Cuda with optimus gpgpu-only access

I have a Dell XPS L502 with an Nvidia 525M graphics card. At the moment, I'm only interested in using the capabilities of gpgpu on the map.

I installed Ubuntu 12.04 as a dual boot from the Windows 7 that came with the machine and performed several installation procedures to install the CUDA driver and developer kit from Nvidia (many Ubuntu reinstallations). In all cases, the display drops to 640x480. Best of all, I can determine that this has something to do with Optimus and Linux technology. I tried Bumblebee to no avail.

I really don't need to use an NVidia card to control the display. Is there a way that I can simply install the NVidia drivers so that the program can use the CUDA features for the video card and I still get full resolution on the display?

+4
source share
2 answers

I had a similar problem with my Alienware M11xR2, and posted the solution on the NVIDIA forums. Unfortunately, at the moment the forums do not work, but basically the process is as follows:

  • Install Nvidia drivers, but when prompted to change the X11 configuration, select "No." This is because the Nvidia card cannot be used as a display device.
  • Install the CUDA SDK and run one of the root samples . I found this a necessary step. After that, you will be able to run additional CUDA programs as a regular user.

Hope this helps.

+1
source

With the new version of CUDA 5, there is an installation guide, there you have only one file that installs drivers, tools and sdk (even nvidia nsight). And another thing that caught my attention is that you also have optimus options during the installation process.

I also have Alienware M14x, and I understand your problem, but I also wanted the drivers to work for me, so I did not try too hard on this.

Perhaps you could give this attempt and comment on all of us.

Here you can find a candidate for the release of CUDA 5: CUDA 5

and here is the installation guide (maybe read this first): CUDA 5 Launch Launcher for Linux.

+1
source

All Articles