[mythtv] Anyone looking at the XvMC/OSD glitches?

Joe Votour joevph at yahoo.com
Sat Nov 27 21:34:07 UTC 2004

Also regarding XvMC, there seems to be some major
sound issues associated with it (at least in LiveTV).

If I use OSS, then I get the odd "prebuffering pause",
if I use ALSA, I get continuous "prebuffering pause"
messages.  I've verified this with two sound cards
(the onboard Via 82xx as well as a SoundBlaster Live)
on my Epia.  This is solved by turning XvMC off
(although at a major CPU and visual quality hit).

When using OSS, things are at their worst when in the
program guide, where moving the "cursor" will cause a
prebuffering pause.  Using ALSA doesn't seem to matter
whether I'm in the program guide or not.

This is using some RPMs of CVS (from 11/25), using
ALSA version 1.0.7-11 from ATrpms, and the latest
version (11/16) of the Unichrome drivers.

-- Joe

--- Daniel Thor Kristjansson <danielk at mrl.nyu.edu>

> On Wed, 24 Nov 2004, Stephen Hocking wrote:
> ]When playing  HDTV live, One can get reliable
> playback and non-flickery OSDs 
> ]by disabling XvMC. However, this makes the CPU
> requirements insanely high. Has 
> ]anyone looked at what it would take to plunk the
> OSD display into the data 
> ]being sent off by the XvMC code, rather than it
> being dropped on top of it, 
> ]causing all sorts of weird interactions?
> No one has looked into that. Are you volunteering?
> :)
> My guess is that you could create fixup data and
> insert it. But you 
> could probably only do this in I and P frames. For
> HDTV that would 
> probably be every 5 frames, giving you an OSD frame
> rate of 6 fps. You 
> would then need to zero the motion vectors near the
> OSD in both P and B 
> frames.
> It may be easier to just turn on frame skipping when
> the OSD is on 
> screen. There are some remnants of a frameskipping
> patch here:
> -- Daniel
> > _______________________________________________
> mythtv-dev mailing list
> mythtv-dev at mythtv.org

Do you Yahoo!? 
Yahoo! Mail - Helps protect you from nasty viruses. 

More information about the mythtv-dev mailing list