High-Performance Computing Workshop [Beginners and Intermediate Level]In-Person Event

2 Sept 2025, 09:00 → 3 Sept 2025, 16:30 Europe/Berlin

114 (CASUS)

Andreas Knuepfer

High-Performance Computing (HPC) clusters are used for large scale data processing and data analysis as well as highly parallel simulations of ever increasing fidelity. This course is meant to introduce the principles of how to work on HPC clusters in general and invites to collect first hands-on experience on the HZDR cluster. It covers all steps from connecting to the cluster to running parallel code it plus useful tips and tricks for convenient day to day work.

The workshop jointly organized by FWCI-H at HZDR in Dresden Rossendorf and the Scientific Compute Core (SCC) department at CASUS in Görlitz. It will take place in Görlitz and is open for attendees from academia, research institutions, and industry.

Tuesday 2 September

Welcome address and Introduction to the HZDR HPC cluster

Overview of the High Performance Computing cluster at HZDR and its architecture.

Accessing HEMERA and interactive jobs

Logging in to the cluster via the Secure Shell, running interactive jobs

Coffee Break

Batch jobs & module system

Working with the SLURM scheduler, using the module system to load software and libraries.

Lunch break

Job scripts and Array Jobs

Writing SLURM job scripts and running basic parameter scans. Bonus: Using AI chat bots to generate job scripts for you.

Coffee Break

Bonus: Log in without a password

Using SSH keys to log in without the need to enter your password.

Wednesday 3 September

Jupyter Notebook on HEMERA

Running your own Python environments and Jupyter notebooks with custom kernels.

Conveners: Franz Poeschel (CASUS/HZDR), Mani Lokamani (HZDR)

Coffee Break

MPI Jobs

Running parallel programs using MPI on Hemera.

Lunch break

Advanced SLURM

Scheduling interdependent jobs and using signals to communicate with running jobs.

Coffee Break

Overview about the new HZDR Cluster ROSI

Bonus: On-demand Session

Topic chosen by participant vote from one of: Tricks for the command line, TUD clusters, Remote VSCode integration

Convener: Andreas Knuepfer (HZDR)