Crunching on graphics card while using integrated for video

Me and several other people on mersenneforum.org, now that GPU computing is becoming more mainstream, would like to be able to run CUDA programs on our graphics cards and update our screens using integrated graphics cards to avoid slow screen updates. Is there any way of doing this? No one there can work it out. The main thread of interest is mersenneforum.org/showthread.php?t=13892 (sorry can't link as I am a new user). I thought that posting on a hardware orientated forum might provide an answer as we are all amateurs in this field.

Referenced link


Hrm, this is a difficult one.

There are several factors that play into this-- the BIOS, for one, the OS and its HAL (Hardware Abstraction Level), the CUDA program itself.

If you guys can get a look at Vijay Pande's Folding@home programs, maybe you could get some insight into how he shares crunching with regular graphics use.

Otherwise, it would be a matter of rewriting the BIOS to use the internal graphics module on the motherboard for OS use, get the OS to use that resource while still being able to access the add-in GPU(s) (meaning the HAL must be configured for it) and then get CUDA to use that add-in.