[MPlayer-dev-eng] Re: Another SiS vidix issue

[Alban Bedel]

Hi Jake Page,

on Tue, 14 Oct 2003 11:09:25 -0700 (PDT) you wrote:

> On Tue, 14 Oct 2003, Thomas Winischhofer wrote:
> 
> > Marcus Thiese wrote:
>  
> >  > > I think the xvidix driver is a
> >  > > great way to test VIDIX development, but Xv is a better option
> >  > > for real playback, since it's better integreated with X11.
> >  >
> >  > But shouldn't the xvidix driver be faster? If it directly accesses
> >  > the graphics card it should and I'm always heading for speed :-)
> > 
> > And what, if not "directly accessing the graphics card", do you think 
> > the X Server does?
> 
> Exactly - as Thomas said, & I mentioned originally, the VIDIX driver was
> 
> based almost entirely on the Xv driver in the first place.  The goal
> (for me at least) was to be able to play video using the hardware
> overlay on the framebuffer withOUT X11.  When in X, use Xv, you'll get
> the same performance with much better integration...
That's not my experimented with a banshee. Using xv is the worst, then
come tdfxfb and finnaly tdfx_vid. However using tdfx_vid i have mostly
the same performence on the framebuffer and under X (on a k6-333).
Try a vidix/*_vid vs. xv on a box with a <= 500 MHz CPU and you'll
see that xv is just slooooow.
	Albeu
-- 

Everything is controlled by a small evil group
to which, unfortunately, no one we know belongs.