[FFmpeg-devel] r9017 breaks WMA decoding on Intel Macs

[Zuxy Meng]

Hi,

2007/5/31, Guillaume POIRIER <poirierg at gmail.com>:
> Hi,
>
> On 5/31/07, Trent Piepho <xyzzy at speakeasy.org> wrote:
> > On Thu, 31 May 2007, Guillaume POIRIER wrote:
> > > On 5/30/07, Michael Niedermayer <michaelni at gmx.at> wrote:
> > > > On Wed, May 30, 2007 at 02:07:19PM +0200, Guillaume POIRIER wrote:
> > > > > Ok, fine with me. Michael, do you think that the patch I posted
> > > > > earlier (100% based on Trent's, only fixing minor issues) should be
> > > > > applied?
> > > >
> > > > well, after actually reading the code ... the loops should be written
> > > > in asm not by using for() / while() this will make the code faster
> > > > and it will make the n+%m code naturally dissapear
> > >
> > > Well, after getting a hint from Apple guys, here's an updated patch
> > > that gets rid of all assembler warnings without touching the code as
> > > much as Trent had done.
> > >
> > > Now, maybe writing the loop in ASM ourselves will help, but at least,
> > > this patch doesn't harm and works on all supported platforms!
> >
> > This won't assemble correctly if any of the memory operands already have a
> > displacement. Loren explained why:
> > http://article.gmane.org/gmane.comp.video.ffmpeg.devel/51453
>
> I agree. This patch's main purpose is to show how to write this
> routine with proper syntax (Bear in mind that all assemblers produce a
> warning here. To a certain extend, we may consider that the reason it
> works with FSF AS is just by pure luck.)
>
> When I have the time, I shall benchmark both yours and my patch to see
> what are the speed figures.... But IMHO, it's a bit pointless, because
> whatever the speed figures may look like, we are comparing 1 solution
> that appears to work by luck, and another that is more reliable. Speed
> isn't what your patch is after.

Hmm I really miss Intel syntax here, which I guess would be something
like "MOVLPS XMM6, [EAX+2*EBX+imm]" so using "+16" at the end doesn't
hurt in any case. But gcc simply doesn't parse".intel_syntax" in
inline asm.

-- 
Zuxy
Beauty is truth,
While truth is beauty.
PGP KeyID: E8555ED6