Workshop on Domain-Specific Languages for Performance-Portable Weather and Climate Models

Material of the workshop held together with NOAA in 2020. All presentations are available in the slides directory. Recordings of the sessions are available on request.

Setup Environment on Orion

1. ssh <user>@orion-login.hpc.msstate.edu (log on to Orion)
2. git clone https://github.com/VulcanClimateModeling/dsl_workshop.git workshop (check out the repository for the workshop)
3. Add the line if [ -f ~/workshop/setup/jupyter_env ] ; then source ~/workshop/setup/jupyter_env ; fi to the file ~/.bashrc just before the line if [ -z "$PS1" ]; then return; fi
4. Make sure current .bashrc is active by typing exec bash
5. Go to workshop directory with cd workshop
6. module purge (remove any previously loaded modules)
7. source ./setup/module_env (load module environemnt for workshop)
8. Install GT4Py and setup Python virtual environment as Jupyter kernel with ./setup/setup_venv

Note: If you ever log on to Orion to work interactively in a terminal, it's important you always repeat steps 6. and 7. in order to have the correct module environment loaded.

Important: At the end of the workshop you will want to remove the modification to your ~/.bashrc again if you continue to use the Orion system.

Setup Environment in Docker

1. run setup/docker_setup to build this image (slow)
2. run the run_docker.sh script to start up the notebook in a container, a link is provided for access

Connect to Jupyter on Orion

1. Point your browser to https://orion-ood.hpc.msstate.edu/
2. Login
3. Click "Jupyter Notebook" under the "Interactive Apps" dropdown menu
4. Configure your session (Account Name = "gfdlhires", Partition Name = "orion", QOS = "batch", Resource Reservation = "vulcan") and launch you session. See image below.
5. Once the session has started, click "Connect to Jupyter"
6. The material for the workshop is under workshop/notebooks
7. Check you setup by executing the Check-Setup.ipynb notebook

Jupyter Notebook App Setup