High Performance Parallelization of COMPSYN on a Cluster of Multicore Processors with GPUs

In this work we propose a high performance parallelization of the software package COMPSYN, devoted tothe production of syntethic seismograms, on a cluster of multicore processors with multiple GPUs. To design andimplement the proposed high performance version, we started from a na¨ıve parallel version of COMPSYN. Thena¨ıve version consists in a simple parallelization on both device side, obtained by exploiting CUDA, and host side,obtained by exploiting the MPI paradigm and OpenMP API. The proposed high performance version implementsseveral practical techniques of CUDA programming and deeply exploits the GPU architecture, thus achieving a muchbetter performance with respect to the na¨ıve version. We compare the performance of the proposed high performanceversion and that of the na¨ıve one with the performance of the version running on the cluster of multicore processorswithout invoking the GPUs. We obtain for the high performance GPU version a speedup of 25x over the versionrunning on the cluster of multicore processors without GPUs against the 10x of the na¨ıve version. Regarding thesequential version, we estimate about 380x the speedup of the high performance GPU version against the about 140xof the na¨ıve version