Moving Your HPC Workloads to LUMI (Hybrid)
9–10 November 2026, 09:00-17:30 CET
University of Copenhagen Panum and Mærsk Tower
Blegdamsvej 3B, 2200 Copenhagen, Denmark
This free two-day hybrid course introduces participants to LUMI’s architecture and working environment. It covers software compilation and management, data transfers, Slurm jobs, CPU and GPU binding, LUSTRE file-system performance, Python environments and containers. After completing the course, participants should be able to work effectively on both the LUMI-C CPU and LUMI-G GPU partitions.
The course is intended for current and prospective LUMI users and support staff who already understand basic HPC concepts. Participants need basic knowledge of the Unix shell and HPC cluster computing; familiarity with a programming language (C, Fortran, or Python) is highly recommended. Please note that this is an introduction to the specific features of LUMI, not a general introductory HPC course.
Participation is free of charge.
Registration closes on 26 October 2026.
View the program details and register
LUMI Profiling and Optimization Workshop (On-site)
11–13 November 2026
09:00-17:30 CET on Wednesday and Thursday; 09:00-12:30 CET on Friday
University of Copenhagen Panum and Mærsk Tower
This free, advanced-level, three-day workshop offers an in-depth introduction to profiling and optimising workloads on LUMI. Through expert-led presentations and practical exercises, participants will explore LUMI’s architecture, compiler and I/O optimisation, performance-analysis tools, Cray MPI on Slingshot, as well as specialised AMD tools for profiling GPU workloads.
The on-site workshop is intended for experienced HPC users with a solid understanding of LUMI who also have an active LUMI project or application. Participants should be comfortable with HPC cluster computing, Linux command line and have programming experience in C, Fortran or Python. They must also join the course-specific training project and create a LUMI account; instructions will be provided after registration.
Participation is free of charge.
Registration closes on 27 October 2026.
View the program details and register
Moving Your AI Training Jobs to LUMI: A Hands-On Workshop (Hybrid)
25–26 November 2026
09:00-17:00 EET on Wednesday; 09:00-16:30 EET on Thursday
University of Latvia, Riga, Latvia
This free two-day workshop is designed for users who want to move AI training workloads from laptops, workstations, or cloud environments to LUMI’s GPU platform. Participants will learn to use AI containers, distribute AI workloads across multiple GPUs within a single LUMI-G node, and gain insight into advanced topics for optimizing AI training processes on the LUMI supercomputer. In-person participants are encouraged to bring their own AI training scripts and will receive personalised support in adapting and running them on LUMI.
Participants should have basic Linux command-line skills and experience with Python, an AI framework such as PyTorch, TensorFlow or JAX, and experience with training an AI model on at least one GPU.
In-person attendees must bring their own laptop and charger. Online participation includes the lectures only; hands-on exercises and personalised support are available exclusively on site in Riga.
Participation is free of charge, but in-person places are limited to 25 and allocated on a first-come, first-served basis.
Registration closes on 29 October 2026.