[Arm-netbook] Any ARM SoC has Open-Source access to hardware video decoder ?

[Michael Zucchi]

On 02/07/12 00:34, Alexey Eromenko wrote:
> While I agree that 3D is nice, the use-cases for it are limited.
> Much more needed is the VPU stuff (hardware video decoder), more than GPU (3D).
> Once any SoC will do VPU, it will become so much more useful...

What's this obsession with video decoding?  Sure it's a nice have, but 
really just how many silent hi-def home-theatre set top boxes does the 
(esp free software) world need?  Sure i'd like a 1080p XBMC box @ 2W of 
silence too, but there has to be more to all these low-power linux-arm 
efforts than just even more time spent in front of the idiot box ...

Accelerated 3d can do a bit more than just decode some video frames faster.

e.g.
  - interactive user interfaces
  - animation
  - games
  - accelerating 2d, text, curve rendering
  - real-time simulation display
  - highly parallel computation
  - freeing up the cpu to do useful work that is more likely to be 
required when doing those things: when playing a video there usually 
isn't much else required.

And it's not like the VPU which just improves the battery life or ups 
the decodable resolution/bit-rate, without 3d acceleration some things 
just aren't possible at all.  It's not like you can just transcode a 
game to a simpler format.

Once you're talking megapixel/HD resolutions, trying to create 
interesting and useful user interfaces whilst leaving accelerated opengl 
on the floor is a big handicap and you don't want the cpu tied up 
pushing so many pixels.  XBMC doesn't work (very well at all) without 
accelerated opengl for this very reason (it's not the only reason, but 
it helps).

This whole '3d acceleration is not important' line seems a bit of a 
short-sighted and blinkered view of how people can use computers.

All new user-facing hardware will have 3d acceleration and that they 
have it is reason enough to need it too.