Up to now nvidia-docler2 is deprecated but it is possible to upgrade as explaned here, it mantains the retrocompatibility. Now you can install nvidia-docker2 sudo apt-get update sudo apt-get install nvidia-docker2 Upgrade nvidia-docker2 Install nvidia-docker ( ) curl -s -L | sudo apt-key add -Ĭreate file /etc/apt//nvidia-docker.list sudo nano /etc/apt//nvidia-docker.list > deb $(ARCH) / > deb $(ARCH) / > deb $(ARCH) / Setup docker without sudo sudo groupadd docker sudo gpasswd -a $USER docker sudo reboot Install docker-compose # install docker compose sudo curl -L “ $(uname -s)-$(uname -m)” -o /usr/local/bin/docker-compose sudo chmod +x /usr/local/bin/docker-compose Install nvidia-docker2 Install dockerīefore install docker-ce correct the file /etc/apt//additional-repositories.list sudo nano /etc/apt//additional-repositories.list > deb bionic stableĪdd the key and after intall docker-ce sudo apt-get update sudo apt-get install docker-ce This force the new driver to overwrite old function. I try to fix broken package with synaptic but it does not work so I solve in this way: cd /var/cache/apt/archives sudo apt install -o Dpkg::Options::='-force-overwrite' -f. When I update my cuda driver from 10.0 to 10.2 I have some problem, because some function should overwrite other old package function.

To update Cuda It is possible to use Drive manager to obtain the new driver version and after the package manager will find automatically the new Cuda version to update. wget sudo dpkg -i cuda-repo-ubuntu1804_10.0.130–1_b sudo apt-key adv -fetch-keys sudo apt-get update sudo apt-get install cuda Install cuda-toolkit sudo apt-get install nvidia-cuda-toolkit Update Cuda

Method 2: sudo apt-get install nvidia-driver-410 Install cudaĬuda 10 is not supported by tensorflow, if you want to use it without docker try to install 9.0 version (Or build it in your machine as shown here). Setup Linux Mint 19 Tara with nvidia cuda for machine learning Install nvidia driver

Install cuda linux mint rebevca