You are not logged in.
Before it's officially released, xorg-server 1.6.1 arrived at arch repos.
Let's use this topic to discuss performace, bugs and news about this update.
Last edited by lycaeum (2009-04-15 19:56:29)
Arch Linux X86_64 (Pure 64bits Blend)
Kernel 2.6.29.3 w/ KMS and ext4fs -
Xorg 1.6.1 w/UXA - Gnome 2.26.1 - compiz 0.8.2
Offline
From http://lists.freedesktop.org/archives/x … 45086.html
Adam Jackson (6):
randr: Fix thinko in xf86TargetPreferred
selinux: Don't bother relabeling resources that are being destroyed
selinux: Add support for avc_acquire_netlink_fd()
Fix Changelog generation
Fix distcheck with dolt
xserver 1.6.1
Eamon Walsh (3):
Correct access mode in call to dixLookupWindow() within RRSelectInput.
xselinux: Don't require incoming context strings to be null-terminated.
xselinux: Don't BadAlloc in List* requests if there are no items to list.
Eric Paris (1):
This patch changes all places in the X code to use _raw functions. The
Kristian Høgsberg (1):
glx: Fix drawable private leak on destroy
Peter Hutterer (5):
Xi: add XIPropToInt() auxiliary function.
Xi: add XATOM_FLOAT to server-defined properties.
Xi: don't crash on a NULL property name, just return None.
xkb: Fix wrong colour reference in XKB geometry copying. #20081
Xext: set POINTER_SCREEN flag in XTestFakeInput if necessary. (RH #490984)
nd
Looks like there isn't much good stuff for us in this new release. fixed some keyboard issues and minor leaks in glx
Last edited by lycaeum (2009-04-15 20:39:06)
Arch Linux X86_64 (Pure 64bits Blend)
Kernel 2.6.29.3 w/ KMS and ext4fs -
Xorg 1.6.1 w/UXA - Gnome 2.26.1 - compiz 0.8.2
Offline
Before it's officially released, xorg-server 1.6.1 arrived at arch repos.
Let's use this topic to discuss performace, bugs and news about this update.
I'm running the 4500MHD chipset inside a acer extensa 5630.
I am very pleased by the 1.6.1 release, since now exa not only matches the performance I had with xaa on 1.5.x, it feels even faster.
When I upgraded to 1.6.0 I was kind of disappointed that flash became very slow in general and flickering on some sites. Somtimes the GUI (gpodder in my case) could be watched while doing its drawings.
All gone, though I had to add
Option "MigrationHeuristic" "greedy"
to the device section in xorg.conf to get rid of graphic artefacts (with this applet: http://kde-look.org/content/show.php/me … ent=86630) and to get the disired speep up for flash.
Thanks for your work, xorg and archlinux devs.
-- Maik
Offline
lycaeum wrote:Before it's officially released, xorg-server 1.6.1 arrived at arch repos.
Let's use this topic to discuss performace, bugs and news about this update.
I'm running the 4500MHD chipset inside a acer extensa 5630.
I am very pleased by the 1.6.1 release, since now exa not only matches the performance I had with xaa on 1.5.x, it feels even faster.
When I upgraded to 1.6.0 I was kind of disappointed that flash became very slow in general and flickering on some sites. Somtimes the GUI (gpodder in my case) could be watched while doing its drawings.
All gone, though I had to add
Option "MigrationHeuristic" "greedy"
to the device section in xorg.conf to get rid of graphic artefacts (with this applet: http://kde-look.org/content/show.php/me … ent=86630) and to get the disired speep up for flash.Thanks for your work, xorg and archlinux devs.
-- Maik
Have you tried out KMS?
freenode.net #linuxn00b
Offline
Have you tried out KMS?
No
Offline
Maybe Catalyst dont support xorg 1.6.1 ?
Offline
As for 1.6.0 vs 1.6.1: the 1.6.0 we had in our repositories was 1.6.1 without the last commit that fixes some memory leaks. I don't expect performance increases with 1.6.1 compared to our 1.6.0 package.
Offline
and intel released xf86-video-2.7.0, compiled and installed here but kdm loads with artifacts too(like 2.6.99.903), but now the kde4 loads with artifacts too, so uninstall 2.7 :-/
Offline
I'm experiencing a huge memory leak with the 1.6.1 package.
Offline
I'm experiencing a huge memory leak with the 1.6.1 package.
I'm too experiencing some major memomy leaks as my log files fill out my /var partition all the time with memory errors.
Offline
I'm re-posting this from the testing forums, due to lack of replies...
