Xvideo performance on Radeon 7500 vs Intel 915
michel at tungstengraphics.com
Mon Dec 4 04:21:34 PST 2006
On Sun, 2006-12-03 at 21:52 +0100, Roland Scheidegger wrote:
> Ken Mandelberg wrote:
> > After:
> > mplayer: user: 124sec sys: 3secs real:198
> > Xorg: user: 34sec sys: 32secs
> > So unless there is a statistical anomaly here, it seems to have reduced
> > the user space cpu time of Xorg by about 30%.
> > Still it doesn't perform well enough to use except in a pinch because of
> > frame drops (about 30%) . I guess the hardware is just not up to the
> > task for HD, especially 1080i.
> > For comparison with my Intel 915-Pentium M (2GHZ) on the same clip
> > mplayer: user: 116sec sys: 3secs real:249
> > Xorg: user: 2sec sys: 0secs
> I think the dropped X cpu usage is about right for the patch, but there
> still seems to be some problem. While a p4-m might be borderline fast
> enough, and you can't expect X cpu time to be as low as on the i915, x
> cpu usage looks quite high (if using dri (dma for xv)). You didn't say
> if you tried increasing agp mode, though it shouldn't really make a
> difference. Especially x.org system cpu time looks high, what is it
> waiting for there.
sysprof or oprofile should tell, but my first guess would be for
indirect buffers for the upload.
Earthling Michel Dänzer | http://tungstengraphics.com
Libre software enthusiast | Debian, X and DRI developer
More information about the xorg