[mythtv-users] cpu usage question / recording setting recommendations

Pierre-Olivier Bouchard petecool at vl.videotron.ca
Tue Jun 17 23:19:39 EDT 2003


Brian Lalor wrote:
> Well, I've finally got everything configured!  I'm very impressed, but 
> I'm having difficulties getting my recording settings down.  I've got an 
> Athlon XP 1700+, 512MB RAM, a 120GB Hitachi drive, Matrox G400 and a 
> WinTV-dbx.  I'm recording the audio via btaudio and using alsa for my 
> on-board via82xx sound card.  I initially used the default recording 
> settings (RTJPEG) which produced very good quality recording as well as 
> live TV, but *huge* honkin' files.  Now I've switched to MPEG4.  First I 
> tried the default settings, but on playback, the recording was jerky.  I 
> believe the load was also extremely high.  I didn't check system load 
> when playing that show.  Then I went to 1800 and found it better, but 
> CPU usage was extremely high (>95% used by mythbackend).  I'm now 
> recording something at 2000 and 320x480 resolution.  CPU usage for 
> mythbackend alone is about 90% (overall system load ATM of 
> 0.83/0.91/0.96).  I can't watch any recordings, however, because the 
> playback and sound are jerky.
> 
> I would have figured that my system would have no problem recording and 
> decoding an MPEG4 stream at the same time, but it seems that I'm having 
> difficulty just encoding to an MPEG4 stream at the default settings. 
> I've compiled Myth 0.9.1 on my box and I explicitly specified my 
> processor (--cpu=athlon).  Can someone point me in the right direction?

Something is VERY wrong with your system, I can do LiveTV at that 
resolution and bitrate setting, and I'm only using a Celeron 1.1A (Much 
much less powerful).

LiveTV might have problems if your video card Xv setup is wrong, it 
would then use a CPU intensive display mode... but if even a single 
recording won't work, then your problem is bigger than with mythtv... 
either a kernel or hardware problem, I guess.


Pete



More information about the mythtv-users mailing list