[MPlayer-users] Bicubic scaling in vo_gl - does it ever work?

[Guillaume POIRIER]

Hi,

On 6/12/06, Vladimir Mosgalin <mosgalin at vm10124.spb.edu> wrote:
> Hi To MPlayer-users mailing list!
>
>  On 2006.06.12 at 21:51:20 +0400, Vladimir Mosgalin wrote next:
>
> >
> > I tried running mplayer under gdb, but the results are no good.
> > 0x00582bc0 in s1557 () from /usr/X11R6/lib/modules/dri/fglrx_dri.so
> > (gdb) where
> > #0  0x00582bc0 in s1557 () from /usr/X11R6/lib/modules/dri/fglrx_dri.so
> > #1  0x0057dab8 in s1533 () from /usr/X11R6/lib/modules/dri/fglrx_dri.so
> > #2  0x005770c0 in s1531 () from /usr/X11R6/lib/modules/dri/fglrx_dri.so
> > #3  0x00580028 in s1545 () from /usr/X11R6/lib/modules/dri/fglrx_dri.so
>
> OK, more interesting info. I recompiled mplayer in debug mode and found
> two things:
> 1) It doesn't hang. It just becomes really slowwwww.. Like 30 seconds
>    per frame or so.
> 2) It spends all of that time switching between functions with
>    meaningless names in fgrx_dri
>
> How come? My system can do over 10k glxgears! Is that scaling shader
> THAT slow?
> OK, joking. Can something be done to find out what exactly is causing a
> problem? Even software-emulated -vo gl is MUCH faster than that.

Don't know if it's relevant, but Reimar, who implemented it, has a
5700 card. It's possible that ATI cards react differently....

Anyway, an update has just been committed to svn, so why don't you try it out?

G
-- 
"Success consists of going from failure to failure without loss of enthusiasm."
 -- Winston Churchill