nitro wrote, on 2007-05-04 17:36:
http://www.imgtec.com/PowerVR/products/Video/MVDA2/index.asp
The bad point of this kind of chip are the limited amount of supported
codecs, so this kind below would be better ; also because it's OpenGL|ES
2.0 compatible ;)
"[...] Video processing for free, with the real-time programmable
architecture providing extensive accelerated functions support for
multi-standard video decode and encode." -- ak vertex&fragment shaders
that seems to be extended in this chip to access other kind of resources
(maybe a kind of fast texture wrapper around raw video blocks ?).
http://www.imgtec.com/PowerVR/products/Graphics/SGX/index.asp?Page=2
Now I think the main problem would be the price of a chip like that.
Why not use an FPGA with a bunch of arithmetic operations widely used in
audio / video compression (eg. DCT) and write a media library that
forward most of the job on the FPGA. I don't know if there is more
complete solutions available, but the basic idea is here :
http://www.opencores.org/projects.cgi/web/video_systems/overview
(well I don't have the whole mailing list archived here, so it has maybe
been already mentioned before)
That sounds like a new project for the Open-Graphics Project:
http://www.opengraphics.org and mailing list available from
news.gmane.org as the newsgroup gmane.comp.graphics.opengraphics
Arthur.
_______________________________________________
OpenMoko community mailing list
[email protected]
http://lists.openmoko.org/mailman/listinfo/community
