This article might eventually become obsolete!

To enable GPU on a Linux Cameyo server you have to do the following:

  1. Stop the Cameyo server
  2. Make sure the server is at Level 8
  3. Do NOT enable GPU in the UI

  4. Locate the VM in GPU (using the Cloud ID on the server page)

  5. Edit the machine in GCP

  6. Go to Availability policies and set [On host maintenance] to [Terminate VM instance]
    Otherwise you won't be able to add a GPU

    if no GPU is available, this means that your server is in a region that doesn't support GPU, then please create on in an appropriate region (see: GCP-Cameyo Regions Mapping)

  8. Click ADD GPU and select an NVIDIA Tesla T4 with 1 GPU (and GRID enabled?)

  9. Save these settings
  10. Start the Cameyo server through the UI

  11. Open an Admin session
  12. Open a Terminal session

  13. Check if GPU is now enabled, enter the following commands::
    sudo apt-get install pciutils -y
    lspci | grep NVIDIA

    this installs the lspci tool and shows if the NVIDIA GPU is present
    if not, it returns nothing. In this case the GPU is not active and it makes no sense to continue. Re-check in GCP if the GPU is really enabled.

  14. Add nvidia-detect using the following commands::
    sudo apt-get install nvidia-detect -y

  15. Run nvidia-detect to see what drivers get recommended


    you then will see:

  16. Add the necessary repositories and install the NVIDIA drivers with the following commands:
    sudo apt install software-properties-common -y
    sudo apt-add-repository contrib
    sudo apt-add-repository non-free
    sudo apt-get update
    sudo apt install nvidia-smi -y
    sudo apt install nvidia-driver -y
    sudo apt-get install linux-headers-$(uname -r)

  17. Restart the server (you can also do via UI if you want):
    sudo reboot now
  18. Open an Admin session and start a Terminal session

  19. Run nvidia-smi to check if the GPU is detected and running with the right driver.
    You should now see: