[FFmpeg-devel] [PATCH] Optimization for add_8x8basis

Michael Niedermayer michaelni
Mon May 14 18:55:28 CEST 2007


Hi

On Mon, May 14, 2007 at 07:22:11PM +0800, Zuxy Meng wrote:
[...]
> > > -}
> > > +#undef SETW_ONE_MMX
> > > +#undef PMULHRW
> > > +#define SETW_ONE_MMX(x)
> > > +#define PMULHRW(x, y, s, o)\
> > > +    "pmulhrw " #s ", "#x "           \n\t"\
> > > +    "pmulhrw " #s ", "#y "           \n\t"
> > > +ADD_8X8BASIS(3dnow, 512, 0)
> > > +
> > > +#ifdef HAVE_SSSE3
> > > +#undef PMULHRW
> > > +#define PMULHRW(x, y, s, o)\
> > > +    "pmulhrsw " #s ", "#x "          \n\t"\
> > > +    "pmulhrsw " #s ", "#y "          \n\t"
> > > +ADD_8X8BASIS(ssse3, 512, -1)
> > > +#endif //HAVE_SSSE3
> >
> > this is messy, especially if its done with many functions
> > (i know that as i did it with some of the motion compensation code ...)
> >
> > i would suggest that the function is put in a file
> > and that #included 3 times with the different defines
> 
> Wouldn't a new file for just one function (or two, if try_8x8basis is
> added later) be considered a bit costly?

hmm, i dont see why ...
we could even make the qns stuff optional and omit that file then if the
user disables it during compile time ...

[...]

-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

No human being will ever know the Truth, for even if they happen to say it
by chance, they would not even known they had done so. -- Xenophanes
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: not available
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20070514/2ad51d3f/attachment.pgp>



More information about the ffmpeg-devel mailing list