[mythtv] mythtv-vid g450 directfb -- Some sucess!!

[Daniel Kristjansson]

On Thu, 2007-08-02 at 12:50 -0400, Peter Hartmann wrote:
> Hey Daniel,
> Thanks for getting back to me. (and thanks for all your hard work) Are
> you saying that mythtv with directfb output needs more resources than
> df_xine and mplayer with  -vo dfbmga:crtc2  ?   Those are perfectly
> smooth.   Here's a typical usage from top:

MythTV's DirectFB support is not very optimized, so yes.
DirectFB is not actively maintained. I just make sure it
still compiles and shows a picture. No one has spent much
time on it.

> I think you may find this interesting...
> 
> 2007-08-02 12:37:48.860 AFD: Opened codec 0x869aed0, id(MPEG2VIDEO) type(Video)
> 2007-08-02 12:37:48.860 AFD: Opened codec 0x869b230, id(MP2) type(Audio)
> 2007-08-02 12:37:48.923 Opening OSS audio device '/dev/dsp'.
> 2007-08-02 12:37:48.972 === WE ARE GOING TO LAYERCALLBACK
> 2007-08-02 12:37:48.973 === Checking for Matrox CRTC2 Layer
> 2007-08-02 12:37:48.973   === Not the right one, trying next...
> 2007-08-02 12:37:48.973 === WE ARE GOING TO LAYERCALLBACK
> 2007-08-02 12:37:48.973 === Checking for Matrox CRTC2 Layer
> 2007-08-02 12:37:48.973   === Not the right one, trying next...
> 2007-08-02 12:37:48.973 === WE ARE GOING TO LAYERCALLBACK
> 2007-08-02 12:37:48.973 === Checking for Matrox CRTC2 Layer
> 2007-08-02 12:37:48.973   === Found CORRECT LAYER
> 2007-08-02 12:37:48.973   === Calling GetDisplayLayer...
> 2007-08-02 12:37:48.974     === DFB_OK, Cancelling Enumeration :)
> 2007-08-02 12:37:48.974     === Testconfiguration skipped

This looks ok. We checked a bunch of layers and found one we could
work with.

> 2007-08-02 12:37:49.007 CreateBuffers
> 2007-08-02 12:37:50.521 TV: Changing from None to WatchingLiveTV
> 2007-08-02 12:37:50.532 Using realtime priority.
> 2007-08-02 12:37:50.549 Video timing method: DRM
> 2007-08-02 12:37:53.487 NVP: Timed out waiting for free video buffers.
> 2007-08-02 12:37:53.897 NVP: prebuffering pause
<snip>

The prebuffering pauses may be due to a bug..but they could also
just be due to some bottleneck.

> I'm looking into the possibility of the timeout as the problem.

I doubt that, but then I don't know which timeout you are
referring to..

-- Daniel