The implicit parallelism is an active domain of computer-science to hide intricate details of parallelization from the end-user. Some solutions are specific to a precise domain while others are more generic, however, the purpose is always to find the adapted level of abstraction to ease the high performance and parallel programming. We present SkelGIS, a header-only implicit parallelism C++ library to solve mesh-based scientific simulations. In this paper is detailed the implementation of SkelGIS for the specific case of network simulations, where the space domain can be represented as a directed acyclic graph (DAG). This implementation is based on a modified, optimized and parallelized version of the Compressed Sparse Row format, which is completely described in this paper. Finally, experiments on different kinds of clusters and different sizes of DAGs are evaluated. © 2014 Springer International Publishing Switzerland.
CITATION STYLE
Coullon, H., & Limet, S. (2014). Implementation and performance analysis of SkelGIS for network mesh-based simulations. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 8632 LNCS, pp. 439–450). Springer Verlag. https://doi.org/10.1007/978-3-319-09873-9_37
Mendeley helps you to discover research relevant for your work.