Parallel Programming

Affiliated with the computer science department of RWTH Aachen University, our group develops tools that support simulation scientists in exploiting parallelism at massive scales.

IBM BlueGene/P in Jülich with 294,912 cores.

Driven by the increasing demand for computing power and accelerated by current trends in microprocessor design, the number of processor cores on modern supercomputers is growing from generation to generation. For example, the IBM Blue Gene/P system at the nearby Jülich Supercomputing Centre already features almost 300,000 cores. To efficiently utilize these expensive devices and to advance the potential of numerical simulations running on them, we design programming tools for the optimization of parallel codes. These tools must be both (i) effective so that performance expectations can be met and (ii) easy to use so that programmer productivity is maximized. A specific challenge to be addressed is the extreme scalability requirements imposed on software tools when deployed on very large processor configurations.

Moreover, as a complement to our research activities, we teach the skills required to write correct and efficient parallel programs in science and engineering. Students will be introduced to this domain with the goal of subsequently applying their knowledge in a variety of simulation projects or even of contributing to the software tools we develop ourselves.