[mythtv] HD-3000 and changes from 18.1 to SVN

[Chris Dos]

>>Daniel Kristjansson wrote:
>>
>>>I'm not sure if a Sempron 64 1600+ can
>>>do software decoding of HDTV. If it can, you should use that.
>>
>>The Sempron 64 2600+ (1600 mhz) pushed to 2000 mhz still doesn't have 
>>enough horse power to to software decoding HD.  The processor is only 
>>running about 30% while using XvMC and playing HD in 18.1.  I also 
>>compiled the Debian source packages with athlon-xp and mmx options.
> 
> Hmm, what are your exact compile options? I have a P4 @2.8 Ghz and it
> has no problem with software decoding. I'm using:
>  "-march=pentium4 -O3 -fomit-frame-pointer"
> Also have you tried libmpeg decoding. Issac has an AMD box and says
> that is faster for him.

My compile options are: --cpu=athlon-xp --tune=athlon-xp --enable-mmx
Everything I've read shows that raw clock speed is whats necessary to do 
software decoding.  I think the 2.8 ghz is right on the edge of what is 
necessary.  Most people that I've read about recommend 3.2 ghz or better.  So 
that leaves pretty much Intel CPU's as only the absolute fastest AMD processers 
are at the 2.8+ ghz or higher.  I personally have issues with Intel's behavior 
and I don't like to support them by purchasing their products.

I've tried (selected the option) for libmpeg decoding and my processor still 
couldn't keep up.  I'll try it again to be sure.  What should I see in the 
frontend log that shows that libmpeg is working?

> 
>>>I think the A/V diverged code is somewhat broken,
>>>but that just amplifies the nVidia XvMC problem.
>>
>>I wonder why when I disabled the sound in MythTV HD played perfectly 
>>using XvMC.  That was very strange.
> 
> Nah, that isn't so strange. The A/V diverged code really messes things
> up. If instead of dropping frames you slow down the audio it also works
> ok. I've watched HDTV with XvMC on a P4 @ 1.8 Ghz this way. If you turn
> of audio there is no A/V diverging..

Well, based what you just said, I would say the A/V diverged code indeed broken.  :)

>>Actually, I think I remember reading that it was 
>>the fading out of the OSD that was messing up XvMC.
> 
> Well we disable OSD fading for XvMC to avoid that problem. It looks
> like there have been some changes since I was last in there, the 
> defines at the top are no longer applied, but it looks like the
> same functionality is in the "xvmcBuffers" structure. If you want
> to be risky you can try playing with OSDNum and OSDResNum at
> lines 29 and 30 in videoout_xv.cpp. You can also make sure that
> the video output thread is running real-time.

I'll look at changing those settings and running it again.  But it looks like 
that's just going to change the OSD settings and not anything related to the 
diverge code.  Is that correct.  Right now the OSD is a lower priority.  Getting 
HD to work again for the wife's soap is the priority.  :)  Though I'm back to 
running the 18.1 so she's happy for right now.

>>Then I wonder if something else that has been changed.  Because 18.1 
>>plays great (after increasing the buffers) except for the OSD flutter. 
>>But the new SVN has the same problems that 18.1 did when I had the 
>>default settings on the buffers.
> 
> It is probably the OSD reservation, try decreasing it.

That may just address the OSD flutter problem, not the diverge problem.  Is that 
correct?  Increasing the buffers in 18.1 allowed HD to play back perfectly fine.

> 
>>I think I remember reading that the VIA XvMC is limited to something 
>>like 1024x768 resolution and it doesn't do native HD resolutions.
> 
> That is the original VIA, there has been a "Pro" for a while now.

Well now, thats quite excellent.  Are the drivers open source?  I don't like 
having to use a binary nVidia driver to get XvMC to run.  Are their any Pro 
cards that are fanless and around how much are they.

>>Still, it's strange that if I turned off the sound, MythTV would play HD 
>>fine but after I turned sound back on, diverged frames.
> 
> Yeah, I think that code may be broken. But it has always been so,
> it isn't a new problem.

Bummer.  Hopefully someone will take it on to fix this code.

> 
>>I think if we can possibly isolate the changes in the DVB code between 
>>18.1 and the current SVN, it will gives us some insight into what is 
>>occuring and make HD work for XvMC in the current SVN.  Though the 
>>differences might be so vast it might be a huge deal.
> 
> That might be useful. I don't think the changes have been vast,
> videoout_xv.cpp has sat pretty much untouched since 18.1.

Well, that is good news.  Hopefully it won't be very difficult it isolate then. 
  For those that want my debs that do work for HD they can find them at:

http://www.chrisdos.com/mythtv-18.1-9_debs.tar.bz2

They are based on the sources 18.1-4 debs from:

http://dijkstra.csh.rit.edu/~mdz/debian/dists/unstable/mythtv/

I've changed the version number to 18.1-9 so they won't get upgraded during a 
apt-get.  My debs are compiled for the Athlon-XP or better.

What is our next step in finding out what is going on so it can be fixed?  I can 
also provide the src to the debs that I posted above.

	Chris