How to avoid separate video display card?

I plan to attach four Tesla/ Fermi GPUs to four x16 slots on a motherboard in a Supermicro box. These are co-processors, not for display. I have been told that with four PCIe x16 slots occupied by the Fermis, Windows will not allow me to use the on-board Matrox video for display. It is said to need a separate video card in an actual PCIe slot (a left over: x4 lanes in an x16 slot). Is this true? Is there any work around?

If u r not gonna use the display heavily, why dont u use “RealVNC” to view the console? (CUDA and RDP had problems earlier. Not sure if that is resolved now.)

If u r not gonna use the display heavily, why dont u use “RealVNC” to view the console? (CUDA and RDP had problems earlier. Not sure if that is resolved now.)

Thanks. I had planed to use Windows remote desktop to launch processes when the computer is running. I checked the RDP issue you described, and I think it is fixed:

http://forums.nvidia.com/index.php?showtopic=109260

If remote desktop does not work for some reason, then I would be pleased to try RealVNC. Thanks for the suggestion.

However, I would also like to have a directly attached display for some admin tasks such as, for example, building the operating system or looking at the screen during bootup. I can put a graphics card in temporarily for this purpose, but it takes a slot I had planned to use for something else (RAID and 10 Gb LAN). I was hoping that the simpler approach of using on-board graphics could be made to work.

Thanks. I had planed to use Windows remote desktop to launch processes when the computer is running. I checked the RDP issue you described, and I think it is fixed:

http://forums.nvidia.com/index.php?showtopic=109260

If remote desktop does not work for some reason, then I would be pleased to try RealVNC. Thanks for the suggestion.

However, I would also like to have a directly attached display for some admin tasks such as, for example, building the operating system or looking at the screen during bootup. I can put a graphics card in temporarily for this purpose, but it takes a slot I had planned to use for something else (RAID and 10 Gb LAN). I was hoping that the simpler approach of using on-board graphics could be made to work.

http://forums.nvidia.com/index.php?showtopic=159208

Use the TCC driver and you should be able to use your onboard VGA device alongside the Teslas.

http://forums.nvidia.com/index.php?showtopic=159208

Use the TCC driver and you should be able to use your onboard VGA device alongside the Teslas.