Re: performance problems double buffering to 1600x1200 window

[Robert Gibbs <gibbsrc gmail com>]

No.  Doing my own double buffering gives me a 5 to 10X speedup when I have an NVidia card (any type) and am using the Nvidia closed source driver.  I am using the wall clock timer from my original example program (not g_timer) so my numbers are not scaled the same as yours.

One computer I tried was a 2 year old Dell M90 Precision laptop running Fedora 11.

I saw about 40 FPS.

After running the test, I noticed that it was using the NV driver.  I installed the Nvidia driver and retested.  I saw 400 FPS. I made no other changes except to go to run level 3 and back to install the driver.

Another computer was a 5 year old IBM T41 laptop with ATI video running Ubuntu 9.10 using the open source driver.  I saw 10 FPS (unacceptable).  I did not try the ATI drivers as I have had problems with them in the past.

Ideally, I would like to be able to update at at least 20-30 FPS (wall clock time) even on old hardware like the IBM above.  I am sure I could do this with OpenGL, but rather than try that right now, I am going to experiment some more by making Xlib calls directly.

Bob

On Wed, Feb 3, 2010 at 6:17 AM, <jcupitt gmail com> wrote:

On 2 February 2010 03:14, Robert Gibbs <gibbsrc gmail com> wrote:
> Conclusion: self implemented GdkPixmap helps only when the X server can use
> certain optimizations from the graphics card.

Or perhaps that's backwards: doing your own double-buffering can help
if your X server does not support fast pixmap allocate/deallocate. It
looks like this problem only happens with some older nvidia cards (is
that right?)

John