

Moreover, I was not prompted for the driver installation only EULA acceptance and the directory location for cuda.Ĩ. Instead I ran the following commands: sudo. However, the sudo sh cuda_n did not work for me.
Install nvidia cuda toolkit ubuntu 16 install#
Note that you don’t want to install the drivers 375 that comes with the cuda-8.0. Note, I initially ran “sudo service lightdm stop” but ofr some reason, it caused my screen to go blank.ħ. In tty mode: cd ~/Downloads/nvidia_installers Keep the next commands handy in another machine since now you are in tty.Ħ. Once there type: Ctrl + Alt + F1, and login to your user. Nothing should have changed in loading up menu. Create the /etc/modprobe.d/nf file with the 2 following lines: blacklist nouveauĥ. I used synaptic and did a purge, AKA completely uninstall programs and configuration. It is better to start with a fresh OS install otherwise completely uninstall anything in the ubuntu repositories with nvidia-*. cuda_8.0.61_n -extract=~/Downloads/nvidia_installers ģ. Open up a terminal and extract the separate installers via: $ mkdir ~/Downloads/nvidia_installers
Install nvidia cuda toolkit ubuntu 16 64 Bit#
I have tested the 64 bit version but I think the 32 bit will work too if your machine is 32 bit.Ģ. Go to and download CUDA toolkit 8.0 for Ubuntu 16.04 and select runfile local as your installer type. The runfile documentation was not that helpful. Thus, you have to install with the runfile, to opt-out of installing the driver. Note that you can not just install CUDA 8.0 by apt-get because it comes with a driver version ( 375.26) that doesn’t support the GTX 1080 Ti. So here I repeat most of his instructions with minor changes that I made to make it work for my own setup. Install CUDA without the driverįor CUDA install without drive, Abhijeet Kislay’s post helped me a lot. Although his is for CUDA 7.5, I was able to make it work with CUDA 8.0 with some minor changes. I recommend that you start with a fresh install of Ubuntu. I thought to write about my experience here.

There are not many online resources for GTX 1080 Ti and a few that I found did not work for my setup. I had to reinstall ubuntu 2 times to make it work. Keras, Tensorflow and Theano Since the card (as of this writing) was relatively new, the process was a bit cumbersome. I used his post and other on line material to set up the NVIDIA’s new GTX 1080 Ti graphics card for use with CUDA-enabled machine learning libraries, e.g. Nelson Liu’s blog has an excellent post about installing GTX 1080 Ti Drivers with Cuda 8.0 on Ubuntu 16.04.
