Evaluation of Tsunami Inundation Simulation using Vector-Scalar Hybrid MPI on SX-Aurora TSUBASA

Authors: Akihiro Musa and Takashi Soga (Tohoku University, NEC Corporation) and Takashi Abe, Masayuki Sato, Kazuhiko Komatsu, Shunichi Koshimura, and Hiroaki Kobayashi (Tohoku University)

Abstract: A real-time tsunami inundation forecast system has been developed since the 2011 Great East Japan earthquake occurred. Reducing the processing time and downsizing the system have been required for the forecast system. To this end, we develop a vector-scalar hybrid MPI code of a tsunami inundation simulation on NEC vector supercomputer SX-Aurora TSUBASA. The vector-scalar hybrid MPI code can run on both Xeon processors and SX vector processors as a single program. Experimental results show that the execution time of the vector-scalar hybrid MPI code is less than five minutes. As a result, the forecast system on a workstation type of SX-Aurora TSUBASA can deliver the prediction information within 17 minutes including the fault estimation and visualization in addition to the simulation.

Best Poster Finalist (BP): no

Poster: PDF
Poster summary: PDF


Back to Poster Archive Listing