how to use the onboard VGA output together with CUDA 4.0?

Has anyone set up CUDA 4.0, so that the video of the screen will be sent to the onboard VGA output of
the motherboard, while using all PCI-E slots for Tesla cards?

If so, how did you do it?

I am trying this on a Tyan S7025 motherboard, and am having issues.
If I select the VGA output, then CUDA 4.0 gets disabled.

Would appreciate help.

You probably need to manually load the kernel driver and create the /dev entries for your cards. How to do so is covered in the “Linux Only” section of the toolkit release notes.