HPC MUSE2022 Workshop1

MUSE HPC 2022 Workshop #1

HPC Hands-On Labs

Hands-On Lab #1 - Logging into Open OnDemand

1. Login to ELSA's Open OnDemand web portal with your TCNJ login and password
   • Open browser and go to https://ondemand.hpc.tcnj.edu/
   • Create your home directory if necessary by following the on-screen instructions

Additional help for this tutorial can be found at Accessing the Cluster via Open OnDemand

Hands-On Lab #2 - Open a Terminal in Open OnDemand

1. Login to the Open OnDemand port if you're not already logged in
2. Using the Open OnDemand menu, go to Clusters and choose _ELSA OpenHPC Cluster Shell Access
3. Execute Linux commands at the terminal prompt, e.g. ls
4. When you are done with the terminal, type exit at the command prompt and close the browser tab.

Additional help for this tutorial can be found at Accessing the Cluster via SSH using Open OnDemand

Hands-On Lab #3

1. Open a terminal shell
2. Add the ELSA tutorial module module add elsa-tutorial
3. Run the ELSA tutorial setup script elsa-tutorial-setup.sh
4. Change to elsa-tutorial directory cd elsa-tutorial
5. Look at the a submit scripts cat submit-serial.sh
6. Look at the other two submit scripts in the tutorial directory using either cat or the Open OnDemand file browser
7. Edit the three submit scripts to change the email address to your email address. Use either nano or the Open OnDemand edit file option in the file browser
8. Run the serial job sbatch submit-serial.sh (this one can take 2-3 minutes to complete)
9. Run the other two jobs (these run much faster)

Additional help for this tutorial can be found at Submitting Your First HPC Job

Hands-On Lab #4

1. Download the zip file from
2. Unpack the zip on your local machine
3. Make a new folder on in your ELSA cluster home directory (e.g. called workshop1)
4. Upload at least the mpi-pi-numinteg program (optionally upload the other files) to the folder you created above
5. Copy the submit-mpi.sh script from the elsa-tutorial folder
6. Edit the copy making any necessary changes
   • Change the command to run from mdart 50000 10000 to mpi-pi-numinteg
   • Use 2 nodes with 16 ntasks
7. Run the job
8. Review the job.jobnumber.out file