Ivor Hewitt wrote:
On Monday 11 Apr 2005 22:23, Daniel Kristjansson wrote:

I've updated the xv/xvmc merge patch with the help of some VLD
debugging by Ivor Hewitt. Hopefully this version functions with
XvMC-VLD..

The patch is at:
 http://www.mrl.nyu.edu/~danielk/mythtv/xv-xvmc-merge-v7.tbz

I've also continued to clean-up the buffer init code, and I've
made the buffer management a little more aggressive. When
AGGRESSIVE_BUFFER_MANAGEMENT is defined (the default in this
patch) videoout_xv reclaims buffers as soon as their children
are done rendering, instead of waiting until the children are
done displaying. Also with AGGRESSIVE_BUFFER_MANAGEMENT we
don't sync the past and future frames in DrawSlice, which
means less time is spent holding their frame locks. Finally,
I've lowered XVMC_OSD_NUM from 3 to 2, which seems sufficient
for OSD buffering.


Daniel,

VLD now displays. Although the lower half of the image misrenders with mpeg artifacts - not corrupt stream artifacts though - its as if the lower part of the screen isn't getting newer mpeg slices to render than the top part has got and is trying to catch up.

Same with or without AGGRESSIVE_BUFFER_MANAGEMENT.



Two tiny changes I forgot to mention before:-

Need to include <map> in videobuffers.h

The log "CreateXvShmImages() shmget() failed:" in videoout_xv.cpp:981 references undefined, "errno" instead of "XJ_shm_infos[i].shmid"


Regards,



------------------------------------------------------------------------

_______________________________________________
mythtv-dev mailing list
[email protected]
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev
Hi,

I have just tried this patch on my setup (Via M10K, Fedora3).
It appears to work fine with me, (limited testing), I do not
see the MPEG curruption artifacts that Ivor is seeing.

A little point, "#include <errno.h>" needs to be included in videoout_xv.cpp for Fedora3.
Also, obviously, debug messages need removing :)


Unfortunately, it does not fix the AVSync problem I currently have
with Via XVMC VLD decoding when the input stream has more than one
audio stream. I thought it might have ..

Thanks for the work Daniel, I think this will be a good improvement.

Terry
_______________________________________________
mythtv-dev mailing list
[email protected]
http://mythtv.org/cgi-bin/mailman/listinfo/mythtv-dev

Reply via email to