Red Hat Bugzilla – Bug 124065
xmms causes high cpu loads for X
Last modified: 2007-11-30 17:10:43 EST
From Bugzilla Helper:
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:1.6)
Description of problem:
X normally takes up about 1% cpu load when idle. However if you start
xmms, this jumps up to about 4-5% cpu load.
Version-Release number of selected component (if applicable):
Steps to Reproduce:
1. run top and monitor X cpu load
2. start xmms
3. observe cpu load for X increasing by 500%
Actual Results: X cpu load remains at a constant 5% cpu load or higher.
Expected Results: X cpu load should be at 1% or so when idle.
Additional info: xorg-x11-6.7.0-2
Note: After compiling a vanilla kernel and install nVidia drivers, I
can no longer reproduce this bug.
I am seeing this, too. It does not happen with all songs, and happens
only if xmms is shaded.
The effect disappears when Title Scrolling is disabled in the xmms
main window (even though the scrolling title is not even visible in
The X CPU load is ~20% on a 1200MHz Duron.
Can you narrow this down to particular songs? How often does it happen?
I tried to find a freely available MP3 to reproduce this, and found
"Every OS sucks", by Three Dead Trolls in a Baggie, available for
Reproducing the effect:
- delete ~/.xmms/xmms.m3u
- start xmms
- unshade the main window
- disable autoscrolling title (right click on title)
- start mp3 playing (CPU usage is low. 6-8%, backgorund noise)
- shade main window (cpu usage is low)
- unshade main window, enable title scrollong (cpu usage is low)
- shade main window (cpu usage goes up to ~25-30%)
- look at top, see X take 18-20% CPU time
- unshade main window (cpu usage is low again)
RedHat should remove xmms and replace it with Beep Media Player.
Moved upstream here:
While I agree that BMP looks cool, it shows exactly the same bahaviour :)
I was under the impression that beep-media-player fixed the problem,
but you are right. Actually, loading the nVidia module and using
nvidia driver fixes the problem for me. If this were an xmms bug,
then why would using a different driver fix the problem? I'm thinking
it's something with xorg-x11.
Ralf, do you have an nvidia card too?
Yes, I do (GeForce2MX, NV11).
I nocticed something interesting today. The problem goes away (almost
completely) when not using a framebuffer console.
I was using vesafb as the console driver, and using this, the
behaviour I described above occurs.
When not using a fb console, "scrolling when shaded" still uses more
CPU than scrolling while not shaded", but the effect is much smaller
(5-7% unshaded, 7-9% shaded).
Using rivafb shows the same effect as using no fb at all, so this is
somehow tied to the vesafb.
I think it may have something to do with mtrr settings, these are
different when using vesafv vs no fb/rivafb (the two latter are equal)
I should probably not that I too have a GeForce2MX, NV11.