[mythtv] MythTV 0.26, XVideo, and OpenGL

[Greg G.]

On 19.04.2012 21:47, Stuart Morgan wrote:
> We support vendor neutral APIs*  e.g. OpenGL. It's really up to the drivers to 
> support OpenGL and if they do that then it should just work with MythTV. We're 
> not talking about a proprietary protocol, or even something new since OpenGL 
> has been around for 20 years. We're not expecting high demand 3D processing at 
> 100 fps with AA, dynamic lighting effects and the rest, just very basic 2D 
> performance. It's shocking really that we were able to play Quake 2 et al 15 
> years ago on hardware less powerful than you'd find in many mobile devices, 
> yet even today you can't get decent 2D performance from enormously more 
> powerful hardware on linux.
> 
> Right now it seems only the proprietary Nvidia drivers work well enough. The 
> drivers for Intel and ATi hardware display underwhelming performance and 
> stability.

I just receveived some interesting feedback from an OSS DRI developer.
When reading the mythtv logs I always thought that the KMS blanking
method is the best to choose. Hower, it seems that for OpenGL output
there are special APIs which would be more approriate:

On 01.05.2012 09:49, Michel Dänzer wrote:
> On Fre, 2012-04-27 at 17:21 +0200, Greg G. wrote:
> > On 26.04.2012 07:58, Michel Dänzer wrote:
> >> What are the other options for 'blanking method' (I assume that's
> >> related to sync to vertical blank)? Do any of them work better?
> > Myth automatically chooses that mechanism, I can't change it. IIRC
> > if there is now KMS vblank it used to use an RTC timer. But surely
> > KMS blanking is the best way to do it.
>
> No, the corresponding GLX APIs would be better.

To me this sounds like a very promising option to explore. What do you
think?

Regards
Greg