Faster rendering to screen

[dhaude physnet uni-hamburg de]

Hello people,

I have an application that requires playing a number (100 or so)
of PAL-sized images (768x576px) on an X-Window system at 25fps.
At the moment I'm keeping those frames as a set of GdkPixbufs and
play them out via gdk_pixbuf_render_to_drawable()  to a
GtkDrawingArea. The rendering function call takes more than 50
msec each time, limiting the playback rate to less than 20fps.

This is a 700MHz PIII with a Matrox G400 an 512MB of RAM running
xserver-xfree86.

Curiously, I remember the playback being faster with
xserver-svga, but after downgrading again I don't see a
difference.

Is there any rather simple way to get the data on-screen
sufficiently fast? Like rearranging the pixbuf data in some
fashion and then dump them out to X with a more low-level
function? What approach do the mplayer people employ (just
curious; I don't need anything THAT fancy!)?

Thanks for any tips,

--Daniel