[FFmpeg-devel] [PATCH] Some ARM VFP optimizations (vector_fmul, vector_fmul_reverse, float_to_int16)

[Michael Niedermayer]

On Mon, Apr 21, 2008 at 02:12:51AM +0300, Siarhei Siamashka wrote:
> On Monday 21 April 2008, Michael Niedermayer wrote:
> > On Sun, Apr 20, 2008 at 11:33:43PM +0300, Siarhei Siamashka wrote:
> > [...]
> >
> > > I also still see absolutely no problem with stack realignment code, it
> > > takes only a few lines in the source and DOES NOT expand to anything on
> > > modern systems which use EABI. So all the talks about the extra code in
> > > the inner loop do not make any sense. If you still want to screw up ARM
> > > legacy systems which can't upgrade their broken toolchains, why do you
> > > still care about gcc 2.95 support? I don't see the logic...
> >
> > Then you must be blind.
> > If i didnt care about old systems i would have approved your stack realign
> > because it, as you say does not affect new systems its #ifdefd out on them.
> 
> Sorry, but you proposed to break them completely:
> http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/2007-September/035420.html
> 
> Wasn't your statement meaning that if somebody's compiler is "broken" (does
> not support *nonstandard* stack alignment option), he is screwed and
> deserves it?

i think i assumed that gcc on arm does support maintaining stack alignment,
thus i assumed you could place the realign in avcodec_decode_video() instead
of the innermost loop.

[...]
-- 
Michael     GnuPG fingerprint: 9FF2128B147EF6730BADF133611EC787040B0FAB

I know you won't believe me, but the highest form of Human Excellence is
to question oneself and others. -- Socrates
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 189 bytes
Desc: Digital signature
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20080421/ff260957/attachment.pgp>