Moving your HPC workloads to LUMI
A two-day course in Copenhagen and online on the specifics of LUMI: hardware, software, compiling and running jobs.
This course introduces the architecture and setup of LUMI, one of Europe’s most powerful supercomputers. It covers the hardware, compiling and managing software, and running jobs – it is about the specifics of LUMI, not a general introduction to HPC. The course is given by the LUMI User Support Team in Copenhagen, and you can join on site or online.
What you will learn
- Connecting to LUMI and moving data to and from it
- LUMI’s hardware, and compiling software for it
- Managing software with the module system and EasyBuild
- Submitting and managing Slurm jobs, including job arrays and GPU/CPU binding
- Spotting and avoiding I/O bottlenecks on the Lustre file system
- Python environments and containers on LUMI
Who should attend
Current and future LUMI users, and support staff from the LUMI consortium countries, who already have some HPC experience. You should know the Unix shell and how a cluster works (Slurm, login and compute nodes, SSH); C, Fortran or Python is highly recommended. The course is not focused on AI workloads, though some AI topics are included.
If you do not have a LUMI account, you get temporary access for the course exercises. Lunch on site is provided by DeiC.
More events