Lecture
Monday, 26 March 2018
11:45am - 12:30pm
Campus Stuttgart-Vaihingen, Pfaffenwaldring 47
Room V 47.03
Citius, altius, fortius!
What makes HPC actually “high-performing”? Much more than the large-scale computational problems we are facing today and the resulting involvement of state-of-the-art large-scale systems typically mentioned in that context, it is the permanent pursuit of efficiency at all levels to obtain the high performance desired. But what exactly is performance, and how is efficiency defined, given the completely different meanings that exist of both notions? And what is necessary to sustainably ensure that efficiency – including hardware-conscious programming and code tuning, but going far beyond that? Which lessons from HPC related to both algorithm and performance engineering can and should be learnt and adopted also in “normal” programming contexts, outside the high-end systems; and which are the challenges specific to those high-end systems at the eve of exa-scale? Finally, last but not least, how can those contents be taught suitably in an educational context? The talk will address and discuss these questions.
Biosketch
Hans-Joachim Bungartz is a Professor of Informatics and Mathematics at TUM and holds the Scientific Computing chair in TUM’s Informatics Department. After having earned degrees in Mathematics and Informatics from TUM, he became Associate Professor of Mathematics at University of Augsburg, then Full Professor of Informatics at University of Stuttgart, and returned to TUM in 2005. Since 2013, he has been both Dean of Informatics and Graduate Dean, with responsibility of doctoral education TUM-wide.
Dr. Bungartz has served on various editorial, advisory, or review boards. In 2011, he became chairman of the DFN. Furthermore, he is a board member of Leibniz Supercomputing Center. In 2016, he joined the steering committee of the Council for Doctoral Education of the European University Association.
His research interests are where Scientific Computing, CSE, and HPC meet. This includes parallel computing, hardware-aware numerics, high-dimensional problems, and aspects of HPC software, with applications such as fluid dynamics or plasma physics. Most of his projects are interdisciplinary ones – e.g., he is one of the coordinators of DFG’s Priority Program SPPEXA.
