[Open-graphics] decoding video in real-time

[Nicolas Boulay]

What are the interest of having most of the decoding done in the graphic card ?

I don't see any interest beside the low power consumption but at a
great design cost to match all the codec.

But what's up if we could use open source codec ?

If the gpu include a (fast) cpu. You can compile FOSS codec and use it
directly on the board. You can imagine the same for a X server.

I don't know the size of ARM core but maybe a cpu as the new Cortex A8
could do the job ?

So you will have a cpu in the graphic card to upload all the graphical
stuff (video, xserver, ...).

Generic cpu must be used to avoid the rewrite of all the codec.

Is it realistic ?

The interest will be only on the low power, but that became even more
important (especially for laptop). If we need only high quality, only
the end of the pipeline could stay in the gpu (yuv conversion,
scaling, filtering, etc...)

Nicolas

2007/4/16, Nicolas Boulay <nicolas.boulay at gmail.com>:
> I miss this one :)
>
> Globaly asynchronous/localy synchronous are a design technic where
> each module use a standard synchonous logic, but connection between
> each block are asynchrone. This permit different clock speed and avoid
> all the nighmarre of having a clock at 1ns periode with 20ns of delay
> depending where you read the clock inside the clock tree. The bigger
> the design the worst the problem is.
>
> Most (all?) SoC design are done like this.
>
> So i didn't speak about asynchronous logic at all.
>
> 2007/3/24, Attila Kinali <attila at kinali.ch>:
> > On Sun, 18 Mar 2007 21:04:49 +0100
> > "Nicolas Boulay" <nicolas.boulay at gmail.com> wrote:
> >
> >
> > > I just speak about the fact that 4 Ghz logic is not impossible.
> > > Because 6 Ghz (or 130 ps of propagation time) still exist. Usualy the
> > > problem are in the clock propagation time, but globaly asynchronous
> > > design could work at that speed.
> >
> > Yes. But the industry still refrains from using any asynchron
> > design. Even GALS are only used in research, though they should
> > be quite straight forward to use.
> > (Ok, multiple cores are some very coars grained form of GALS,
> > but well...)
> >
> >                         Attila Kinali
> > --
> > Linux ist... wenn man einfache Dinge auch mit einer kryptischen
> > post-fix Sprache loesen kann
> >                         -- Daniel Hottinger
> >
>