Online This course provides a high-level view of the architecture and the usage modes of an HPC system. The applicability of HPC for a selected number of applications will be understood. The challenges of developing and running applications on such an HPC platform will be comprehended.
Objectifs
At the end of this course, the successful attendee will:
- Have knowledge about:
- High-Performance Computer systems are and their general architecture.
- The MeluXina system, and its capabilities.
- The problems that are suitable for HPC systems.
- Differences between computational problems, data analytics and artificial intelligence
- The resources needed to successfully use an HPC system.
- The HPC ecosystem in Luxembourg and the rest of Europe.
- and be able to:
- Estimate the cost of using an HPC system for a given problem.
- Estimate the return on investment.
- in order to:
- Make informed decisions about the use and/or procurement of HPC systems in their organisation.
Programme
- HPC, HPDA and AI in a nutshell, European and national context.
- What kind of ROI can be expected?
- Going back to the basics. Some familiarity with notions such as CPU or memory helps to understand the specificities of a supercomputer such as MeluXina.
- Diving in: Luxembourg’s supercomputer MeluXina and its environment High-Level architecture and use cases.
- Case studies:
- Perfect use cases of a Supercomputer’s capabilities and capacities: examples from computational physics, chemistry, engineering, materials science, life sciences, finance, space).
- Innovative usages: IoT, Edge computing, interface with cloud and on-prem infrastructure.
- Impact and challenges of developing and running HPC, HPDA & AI applications.
- Discussion about what has been the functioning mode of supercomputers over the past decades, including job scheduling, application parallelisation and optimisation.
- Impact and challenges of developing and running HPDA & AI applications.
- The increasing amount of data forces a convergence of Data Analytics and HPCS into what is called HPDA. Discussion on how to leverage a supercomputing infrastructure for data analytics and artificial intelligence workflows.
- Agile supercomputing: Next generation systems and their ecosystem.
- Interface with the cloud to make it more user friendly and foster innovative usages.
- How the ecosystem is evolving, heterogeneity of infrastructure and software.