Re: G.729A (now fixed-point) decoder

On Wed, Mar 19, 2008 at 11:37:08PM +0600, Vladimir
Voroshilov wrote:
> On Wed, Mar 19, 2008 at 10:46 PM, Michael Niedermayer
<michaelnigmx.at> wrote:
> >  > Reference code uses ~((~a)>>b) for
negative numbers.
> >
> >  argh, right, L_shr(a,b) == a>>b
> >  iam silly but you are as well because
L_shr(-1,1)!=0
> 
> L_shr(-1,1) IS equal to zero and
> L_shr(-3,1) IS equal to -2
> in reference code.

That doesn't make sense, you claimed reference code uses
~((~a)>>b) for
negative numbers, for a == -1 that gives -1 and not 0.
So one of your statements must be wrong I think.
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-develmplayerhq.hu
https://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-devel

On Thu, Mar 20, 2008 at 12:10 AM, Reimar Döffinger
<Reimar.Doeffingerstud.uni-karlsruhe.de> wrote:
> On Wed, Mar 19, 2008 at 11:37:08PM +0600, Vladimir
Voroshilov wrote:
>  > On Wed, Mar 19, 2008 at 10:46 PM, Michael
Niedermayer <michaelnigmx.at> wrote:
>
> > >  > Reference code uses ~((~a)>>b)
for negative numbers.
>  > >
>  > >  argh, right, L_shr(a,b) == a>>b
>  > >  iam silly but you are as well because
L_shr(-1,1)!=0
>  >
>  > L_shr(-1,1) IS equal to zero and
>  > L_shr(-3,1) IS equal to -2
>  > in reference code.
>
>  That doesn't make sense, you claimed reference code
uses ~((~a)>>b) for
>  negative numbers, for a == -1 that gives -1 and not
0.
>  So one of your statements must be wrong I think.

You are right, my statement about L_shr(-1,1) was wrong.

Anyway each +/-1 difference is accumulated from frame to
frame and gives
different result.

-- 
Regards,
Vladimir Voroshilov mailto:voroshilgmail.com
JID: voroshilgmail.com, voroshiljabber.ru
ICQ: 95587719
_______________________________________________
ffmpeg-devel mailing list
ffmpeg-develmplayerhq.hu
https://lists.mplayerhq.hu/mailman/listinfo/ffmpeg-devel