[MPlayer-dev-eng] [PATCH] x86_64 mmx/sse/3dnow optimisation support

D Richard Felker III dalias at aerifal.cx
Fri Oct 22 02:02:24 CEST 2004


On Fri, Oct 22, 2004 at 01:38:47AM +0200, Aurelien Jacobs wrote:
> On Fri, 22 Oct 2004 00:21:55 +0300
> Jan Knutar <jknutar at nic.fi> wrote:
> 
> > On Thursday 21 October 2004 14:58, Aurelien Jacobs wrote:
> > > On Sun, 17 Oct 2004 22:40:51 -0400
> > > D Richard Felker III <dalias at aerifal.cx> wrote:
> > > 
> > > > ok, this all sounds much better. i haven't read the patch yet but from
> > > > what you've said i think it's good so far.
> > > 
> > > Commited.
> > > Hope it won't break anything !
> > 
> > Perhaps related:  http://www.mplayerhq.hu/pipermail/mplayer-advusers/2004-October/001043.html
> 
> Unfortunatly it is related !
> Here is a patch that fix this crash. Unfortunatly, I don't have
> my amd64 system handy, so I won't be able to test it on amd64
> during the next 18 hours :-(
> It would be nice if someone on amd64 could test it (the
> important point is to test -vf scale).
> 
> I don't know why I modified this in the first time. This
> probably mean at least 10l for me.

actually it means this for you:
http://mplayerhq.hu/~arpi/eu2004/mplayer/2560/p8151391-10000l.jpg

> Should I commit this before testing on amd64 or should I wait ?

you should do something to ensure that mplayer works on x86 asap!
if you're worried about breaking x86-64, disable asm on x86-64 in the
mean time.

btw you should run through ALL of your patches and make sure they
generate identical code on x86 as they did before you patched.
if any of the generated code is different, you need to run extensive
benchmarks to make sure you did not slow anything down!

rich




More information about the MPlayer-dev-eng mailing list