Kjetil Torgrim Homme
kjetilho at linpro.no
Tue Oct 9 13:14:27 CEST 2001
Andrew Suffield <asuffield at debian.org> writes:
> On Mon, Oct 08, 2001 at 10:38:51PM +0200, Gábor Lénárt wrote:
> > It does not worth to support! If you can select between functions
> > optimized for various CPUs it would take much more memory and a
> > bit more CPU time (function pointers). If you make dynamic
> > loadable "plugins" it would decrease performance about between 10
> > and 20 percent (afaik). The latter because on x86 relocation
> > information should be kept in a register (and shared objects are
> > this kind of thing ..).
> Uhh, this is complete nonsense. If written in any reasonable manner
> it will be no slower. Also, mplayer already links against half a
> dozen shared libraries, and dynamically loads dvd support at run
> time as necessary. You don't even need to use any shared code,
> that's just the neatest solution.
Unfortunately for you, that _is_ the current state of affairs in
mplayer. Please help fix it if it is important for you. Arpi and the
others have different immediate goals.
I'm currently looking into replacing libac3 with the newer a52lib.
This will not be without cost until performance enhancements from
mplayer's libac3 branch is ported back into a52lib, but I think it
will be worthwhile for both projects.
More information about the MPlayer-users