ESiWACE3 milestone M5.4: First online training course

Centre of Excellence in Simulation of Weather and Climate in Europe Phase 3 First online training course Milestone M5.4 Funded by the European Union. This work has received funding from the European High Performance Computing Joint Undertaking (JU) under grant agreement No 101093054. ESiWACE3 Milestone M5.4 Contact details Project Office: [email protected] Website: www.esiwace.eu YouTube:https://www.youtube.com/@esiwace880 X: https://X.com/esiwace LinkedIn: https://www.linkedin.com/company/esiwace3 ESiWACE is on Zenodo, the Open Access repository for scientific results https://zenodo.org/communities/esiwace Disclaimer: Funded by the European Union. Views and opinions expressed are however those of the author(s) only and do not necessarily reflect those of the European Union or the European High Performance Computing Joint Undertaking (JU). Neither the European Union nor the granting authority can be held responsible for them. 2 ESiWACE3 Milestone M5.4 Lead Beneficiary: Sveriges Meteorologiska och Hydrologiska Institut (SMHI), Nikki Brown Other contributing authors: Stichting Netherlands e-Science Center (NLeSC), Alessio Sclocco, Stijn Heldens Helsingin Yliopisto (UH), Victoria Sinclair Type: Dissemination level: PU (for public use) Delivery date: Due date month 36. Actual delivery date 11December 2025 Means of verifications: Short report about the course, indicating the uptake. Achieved: Yes If not achieved, indicate forecast achievement date: - Comments: - 3 ESiWACE3 Milestone M5.4 Introduction The first ESiWACE3 online training course was an interactive workshop on GPU optimisation using the ESiWACE3-developed tool Kernel Tuner, held over two half days on 12 and 13 September 2024. The event was organised in collaboration with NCC Austria as a contribution to the Training Sprint initiative which pairs EuroHPC centres of excellence with national competence centres to arrange training relevant to the EuroHPC community. ESiWACE3 provided the tutors and course material and conducted the training, while NCC Austria organised event registration, computing access through the Vienna Scientific Cluster and other practical arrangements. Both partners advertised the course on their respective websites12 and through their mailing lists and social media. Course content The course was prepared and taught by Alessio Sclocco and Stijn Heldens of the Netherlands eScience Centre. It comprised a mix of lectures and hands-on exercises covering the use of the Kernel Tuner tool and compute performance and energy efficiency more generally, and was aimed at more experienced programmers and modellers. A pedagogical review of the training materials was conducted in advance by Victoria Sinclair of the University of Helsinki, and her recommendations3 were reflected in the final versions used in the course4. 4 https://github.com/KernelTuner/kernel_tuner_tutorial/tree/master/slides/2024_VSC_ESiWACE3 https://github.com/KernelTuner/kernel_tuner_tutorial/tree/master/hands-on/esiwace3 3 https://drive.google.com/file/d/1cfEpCcFhf0Uxh3PplJiFIpLQ36vc_WjL/view 2 https://events.asc.ac.at/event/143/ 1 https://projects9.dkrz.de/news/news/online-training-event-201cgpu-optimisation-with-kernel-tuner201d 4 ESiWACE3 Milestone M5.4 Attendance 33 participants attended the course, of whom just under half (15) were from institutions based in Austria; the remainder represented a broad range of Euro HPC member countries including Italy, Bulgaria, Germany, Cyprus, Sweden, Spain, Iceland, Ireland, the Netherlands and Slovenia. 22 of the participants gave their affiliation type as academic/university, a further six from research institutions that were at least partly publicly funded, three from other public sector, one from an SME and one from “other”. The positions held by the participants reflected the advertised target audience well. 18 were either research & development or research & teaching staff. Three were sysadmin and support staff, one a project manager, seven PhD students, 2 MSc students and two “other”. Gender balance was less positive: over two thirds (25) of participants were male. Five were female, one non-binary/non-conforming and two preferred not to give their gender. This imbalance probably reflects the current poor representation of non-male scientists in roles at a more experienced level in HPC. The scientific domains represented were varied. Just under half of participants (15) worked with mathematics and computer sciences. Other domains of interest included biochemistry, bioinformatics & life sciences; chemical sciences & materials; earth system sciences; engineering; fundamental constituents of matter and universe sciences. From this perspective, therefore, the collaboration with NCC Austria and the advertising network that this allowed access to clearly helped to extend the reach of the training offer beyond ESiWACE3’s core target domain of weather and climate modelling. Feedback 18 participants responded to a request to provide feedback, and all but one of these gave positive ratings (seven gave 10 out of 10, three 9, five 8, two 7) The one negative respondent - rating 4 out of 10 - found the course too superficial and too fast, with too few hands-on exercises, but the highly positive ratings from the majority suggest that the prerequisites for the training were well advertised and the material and pace pitched at the right level. Respondents were overwhelmingly in favour of the online live (instructor-led) format, welcoming the opportunities to ask questions and access support throughout. Several also commented that the format allowed them to join the course from a location and with a terminal set-up that suited them. 5