<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=Content-Type content="text/html; charset=iso-8859-1">
<META content="MSHTML 6.00.2800.1491" name=GENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=#ffffff>
<DIV><FONT face=Arial size=2>Hello - I wonder if anyone has ever seen this, or
knows of any information on it, or what not...any help is greatly
appreciated, as my employer has tasked me with figuring this one out! :) I
paste some of an email I just wrote to another person...</FONT></DIV>
<DIV><FONT face=Arial size=2></FONT> </DIV>
<DIV><FONT face=Arial size=2></FONT> </DIV>
<DIV>Here is the situation I'm trying to deal with - the goal is to run
3<BR>simultaneous instances of mplayer, one on each of three monitors on 3
nvidia<BR>geforce 5200 video cards, using the motion compensation codecs.
I don't<BR>understand xorg that well yet (I'm actually an audio engineer), but
the lead<BR>developer tells me that it's three different X servers as opposed to
one (as<BR>in Xinerama?)<BR><BR>So using the xv video output in mplayer (mplayer
-vo xv) this works fine, I<BR>play back and it's great and I have yet to see a
crash. Using xvmc (<BR>mplayer -vc ffmpeg12mc -vo xvmc ), what happens is,
after about an hour of<BR>looping videos (I'm looping 2-3 minute videos over and
over), mplayer will<BR>start to crash. One instance will crash, then some
minutes later, another,<BR>and another. Mplayer crashes when it starts a
new loop. So it doesn't<BR>crash in the middle of the video. It'll
finish the video a bunch of times,<BR>and then finally, on the 20th time or
whatever number it is (I haven't<BR>counted and I don't think it's the same each
time), it tries to play the<BR>video again and crashes with this
output:<BR><BR>Starting playback...<BR>vd_ffmpeg: XVMC accelerated
MPEG2<BR>trying pixfmt=0<BR>VDec: vo config request - 720 x 480 (preferred csp:
MPEG1/2 Motion<BR>Compensation and IDCT)<BR>VDec: using MPEG1/2 Motion
Compensation and IDCT as output csp (no 0)<BR>Movie-Aspect is 1.33:1 -
prescaling to correct movie aspect.<BR>VO: [xvmc] 720x480 => 720x540 MPEG1/2
Motion Compensation and IDCT [fs]<BR>vo_xvmc: Port 292 grabed<BR>vo_xvmc: Found
matching surface with id=54434449 on 292 port at 0<BR>adapter<BR>FATAL: Cannot
initialize video driver.<BR>trying pixfmt=1<BR>Only buffers allocated by vo_xvmc
allowed<BR>mplayer: vd_ffmpeg.c:942: mc_get_buffer: Assertion `0'
failed.<BR><BR>MPlayer interrupted by signal 6 in module: decode_video<BR>-
MPlayer crashed. This shouldn't happen.<BR>It can be a bug in the MPlayer code
_or_ in your drivers _or_ in your<BR>gcc version. If you think it's MPlayer's
fault, please read<BR>DOCS/HTML/en/bugreports.html and follow the instructions
there. We<BR>can't and<BR>won't help unless you provide this information when
reporting a<BR>possible bug.<BR>alsa-uninit: pcm closed<BR><BR><BR>It will
indeed do this with 2 instances, but it takes longer. It
happens<BR>quicker with larger more resolute video files. All this leads
me to believe<BR>it's a memory leak.<BR><BR>I have upgraded from xorg-x11-6.7 to
6.8.0 and then to 6.8.2 and I've added<BR>the line<BR>Option
"XaaNoOffscreenPixmaps"<BR>to each of the 3 device sections in xorg.conf.
(this seems to have improved<BR>CPU usage DRASTICALLY, and I believe it was only
taking about half an hour<BR>to crash before, but eventually, they still crash),
per this bugzilla discussion:</DIV>
<DIV><FONT face=Arial size=2></FONT> </DIV>
<DIV><A
href="https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=126307">https://bugzilla.redhat.com/bugzilla/show_bug.cgi?id=126307</A></DIV>
<DIV><FONT face=Arial size=2></FONT> </DIV>
<DIV>This and this mailing list archive are the closest I've found to someone
having the same problem:</DIV>
<DIV><BR><A
href="http://www.newsarch.com/archive/mailinglist/mplayer/user/msg10614.html">http://www.newsarch.com/archive/mailinglist/mplayer/user/msg10614.html</A><BR><BR>In
the bugzilla discussion, you can see that the problem is
actually<BR>different. Instead of "only buffers allocated by vo_xvmc
allowed", their<BR>message is "BadAlloc (insufficient resources for operation)",
and presumably<BR>their problem happens right away, whereas ours takes an hour
of looping.<BR><BR>In the mailing list archive, it appears that he has a closer
problem. In<BR>all of my research thusfar, I've found only claims that the
bug is in xorg<BR>and that it's solved.<BR><BR>I am running the following (taken
from some of my
documentation):<BR><BR>HARDWARE:<BR><BR>CPU:
Pentium 4 2.8 GhZ<BR>BIOS: AMBIOS
.08.00.9<BR> Settings: USB
disabled<BR>
On-board audio
disabled<BR>
ACPI enabled<BR>RAM: 512MB 333 MhZ
DRAM<BR>Harddisk: ST380013AS SATA 80
GB<BR>Audio: M-Audio Delta
1010<BR>Video: NVIDIA GForce FX 5200s (3
cards)<BR>Periphs: Uknown serial exchange card of some
kind<BR> 3.5"
Floppy<BR>
ATAPI CDROM
Drive<BR><BR>SOFTWARE:<BR><BR>OS:
Linux i686
2.4.26-gentoo-r9<BR>
Grub Boot Manager<BR>X-Window Manager:
xorg-x11-6.8.2-r1 (gentoo ebuild)<BR>Window
Mangager: fvwm-2.5.10-r3 (gentoo
ebuild)<BR>Video Playback:
mplayer-1.0_pre4-r7 (gentoo ebuild)<BR>Audio Drivers/Tools:
alsa-driver-1.0.6a (gentoo
ebuild)<BR>
Alsa-utils-1.0.6 (gentoo
ebuild)<BR>
Alsa-tools-1.0.6 (gentoo ebuild)<BR><BR></DIV>
<DIV><FONT face=Arial size=2>Thanks in advance for anyone that has any
ideas!</FONT><BR></DIV>
<DIV><FONT face=Arial size=2>---------------------<BR>Aaron Trumm<BR><A
href="http://www.nquit.com">www.nquit.com</A><BR>-----------------------</FONT></DIV></BODY></HTML>