The 4th VSC Users Day was held at the "Paleis der Academiën", the seat of the "Royal Flemish Academy of Belgium for Science and the Arts", in the Hertogstraat 1, 1000 Brussels, on May 22, 2018.
The titles in the program link to slides or abstracts of the presentations.
- 9u50 : Welcome
- 10u00: “Ultrascalable algorithms for complex flows” – Ulrich Rüde, CERFACS and Universitaet Erlangen-Nuernberg
- 11u00: Coffee break
- 11u25: Workshop sessions part 1 – VSC staff
- 12u15: 1 minute poster presentations
- 12u45: Lunch
- 13h30: Poster session
- 14u15: “Processing Genomics Data: High Performance Computing meets Big Data” – Jan Fostier, UGent
- 14u45: Workshop sessions part 2 – VSC staff
- 15u35: Coffee break
- 15u55: “Why HPC and artificial intelligence engineering go hand in hand” – Joris Coddé, CTO Diabatix
- 16u15: Tier-1 supercomputing platform as a service –
- 16u35: Poster award and closing
- 16u45: Reception
- 18u: end
Abstracts of workshops
VSC for starters
The workshop provides a smooth introduction to supercomputing for new users. Starting from common concepts in personal computing the similarities and differences with supercomputing are highlighted and some essential terminology is introduced. It is explained what users can expect from supercomputing and what not, as well as what is expected from them as users.
Start to GPU
GPU’s have become an important resource of computational power. For some workloads they are extremely suited eg. Machine learning frameworks, but also applications vendors are providing more and more support. So it is important to keep track of things happening in your research field. This workshop will provide you with an overview of available GPU power within VSC and will give you guidelines how you can start using it.
All code contains bugs, and that is frustrating. Trying to identify and eliminate them is tedious work. The extra complexity in parallel code makes this even harder. However, using coding best practices can reduce the number of bugs in your code considerably, and using the right tools for debugging parallel code will simplify and streamline the process of fixing your code. Familiarizing yourself with best practices will give you an excellent return on investment.
Performance is a key concern in HPC (High Performance Computing). As a developer, but also as an application user you have to be aware of the impact of modern computer architecture on the efficiency of you code. Profilers can help you identify performance hotspots so that you can improve the performance of your code systematically. Profilers can also help you to find the limiting factors when you run an application, so that you can improve your workflow to try and overcome those as much as possible.
Paying attention to efficiency will allow you to scale your research to higher accuracy and fidelity.