I got a DELL XPS L501X (Optimus machine). Now I know that I wont be able to use the 420M GPU I have on the machine for X and graphics, due to the whole Optimus not supported on Linux thingie (shame on you NVIDIA for that External Image). However, I read a post here on the forum about utilizing a Tesla card along-side an ATI Radeon card, and that kindled a hope that it might be possible to use the 420M card for CUDA only, without loading the X driver, and allowing the integrated Intel GPU to utilize X.
You can even run CUDA programs on the same cuda device that is being used by X but… cuda-gdb will not work.
Using the intel graphics card for X and a CUDA card for computation only sounds like a good choice to me. Please bear in mind that I’ve never tried this setup so I don’t take responsibility if it doesn’t work or goes wrong!
I guess I’ll have to try setting it up like that to see if it’s even possible, as it seems no one has any insight on the matter.
I do have some hope that such a setup will be possible, since the live Ubuntu system I’ve booted, recognizes the existence of the 420M GPU, and offers I’d install the driver for it (honey trap sin’t it?).
It may be a good idea not to use Ubuntu’s nvidia drivers as they may be out of date. I’m not running Ubuntu at the moment so I can’t confirm this. If they are out of date then I would recommend using Nvidia’s latest drivers.
