[Ffmpeg-devel] maintainers list

Guillaume POIRIER poirierg
Tue Feb 7 13:27:52 CET 2006


On 2/7/06, Diego Biurrun <diego at biurrun.de> wrote:
> On Tue, Feb 07, 2006 at 03:25:08AM +0100, Alexander Strasser wrote:


>  Also I changed the platform to consist
> > of os/arch pairs (only one entry as example).
> I have doubts about the os/arch pair, there is not much sense in making
> distinctions too fine-grained there IMO.  The assembler code is the same
> after all.

Except that the Altivec optimizations are mostly intrinsics, therefore
final code generation is still left to the compiler. In theory, GCC
should do the same job on all platforms that it supports, however it's
not the case: Apple's GCC (the one shipped on OSX) is notoriously
better a generating good code from the intrinsics, and FSF's GCC (the
one on all other plateforms, like Linux or BSD) is notoriously
"inferior" in that area (some ppl may even say "broken"), which
explains why some Altivec optimisation have been working on OSX only.

Blame it on whoever you want (Apple for not contributing back to FSF's
GCC or FSF not integrating Apple's patches), it's a sad situation.

In any case, that explains why IMHO for PPC it's justified to tell
which platform are maintained (NB: if I'm not mistaken, even Amiga
uses PPC, and I doubt any maintainer can pretend that they can
maintain that platform).

Just because code is syntactically "valid" GNU C doesn't mean GCC can
always compile it.
  Steven Bosscher - 2005-01-01
MPlayer's doc isn't up-to-date. Visit my updated mirror here:

More information about the ffmpeg-devel mailing list