The HPC group of RRZE has a long standing expertise in performance modeling, performance
optimization and application parallelization for numerical problems. Besides providing
standard services such as porting, debugging and performance analysis for parallel applications
the group is actively involved in many state-of-the art HPC research activities. These include
evaluation of new multi-/many-core computer architectures and GPUs, determining optimal data
access strategies for ccNUMA architectures and many-core designs with multiple memory controllers,
and parallelization approaches for exploiting shared caches.
A strong research focus lies in efficient implementation of flow solvers based on the Lattice Boltzmann method and their application to reactive flows in porous media. Recently the group has been recognized for work on optimal data layouts, handling of complex geometries, and advanced temporal blocking strategies, to name only a few.
The core contributions to the SKALB projects will be as follows: