Einführung in Modulares Supercomputing - WiSe 2024/25

Inhalte:

• Scalable problems for scalable computing systems
• From Accelerated to disaggregated supercomputing architectures
• Amdahl’s Law and generalizations
• From the Cluster-Booster Concept to a Modular Supercomputing Architecture (MSA)
• Resource Optimization by MSA
• The ParaStation Modular Software Architecture
• Comprehensive Software Environment (co-scheduling, resource management, etc.)
• Programming Models (inter-module MPI offloading, OmpSs abstraction layer, resiliency)
• Virtualization by Network Attached Accelerators
• Co-designing applications and workloads (e.g., neuroscience simulations, climate simulation, seismic imaging, data analytics in earth science)
• Hardware implementations and prototypes
• Exascale Supercomputing Technology
• Interactive Supercomputing
• Integrating Future Computing Technologies (Quantum Computers)

Lernergebnisse / Kompetenzziele:

The students will

• understand the implications of Amdahl’s and Gustafson’s Laws on scalability;
• learn the basics of the modular supercomputing architecture (MSA) from idea to production;
• use the theoretical formulation of MSA for resource optimization in HPC;
• understand MSA as a new paradigm for heterogeneous architectures in high performance computing (HPC);
• be able to apply the MSA programming models;
• understand the software environment needed to operate future supercomputing facilities;
• meet various hardware implementations and prototype platforms.