Annually, hpc.bw project organizes a variety of workshops aimed at supporting skill development among HSUper and ISCC users. Continuing this tradition, this year, the HSUper workshop took place on 23rd and 24th of April, at the Helmut Schmidt University/University of the Federal Armed Forces Hamburg (HSU/ UniBw H). This year’s workshop primarily focused on encouraging students and staff at HSU to start integrating HSUper resources in their respective research work. At the same time, experienced users who had already been using HSUper were also invited to participate. In line with the main objective of the workshop, the content was designed around setting a strong foundation about HPC and HSUper for new users. For example:
- How to login to HSUper, transfer data, manage folders and storage space
- How to interact with the system using the web interface or Linux-based terminal
- How to load existing software modules and install required software packages
- How to create new jobs, run them, and monitor progress
A total of 16 participants attended the workshop, including complete beginners interested in exploring HPC systems, students planning to utilize HPC resources in their upcoming projects and theses work as well as advanced users seeking support for specific issues related to their projects. To provide on-site support and guidance, six HSUper experts from the Chair of High Performance Computing at HSU also took part in the workshop on both days.
First day of the workshop began with an introductory session conducted by the Head of HPC Lab, Piet Jarmatz. This was an interactive session with the participants and introduced the fundamental concepts of high performance computing starting with the historical background and addressing the wide range of application areas. In addition, the participants were given an overview of the container based computing centre (CBRZ) familiarizing them with HSUper and introducing the soon expected upgrade, HSUpreme. This session concluded with an ice-breaking session among the participants. The second half of day one was focused on giving the participants hands-on experience in working with HSUper through the workshop tasks, which are available on the HPC portal (Training | HPC Portal). Following the introduction to accessing HSUper and navigating the Linux terminal, participants got to explore topics of data transfer, understanding the storage systems and the underlying hardware architecture towards the end of day one.
After a successful first day of the participants gaining the beginner knowledge on HSUper, the day two of the workshop transitioned towards more advanced topics through presentations delivered by HPC experts, Ruben Horn and Hauke Preuß. The sessions covered the topics of module system and Slurm job submissions along with job priorities, dependencies, checkpointing, and related concepts. These topics are highly relevant and essential for efficiently utilizing HPC resources, as they help users better understand Slurm messages, what could hinder a job from running and limitations set by HSUper.
organized but well-supported set up which allowed them to work through the tasks on their own pace, deepen their knowledge independently, and gain practical experience instead of passively following demonstrations. One of the most appreciated, and a new tool introduced in this year’s workshop was the workshop game. This allowed the participants to track their progress and compare it with the others while serving as a feedback mechanism for the organizers to identify topics that require additional discussion or support. This optional game was designed with a score for each completed workshop task and the participants mentioned this was highly motivating and encouraging them to actively engage with the workshop tasks.
