[MPlayer-dev-eng] Re: DECODING AHEAD - (Initialy Michael's idea)

Tobias Diedrich td at informatik.uni-hannover.de
Tue Feb 26 20:30:00 CET 2002


Arpi wrote:

> for I/P type videos without MC (like FLI, CVID) could use method 1 with
> static buffer or method 2 with double/triple buffering.

Umm... Is it not the other way around (method 2 requires single
buffering because only parts of the memory may be updated (skipped
macroblocks)) ?

It would be nice to have more low level access to the gfx card (vidix
maybe ?) so one could have the codec update a buffer in video memory
and let the gfx card to a dma transfer to a set of other buffers in video
memory for double/triple buffering. This would also allow for the gfx
card to do the color space conversion even if one is using method 2.

Modern gfx cards usually have really fast memory (for texture reads)
but the agp bus is the bottleneck. At least on my machine
(Celeron 433 => 66mhz bus => PC66 speed memory access, BX chipset => 2xAGP)
agp may be just as fast as or even faster then main memory. At least
if the framebuffer is correctly setup to be cached.

-- 
Tobias								PGP: 0x9AC7E0BC



More information about the MPlayer-dev-eng mailing list