Will test difference between standard and latest git kernel later.
Last edited by b9anders (2009-02-26 20:43:22)
Offline
fijam wrote:X Server 1.6 just got released, and according to the arch-dev-public the new stack will hit testing this week. 2.6.29 will be out probably in late March but I am gonna try a -rc and report back soon.
seeing the new intel driver has just been released, I tried the stack again with the new x server and intel driver.
Apart from realising the evdev driver also needed to be updated for hotplugging to work, it went without a hitch. I am now enjoying smooth [uxa] accelerated graphics. Very nice (and somewhat annoying I manage to make it work a week before it hits testing - but hell, I learned a lot).
Will test difference between standard and latest git kernel later.
Can you list the version of the following package you used:
libdrm
xf86-video-intel
xorg-server
kernel
I am quite confused why when I enable "UXA" with intel driver 2.6.2. my X was locked and it gives me funny screen
Offline
b9anders wrote:fijam wrote:X Server 1.6 just got released, and according to the arch-dev-public the new stack will hit testing this week. 2.6.29 will be out probably in late March but I am gonna try a -rc and report back soon.
seeing the new intel driver has just been released, I tried the stack again with the new x server and intel driver.
Apart from realising the evdev driver also needed to be updated for hotplugging to work, it went without a hitch. I am now enjoying smooth [uxa] accelerated graphics. Very nice (and somewhat annoying I manage to make it work a week before it hits testing - but hell, I learned a lot).
Will test difference between standard and latest git kernel later.
Can you list the version of the following package you used:
libdrm
xf86-video-intel
xorg-server
kernelI am quite confused why when I enable "UXA" with intel driver 2.6.2. my X was locked and it gives me funny screen
libdrm: 2.4.5
xf86-video-intel: 2.6.2
xorg-server: 1.6 (note: evdev needs to be 2.1.3 for hotplugging to work)
kernel: arch standard
and otherwise the PKGBUILDS posted in page 1 of this thread (although libx11, libxcb, inputproto and pixman are now irrelevant as the arch packages are up to date) with the git packages built a few days ago, though I suspect those could be replaced with the stable packages released (and not yet in arch).
Last edited by b9anders (2009-02-26 21:00:57)
Offline
phabulosa wrote:b9anders wrote:seeing the new intel driver has just been released, I tried the stack again with the new x server and intel driver.
Apart from realising the evdev driver also needed to be updated for hotplugging to work, it went without a hitch. I am now enjoying smooth [uxa] accelerated graphics. Very nice (and somewhat annoying I manage to make it work a week before it hits testing - but hell, I learned a lot).
Will test difference between standard and latest git kernel later.
Can you list the version of the following package you used:
libdrm
xf86-video-intel
xorg-server
kernelI am quite confused why when I enable "UXA" with intel driver 2.6.2. my X was locked and it gives me funny screen
libdrm: 2.4.5
xf86-video-intel: 2.6.2
xorg-server: 1.6 (note: evdev needs to be 2.1.3 for hotplugging to work)
kernel: arch standardand otherwise the PKGBUILDS posted in page 1 of this thread (although libx11, libxcb, inputproto and pixman are now irrelevant as the arch packages are up to date) with the git packages built a few days ago.
Many thanks !
Do you have a PKGBUILD for your xorg-server 1.6?
BTW, what intel chipset do you have?
Offline
Many thanks !
Do you have a PKGBUILD for your xorg-server 1.6?BTW, what intel chipset do you have?
I just took the pkgbuild from page 1 and changed the --disable-dri2 line to --enable-dri2 and changed the version number. This should also work with the pkgbuild from abs (at least it did with the release candidate I compiled with that build).
I have a Intel GMA X3100 / Mobile GM965/GL960 chipset.
Offline
I am not seeing much difference with the .29 kernel. But I am fairly new to kernel compilation, so I might have missed some optimisations/garbled the ones in place.
Offline
Damn... I am using newest xorg, updated driver, latest drm and mesa from testing and performance still sucks too much... kwin effects are slowwwwwww on my GMA 945.
Excuse my poor English.
Offline
Do you use UXA or EXA? Any numbers you can provide?
My experience is good so far. With the previous stack I'd have ~24 fps in tuxracer 1280x800, now I have a consistant 45fps with EXA and about 35-40 with UXA. When I started tuxracer from terminal I noticed:
get fences failed: -1
param: 6, val: 0
Apparently, fencing is not available in 2.6.28.
Also, flash videos with compositing enabled are no longer choppy.
edit: That is i915GM
Last edited by fijam (2009-03-03 02:53:08)
Offline
Hi
I was trying to recompile xorg-server with all the new stuff from testing cause it seems that there is a memory leak.
Recompilation fails with missing /usr/include/GL/internal/glcore.h. Can someone provide information where this file belongs to?
Thanks,
david
EDIT: I was able to compile it with an updated version of the PKGBUILD but had to disable kdrive.
Last edited by dcrabs (2009-03-05 10:22:05)
Offline
Guys there's something weird: Yesterday I tried Ubuntu's 8.10 LiveCD and guess what. The damned laptop was faster (compiz-and-graphics-wise) than with my current Arch install. So I started searching around, Their Xorg.Conf is just minimal, check this: http://bbs.archlinux.org/viewtopic.php?id=67089 <-- There is my Xorg.conf after I saw ubuntu's. Just minimal. Some other thing I noticed is that enabling Loose Binding Boosts performance a lot. BUT Ubuntu doesn't have it enabled. So there must be some other thing that can be done to boost performance. Their kernel and xf86-video-intel versions are older than ours. Maybe if someone savvy got their hands on a LiveCD could check what's the thing.
Offline
Guys there's something weird: Yesterday I tried Ubuntu's 8.10 LiveCD and guess what. The damned laptop was faster (compiz-and-graphics-wise) than with my current Arch install. So I started searching around, Their Xorg.Conf is just minimal, check this: http://bbs.archlinux.org/viewtopic.php?id=67089 <-- There is my Xorg.conf after I saw ubuntu's. Just minimal. Some other thing I noticed is that enabling Loose Binding Boosts performance a lot. BUT Ubuntu doesn't have it enabled. So there must be some other thing that can be done to boost performance. Their kernel and xf86-video-intel versions are older than ours. Maybe if someone savvy got their hands on a LiveCD could check what's the thing.
It's no wonder, ubuntu from october '08 uses older arhitecture and probably older intel drivers which worked noticabley faster.
Just check this thread and the link in my signature.
...
Last edited by karabaja4 (2009-03-10 03:54:55)
Offline
No, I actually checked xorg-server and xf86-video-intel versions and they were surprisingly new. Xorg-server was version 1.5.something and intel 2.4.2 i think. I was really amazed but I thought too that they were using the good old TTM based server. Something that bothered me, is that the CD used a 2.6.27 kernel which, afaik doesn't include GEM. Maybe it's patched?
Offline
Has anyone tried newest xf86-video-intel-2.6.99.901 RC?
Offline
I tried to compile it... but it gives me errors about incorrect files.
I think the tarball isn't complete!
Sorry for my bad English
Offline
No, I actually checked xorg-server and xf86-video-intel versions and they were surprisingly new. Xorg-server was version 1.5.something and intel 2.4.2 i think. I was really amazed but I thought too that they were using the good old TTM based server. Something that bothered me, is that the CD used a 2.6.27 kernel which, afaik doesn't include GEM. Maybe it's patched?
I tried ubuntu again and yes, the performance is better. It's weird and as you said the xorg.conf is very minimal. What can we do??
Excuse my poor english
Offline
Better than arch testing? I really doubt that. Performance with the old drivers was about the same as with arch (xf86-video-intel 2.4.2 and xorg-server 1.4)
Now things are much better.
My victim you are meant to be
No, you cannot hide nor flee
You know what I'm looking for
Pleasure your torture, I will endure...
Offline
Has anyone tried newest xf86-video-intel-2.6.99.901 RC?
Yes, many bug fixes while performance is even lower on my 965.
I tried to compile it... but it gives me errors about incorrect files. I think the tarball isn't complete!
Yes, there is 2.6.99.902 now.
You need to install an RTFM interface.
Offline
I tried it too but it still leaks memory like crazy with KDE4.2 and desktop effects enabled.
For example the expose effect adds ~1MB to the X process.
Maybe it is not the driver itself but I have no idea where to search.
Offline
Hi. Have somebody solution to low performance of Intel X3100 card? I have spent over hundred hours to test a lot of drivers, and I didn't get any solution. I googled a lot of topics on other forums and I didn't find any solution. Now I have latest stable drivers and still "broken pc" by this...
Thanks!
IBM/Lenovo R61i
Intel core 2 duo 1024MB RAM 120GB HDD
Archlinux 64bit KDEmod 4
without Windows
Offline
Just wait another week (maybe 2 to hit the repos) for 2.6.31 and it will improve... BTW u had some newer topics about this, didn't have to digg this one out
Offline
Yes, but it doesn't matter or no? :-D. Nobody didn't respond long time, I wrote here ;-).
Thanks, I believe you :-D.
IBM/Lenovo R61i
Intel core 2 duo 1024MB RAM 120GB HDD
Archlinux 64bit KDEmod 4
without Windows
Offline