[XviD-devel] Question about bvop decoding

[Edouard Gomez]

skal (skal at planet-d.net) wrote:
> 	Now it's pretty obvious you're loosing
> 	time in Memory I/O and 16/8bits conversion
> 	during steps b) and c). These two could be
> 	merged into a single 'averaging' bwd predicting
> 	step.

Just before i start decrypting your assembly, to map the
function names to more explicit meaning, and then starting
hacking bf interpolate to use them, and of course implementing
a C backend set of functions... (i'll do tomorrow, cause i'm
pretty tired today, too much J2EE coding kills my coding
energy)

How much speedup can i expect doing this change ? Is it only a
few % or the several % i'm looking for to catch up with ffmpeg?

Btw, if this proves to be really helpful, i guess using such a
trick for qpel will be much harder to implement, won't it ?

-- 
Edouard Gomez