[MPlayer-users] Question about -vo gl and nVidia

Dmitry Baryshkov lumag at qnc.ru
Thu May 15 09:20:26 CEST 2003


Hello!
В ?? 14.05.2003, в 20:00, Axel Liljencrantz пишет:
> [Automatic answer: RTFM (read DOCS, FAQ), also read DOCS/bugreports.html]
> 
> > With what card/driver(opensource or one from www.nvidia.com)?
> 
> Closed source binary, latest version. I need TV-out. I want 3d
> acceleration... Please don't hate me.
> 
> 
> > Are you using -vo gl or -vo gl2 ?
> > gl2 should be faster.
> > Also you could try using -vo gl:<num> , where num is number of
> lines
> > to be transferred at once. num <= 0 means full image. Default is 4.
> 
> Tried them both. gl2 seems slightly faster, but only by a few
> percent. gl:1 is really slow, but turning it up beyond 4 seems to
> have very little effect on performance.
> 
> Played around a little bit, and whith all parameters optimally
> tweaked xv uses ~25% CPU, gl2 ~65% and gl ~70%. I've got cycles to
> burn, but the problem is that sometimes the usage for both gl-mode
> peek and mplayer is forced to skip frames. Which really shows.
> 
> BTW, why is it impossible (or at least hard) to do update on vsync
> with the xv hardware? Isn't there a XSync command in X that can be
> used? And if it's not good enough somehow, doesn't graphics hardware
> generate interrupts on VSync which could be caught and used?
I also have old (Vanta) card, and use nvidia driver, but don't have
any 'tearing' with such configuration. Suppose, it's not a problem
of VSync.
Some more questions:
1) Do you have such problem with all movies?
2) Are you using CRT-monitor, or something like TV/etc. for viedeo
    output?
3) Is the problematic video interlaced/hard-telecined? (try playing
   with something like -vf pp=lb or with -vf ivtc )
4) Are you using framedropping/hardframedropping by default? Try using
   mplayer -noframedrop -nohardframedrop <all other options you pass>
> 
> thank you for your time
> 
> /Axel Liljencrantz
-- 
Dmitry Baryshkov <lumag at qnc.ru>



More information about the MPlayer-users mailing list