[MPlayer-dev-eng] MS-ADPCM/Stereo Works

[Arpi]

Hi,

> On Friday 28 December 2001 16:00, Michael Niedermayer wrote:
> [...]
> > > or maybe try to port 3dnow code from libac3. i don't know how different
> > > it is, but for example, idct is a strict algo, shouldn't be different.
> > > (but you know it may be - walken's idct vs. your simple_idct :))
> >
> > well the bias is missing in ac3 at least i cant find it ...
> > ill TRY to port the imdct stuff ;)
> > lets hope i dont port the bugs too
> ok i ported some of the 3dnow code, the equinvalent c code from libac3 too 
> which does work but compare says the difference is larger than +/- 1
> the C code is allso 2% faster than the current C code in liba52
> do u want a patch or should i commit it?
i prefer a patch first, for testing. is it mmx or 3dnow ?
if we want pre2 today, we shouldn't commit new untested code now :(

> committing would require to copy srfftp.h & srfftp_3dnow.h -> liba52/
no prob


A'rpi / Astral & ESP-team

--
mailto:arpi at thot.banki.hu
http://esp-team.scene.hu