Hi!

I have a slackware machine with a Geforce GTX 460 graphics card.

Problem is that the X-server tends to use far too much memory of the available
1GB on that card to that sometimes the cuda applications cannot start.
(which generates a lot of "computation errors")

Is there a way to limit the the X-servers graphic memory usage?
I have looked at Nvidias xorg.conf documentation and have not found any settings for this,
but maybe I am missing something.

There is an onboard Nvidia 8300 on the motherboard with 256 MB of memory, but if I try
to run the GUI on that one and seti on the 460 the machine crashes after a while,
so the 8300 is now disabled.

Any ideas?

---

You will have to edit your /etc/X11/xorg.conf to only be setup for the onboard card, and have no mention of the other card in there. If both cards are the same you should be able specify the cards using the BusID line in the Section "Device" for the driver.

If your unsure of where to start, you can always start with a default auto-generated X config which will try and auto-load all drivers for detected cards and do a basic auto-configure for them. Run 'X -configure' from a root login if you want to go that route..

Make sure you keep a backup of your existing files though before modifying anything so you can always go back to something you know works!