TU 2: High performance computing
[M. Schweitzer, E. Suarez, C. Urbach]
High performance computing, in its broad sense, is required in all RUs of CASCADE. The models considered in the cluster either employ multi-node or even large-scale computing resources already today, or are planning to do it in the future. In modern computer architectures, gains in computing performance are reached by increasing parallelism and combining heterogeneous hardware components. However, the efficient utilization of parallel computing platforms (at all scales) requires expert knowledge and is often cumbersome. The TU “High Performance Computing” will provide this expert knowledge throughout the cluster, so that all the involved scientific areas can efficiently employ the available computing resources to advance the clusters research agenda in a timely fashion.
