SC²S Colloquium - July 25, 2012

From Sccswiki
Revision as of 12:23, 16 July 2012 by Schreibm (talk | contribs) (Created page with '{| class="wikitable" |- | '''Date:''' || July 16, 2012 |- | '''Room:''' || 02.07.023 |- | '''Time:''' || 3 pm, s.t. |- |} == Diana-Mihaela Gudu: Parallel tsunami simulations wi…')
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigation Jump to search
Date: July 16, 2012
Room: 02.07.023
Time: 3 pm, s.t.

Diana-Mihaela Gudu: Parallel tsunami simulations with block-structured adaptive mesh refinement

Due to the catastrophic effects of tsunamis, tsunami simulations are desired to increase the understanding of the phenomena, as well as to serve as mitigation and early warning systems for the affected areas. Even with the advent of supercomputers, the latter one is far out of reach due to the contradicting need for accuracy and speed. A common solution to this problem are adaptively refined grids, which use a fine discretisation in the regions of interest where a higher accuracy is needed, such as the shoreline, and a coarser resolution in the rest of the domain. Therefore, adaptively refined grids offer a compromise between accuracy and time and memory efficiency. However, they raise a number of algorithmic challenges. The group at the Scientific Computing Chair in TUM is working on the SWE package, a code that implements the governing equations of tsunami simulations, the shallow water equations, with several parallelisation strategies. This project augments the SWE package with a block-structured adaptive mesh refinement. The ”block-structured” term means that the adaptive refinement is done at block level, to avoid the possible over- head that cell-level refinement involves. For this, coarsening and refining strategies were implemented, as well as two integration schemes, with global time-step and different time-steps per refinement level. Additional integration schemes, which use time or space interpolation for the ghost layer exchange, will be implemented. The biggest challenge of adaptive mesh refinement is to deal with different workloads across multiple nodes in the case of parallel codes. This project will adapt the existing MPI implementation to accommodate load balancing strategies. Furthermore, a hybrid parallelisation approach is considered, either MPI-OpenMP or MPI-CUDA, depending on the progress of the work. The results so far seem to be quite promising. For benchmarking, simple artificial scenarios, as well as real data scenarios, will be used.