This work presents a set of node-level optimizations to perform the assembly of edge finite element matrices that arise in 3D geophysical electromagnetic modelling on shared-memory architectures. Firstly, we describe the traditional and sequential assembly approach. Secondly, we depict our vectorized and shared-memory strategy which does not require any low level instructions because it is based on an interpreted programming language, namely, Python. As a result, we obtained a simple parallel-vectorized algorithm whose runtime performance is considerably better than sequential version. The set of optimizations have been included to the work-flow of the Parallel Edge-based Tool for Geophysical Electromagnetic Modelling (PETGEM) which is developed as open-source at the Barcelona Supercomputing Center. Finally, we present numerical results for a set of tests in order to illustrate the performance of our strategy.

This project has received funding from the European Union's Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No. 644202. The research leading to these results has received funding from the European Union's Horizon 2020 Programme (2014-2020) and from Brazilian Ministry of Science, Technology and Innovation through Rede Nacional de Pesquisa (RNP) under the HPC4E Project (www.hpc4e.eu), grant agreement No. 689772. Authors gratefully acknowledge the support from the Mexican National Council for Science and Technology (CONACYT). All numerical tests were performed on the MareNostrum supercomputer of the Barcelona Supercomputing Center - Centro Nacional de Supercomputación (www.bsc.es).

Peer Reviewed