Computer Science Colloquium
Optimization of Scientific Applications in High Performance ComputingThu 28.05.2015, 17:15, 60 minutes
S3-055, Science Park 3
AbstractThe tremendous computational power of today´s largest supercomputers offers great possibilities for scientific applications. Nowadays, high performance computer simulations are well-established and are in fact the key to new and ground-braking insights in scientific domains such as Astrophysics, Quantum Mechanics, Climatology, Particle Physics and many more. However in practice, exploiting the huge potential of supercomputers is challenging and requires a lot of computational engineering know-how and experience. In particular, knowledge about modern parallel programming paradigms as well as experience with tools for debugging and profiling distributed applications running on thousands of CPUs is essential.
This talk will provide an overview about the current state-of-the-art with respect to programming, profiling and optimizing applications in High Performance Computing (HPC). Based on examples coming from the optimization work on real-world scientific simulation codes, we will discuss strategies for profiling and tuning massively parallel applications, as well as some important aspects to consider, when developing such applications. The talk will close with an outlook on future trends in HPC and corresponding open challenges posed to computer science.
BioThomas Ponweiser has studied Technical Mathematics with focus on Mathematics in Computer Science at the Vienna University of Technology (TU Wien). Since 2011, he is working at RISC Software GmbH as a Software Developer. Within the EU Project PRACE (http://prace-ri.eu) , he is engaged in the field of High Performance Computing (HPC) with the optimization of scientific simulation codes with respect to performance and scalability in close collaboration with European domain scientists.
Invited by Univ.-Prof. Dr. Volker Strumpen, Institute for Computer Architecture
The Computer Science Colloquium is organized by the Department of Coputer Science at JKU, the Österreichische Gesellschaft für Informatik (ÖGI) and the Österreichische Computergesellschaft (OCG).