High Performance Computing

 

Prof. Dr. Matthias S. Müller holds the Chair for High Performance Computing (HPC) / Computer Science 12 in the department of computer science and is director of the IT Center. The research focus of the HPC group is on parallel programming models and tools in the area of high-performance computing. This includes performance evaluation and error analysis of parallel programs and architectures, as well as research in energy efficiency, programmability and total cost calculations.

In the context of parallel programming models, the focus of the HPC group is on the (further) development of the industry standard OpenMP. For that, the HPC group is member in the OpenMP Language Committee since 2006. In the field of correcteness analysis of parallel programs, the tool MUST is developed for automatic error analysis.

In addition, approaches to assess and improve the productivity of using and programming HPC systems are investigated. In particular, the aspect of total ownership cost optimizations is examined and the benchmarks of the Standard Performance Evaluation Corporation (SPEC ) are further developed. Corresponding research results are continuously incorporated into the improvement of operation and efficiency of HPC systems at the IT Center of RWTH Aachen University.

Research activities at a glance

  • Parallel Programming Paradigms (OpenMP and others)
    • Affinity of data to processing units, tasking, optimization for NUMA architectures, parallelization in object-oriented programming, …
    • Member of the OpenMP Language Committee and Architecture Review Board
  • Correctness Checking
    • Support of the Message-Passing Interface (MPI)
    • Support of hybrid parallel programs (MPI+OpenMP, other paradigms)
    • Development and implementation of scalable algorithms
  • Total Cost of Ownership

    • Energy efficiency in the context of HPC

    • Programmability & productivity
    • Modelling and evaluation of the performance of HPC systems
  • Analysis of Parallel Computer Architectures
    • Member of the Standard Performance Evaluation Corporation (SPEC)
    • Focusing on large shared-memory systems
    • Support of accelerator technologies

As part of its research focus, the HPC group is involved in these third-party-funded projects.

Teaching

The courses of the HPC group of the Chair for Computer Science 12 cover the field of high-performance computing. In particular, programming models and tools for parallel and multi-core computers, as well as performance and correctness analysis of parallel programs are part of the course contents. In these areas, lectures, software labs and seminars are offered regularly. An overview of our course offerings can be found in the Academics section. In cooperation with the IT Center, tutorial and training sessions are held on a regular basis. Also in cooperation with the IT Center, the HPC group trains Mathematical Technical Software Developers (MATSE).

Collaborations

The HPC group is a member of JARA-HPC and the JARA-HPC cross-sectional group Parallel Efficiency. In this context, scientists and HPC users of RWTH Aachen University and Forschungszentrum Jülich are supported in performance analysis and optimization of their simulation applications. In addition, the tools required for this are continually maintained and further developed. In cooperation with the IT Center, methodical consultation and support of HPC users of RWTH Aachen University and the JARA-HPC partition, as well as the continuous further development of the HPC services is undertaken. An overview of the current members of the HPC group as part of the Chair for Computer Science 12 can be found on the corresponding team webpage.