[FFmpeg-devel] [PATCH] pointless #ifdeffery in mpegvideo_ppc.c

[Ivan Kalvachev]

2007/10/2, Reimar D?ffinger <Reimar.Doeffinger at stud.uni-karlsruhe.de>:
> Hello,
> On Tue, Oct 02, 2007 at 12:12:15PM +0200, Luca Barbato wrote:
> > Diego Biurrun wrote:
> > > On Tue, Oct 02, 2007 at 10:39:01AM +0200, Diego Biurrun wrote:
> > >> This patch removes some pointless #ifdefs from
> > >> libavcodec/ppc/mpegvideo_ppc.c and makes the file conditional on
> > >> HAVE_ALTIVEC.  It only contains AltiVec code anyway.  Alternatively, or
> > >> as a next step, it could be directly merged into mpegvideo_altivec.c.
> > >
> >
> > fine, I'm about to commit something more invasive about moving all
> > altivec bits in the same place...
> >
> > NOTICE: x86 people please do the same, gcc uses sse/mmx/whatelse if you
> > implicitly or explicitly enable -msse/mmx/and so on
>
> I do not know about ffmpeg, but MPlayer does not set any of these for a
> "generic" build, and they should not be.
> -maltivec is only set because otherwise the altivec stuff will not
> even build (and which is the thing that creates all the problems and
> some of us consider so horribly broken - support for explicit use of
> extensions should not be mangled into one option with automatic use).

I guess this is one more reason to avoid intrinsics usage.

I wonder why altivec code is written with intrinsics and not written
in assembler.
Is there some technical reason?