[MPlayer-dev-eng] ffmpeg directrender

[Arpi]

Hi,

> > Probably because B frames are direct rendered (and stride is 32-byte
> > aligned due to MGA hw limitation) while I/P frames aren't.
> >
> > Is there any good reason why stride cannot be changed between frames?
> yes, all the functions in ffmpeg get only 1 stride passed (its the same for 
> src & dst) (motion compensation functions for example) there would be quiet 
> alot of code which would need to be changed to support different strides, 
> there is gmc qpelmc halfpel mc and there are averaging version of some of 
> them then x86 asm & c & some alpha code ...

ah, ok.
anyway, it has no sense to do, as with rendering B frames to video ram, both
libmpeg2 and ffmpeg are slow as hell (slower than no DR).
i can't even imagine speed if we would put I/P to vram too :)


A'rpi / Astral & ESP-team

--
Developer of MPlayer, the Movie Player for Linux - http://www.MPlayerHQ.hu