[mythtv-users] mpeg2 playback performance issue makes no sense

Michael J. Lynch mlynch at gcom.com
Tue Oct 12 12:54:35 UTC 2004


I ran into a strange problem that makes absolutely no sense to me and 
was hoping someone could shed some
light on it.  Here's what I did.  First I ripped a DVD and used mplayer 
(via "watch videos") to playback the
ripped VOB(s).  I checked CPU idle time and it was about 70% idle.

Just for grins and giggles I wanted to see how performance of Myth's 
internal playback compared to playing
a file using mplayer.  Since VOB, and NUV files are simply MPG files 
with non-standard extenstions, I figured
I could simply rename or copy one of the VOB files to NUV (with proper 
basename format), inform Myth of the
NUV files existance, and play it back using Myth's internal playback.

First I used Myth to record something.  Then I deleted the resulting NUV 
file and renamed one of the VOB files
from my DVD rip to the file I had deleted.  This effectively informed 
myth of the existence of the recording.  I
probably could have used the rebuilddb script too, but this seemed to 
get the job done.  I went to "watch recordings"
and sure enough the VOB file was watchable under the name of what I had 
recorded.

Now this is where it get's weird.  The playback was skipping and 
generally acting like there was not enough
processor to do the playback, so I checked it out.  Sure enough, 
playback using Myth's internal playback was
pegging the processor.

First, is the test methodology I described valid?  If so, what could 
account for the large difference in processor
usage between "watch videos" and "watch recordings" playing the same file?

My system is 733 PIII with 384M, FC1, and Myth 0.16.

-- 
Michael J. Lynch

What if the hokey pokey IS what it's all about -- author unknown




More information about the mythtv-users mailing list