As some others have reported, I am also having problems with moving windows in the new xorg (1.6.0 as well as 1.6.1).
When I move any window, the CPU usage goes up drastically, and when I keep moving it tops at 100%. The window moving is also a bit slow: when I keep moving, the window lags behind my cursor.
I'm using the catalyst driver, with my integrated radeon 3200. My window manager is fluxbox.
A search on the net has given a few threads reporting the same problem, it is even reported as a bug in xorg: https://bugs.launchpad.net/xorg-server/+bug/179348.
In another thread ( http://www.mail-archive.com/debian-bugs … 34308.html ) someone says this:
I confirm that running the latest DRM allows the driver to use EXA and
everything works fine.
Also the content of /proc/mtrr is the one that is supposed to be.
What is "DRM", and how can I find the latest version? I have set the driver to use "EXA" in my xorg.conf, but this is apparently not supported by the catalyst driver (I get a message stating this in the x log).
So what is best to do? Should I just wait for a new version of the catalyst driver, or a new xorg? This problem is quite annoying; every time I move a window, 30% of my cpu is used...
Thanks in advance for replies!
Edit: I've tried the xf86-video-ati driver. EXA got enabled correctly, but the problem still remained: very high CPU usage when moving windows.
Last edited by ignus (2009-04-26 09:05:21)
Offline
Only guessing as I have no ATI board. But if this occurs at moving windows only, the cpu appears to be processing the display plane contents over and over again in an attempt to keep the window picture up to date.
Did you try to hide window contents while moving, showing frames only?
Does it also occur when resizing (even if not up to 100%).
Some googling (e.g. this:http://www.phoronix.com/forums/showthread.php?t=8071 reveals this as a rather common EXA problem.
Last edited by bernarcher (2009-04-26 09:49:51)
To know or not to know ...
... the questions remain forever.
Offline
Thanks for your reply.
The same happens when resizing windows (on openbox now, which has opaque resize), and when I hide the window contents and just move the window title bar.
I do not have EXA enabled, the Catalyst driver doesn't seem to support it. I get a message in the xorg log that it could be initialized, when I try to enable it in my xorg.xonf.
Last edited by ignus (2009-04-26 10:08:20)
Offline
Hello,
I have a notebook with ATI Xpress 1150 and using open source drivers since about January. As I switched from proprietary driver to xf86-video-ati, framerate decreased from ~1500fps to ~1300fps. It was something I could live with, but since xorg server 1.6 I have ~300fps. All log files are OK, glxinfo says direct rendering is on, but performance is suffering badly. The same problems were noticed here: http://bbs.archlinux.org/viewtopic.php?id=71643
For me it's a bug, which I would submit, but I don't know if it's in the server or in the ati module...any suggestions?
Offline
To those having memory leaks.. I have a desktop using Nvidia graphics, and a notebook using Intel graphics, both with KDE 4.2. Both of them are leaking with 1.6.1 (and the updated 1.6.901-1 package that is current as of this writing). It's aggravating that the Xorg developers seem to be intent on making Linux waste more reources than Vista while not really adding anything to the experience.
There are many ways to get the leak to accelerate so it becomes noticeable. On my desktop with the Nvidia card, I can run Half-Life 2 in wine for about an hour and have X using over 4 GB of RAM.. AFTER I've fully exited wine and there is basically nothing running but an empty desktop. My record is that Xorg had over 6GB resident in memory, and over 2 GB in swap so my system that has 8GB of memory was swapping... thank you X developers for making Vista look lean & mean.
Anyway, here's a python script that will cause the leak to occur, but in a slow manner that won't destroy your system quickly. It assumes you have KDE's "kcalc" program available, but can be easily edited to run different programs. All it is doing is running some simple desktop programs, and then cleanly exiting from them repeatedly. Let it run for a few hours and watch X's memory usage climb higher & higher... for no good reason.
from subprocess import Popen
import time
RUN_CMD = "/usr/bin/kcalc"
# test program that automatically executes, then kills xterms.
# Designed to show that x.org will memory leak simply by running & then
# killing trivial programs.
while (True): # continue until the user kills us
process_list = []
# launch then kill & wait 10 at a time:
for II in xrange (0, 10):
process_list.append (Popen (("kcalc%d" % II), executable=RUN_CMD,
close_fds=True))
time.sleep (10)
for nextProc in process_list:
nextProc.kill ()
nextProc.wait () # no zombies!
Offline
Now, is this a problem of X or of KDE? I don't run KDE currently and still have to experience a memory leak like this one.
KDE 4.2 still is crap. Waiting for 4.5 or later ...
To know or not to know ...
... the questions remain forever.
Offline