[FFmpeg-devel] [PATCH] G.729A (now fixed-point) decoder

Vladimir Voroshilov voroshil
Wed Mar 19 14:27:51 CET 2008


On Wed, Mar 19, 2008 at 6:59 PM, Michael Niedermayer <michaelni at gmx.at> wrote:
> On Wed, Mar 19, 2008 at 12:23:46PM +0600, Vladimir Voroshilov wrote:
>  > On Sun, Mar 16, 2008 at 1:02 AM, Diego Biurrun <diego at biurrun.de> wrote:
>  > > On Sun, Mar 16, 2008 at 12:24:58AM +0600, Vladimir Voroshilov wrote:
>  > >  >
>  > >  > Hopefully this version will be better.
>  > >
>  > >  build system part OK, some nits below
>  > >
>  >
>  > Just a reminder: are any issues except spelling ?
>  >
>  > Michael, did you look into this path (i want to collect issues for
>  > weekend work)  ?
>
>  this is a >90k patch and there are all the SOC qualification patches (
>  which have priority) so be patient.

NP, knowledge that patch was not overlooked is enough.

>  And yes there are issues except spelling, for example why is it not
>  binary identical? Is it either * bugs or * it would be slow/messy/complex?

Small quick example:
reference code uses L_shr() routine everywhere instead of ">>"
L_shr(-1)==0, L_shr(-3)==2

To get binary exact result i have to implement all of such routines
and call them everywhere.
In other words i should get all fixed-point math layer from reference
code, imho.

If this is ok, i can do it, of course.

And yes, i'm not 100% sure that my code does not contains bugs.


>  And yes iam happy its fixed point now :)

Programming in fixed point much harder for me :( (overflows, overflows,...)

-- 
Regards,
Vladimir Voroshilov mailto:voroshil at gmail.com
JID: voroshil at gmail.com, voroshil at jabber.ru
ICQ: 95587719




More information about the ffmpeg-devel mailing list