DockerImage

Docker (development) image with everything set up to work with tomolab and NiftyRec

Info

• This image is based on Ubuntu 18.04 and integrates the development version of cuda toolkit (a smaller release version will be released in the future)
• The image is no greatly optimized. Once built, it needs around 5 GB of disk space, which more or less double one a new container is started from that image

Pre-requisites

• Docker image manager (the one you are more confortable with, depending on you OS: e.g. docker-ce)
• Nvidia Docker 2 (needed to access GPU hardware on the host side)
• Updated Nvidia drivers installed on the host computer

Quick start

• Clone this repository

• Make run.sh executable

• Make all scripts in folder scripts/ executable

• From within the main folder, fire up a terminal and execute sh run.sh or just ./run.sh

• Hopefully, the image will be built and a (self destructive) container will start right away.

• The process will start a jupyter server, which you may access from a browser of your host computer at:

  • http://localhost:8888 --> jupyer notebooks
  • http://localhost:8888/lab --> jupyter lab interface

• Be sure that no other process is blocking port 8888

• To stop the notebook server and destroy the container (the image will not be destroyed) just press Ctrl+C from within the terminal window