Configuration Tesla K80 and Matrox on RHEL7.2 for CUDA and OpenGL with windows remote desktop and ss


I have the problem to configure Tesla K80 on redhat 7.2

The problem is in configuration

  • RedHad 7.2
  • Tesla K80
  • Matrox integrated into motherboard

The needs

  • Use K80 for
    o CUDA computing
    o OpenGL applications
  • To be able access Gnome and software GUIs via
    o Windows remote desktop
    o ssh –x X11 forwarding
  • To have good quality of rendering on the clients posts ( network is local and rapid)

The matrox hardware does not support OpenGL, but provides correctly

  • Windows remote sessions
  • X11 forwarding on the clients’ computers
    It does not compatible with nvidia drivers (segmentation fault errors) and I did not find the way to use it simultaneously with nvidia.
    So I have to uninstall nvidia driver, and thus no more cuda computing options on the server!

When install nvidia and switching Xorg to use nvidia driver

  • Windows remote sessions does not work
  • VNC sessions are of very bad quality (impossible to work)

Question :

  • Is it possible to find the way to get system running using Nvidia driver and simultaneously have system’s display using matrox hardware ?

Thank you for your help !

I have posted a similar question with no response yet.
Onboard GPU/8xK40m X server Failure\No Screen after CUDA7.5 install

In the installation guide for CUDA it discusses that you will have to do changes for non-trivial GPU configurations which some how this is believed to be. I find that this should be detailed from Nvidia more thoroughly because I have seen many people have this issue but have not been able to find a successful fix. I received a message from my vendor to buy a Nvidia GPU card instead of using the Matrox which I find unacceptable.

Have you had any luck on resolving this as well?