I have recently finally succeeded in installing nvidia drivers (367.44) CUDA (7.5) and cuDNN (5.0) on Ubuntu 16.04. Everything worked fine, I could successfully test the setup in Theano python package as described here http://deeplearning.net/software/theano/install_ubuntu.html#test-gpu-configuration, the GPU was correctly detected and used. Now I wanted to install GPU version of Tensorflow and to my surprise it said that GPU cannot be detected. I wanted to check the output of nvidia-smi command, but it could not be found either. I can only see nvidia-detector and nvidia-settings. When I ran nvidia-smi directly from its directory in /usr/bin/nvidia-367/bin it said:

NVIDIA-SMI has failed because it couldn't communicate with the NVIDIA driver. Make sure that the latest NVIDIA driver is installed and running.

It seems like the driver was somehow disabled and I have no idea how to make it work again. Do you know what could happen and what it can be fixed?

  • 'lsmod' will tell you what drivers are loaded. – WinEunuuchs2Unix Sep 27 '16 at 12:21
  • I reinstalled the driver and now it works again. I will observe if the issue comes back. – makkam121 Sep 27 '16 at 15:22
  • Within the last 48 hours another user was complaining that automatic updates were breaking Nvidia drivers.: [askubuntu.com/questions/829468/… – WinEunuuchs2Unix Sep 28 '16 at 1:24

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.