Exploiting nested task-parallelism in the H-LU factorization.

Carratalá-Sáez, Rocío, Christophersen, Sven, Aliaga, José I., Beltran, Vicenç, Börm, Steffen and Quintana-Ortí, Enrique S. (2019) Exploiting nested task-parallelism in the H-LU factorization. Journal of Computational Science, 33 . pp. 20-33. DOI 10.1016/j.jocs.2019.02.004.

Full text not available from this repository.

Supplementary data:


We address the parallelization of the LU factorization of hierarchical matrices (H-matrices) arising from boundary element methods. Our approach exploits task-parallelism via the OmpSs programming model and runtime, which discovers the data-flow parallelism intrinsic to the operation at execution time, via the analysis of data dependencies based on the memory addresses of the tasks’ operands. This is especially challenging for H-matrices, as the structures containing the data vary in dimension during the execution. We tackle this issue by decoupling the data structure from that used to detect dependencies. Furthermore, we leverage the support for weak operands and early release of dependencies, recently introduced in OmpSs-2, to accelerate the execution of parallel codes with nested task-parallelism and fine-grain tasks. As a result, we obtain a significant improvement in the parallel performance with respect to our previous work.

Document Type: Article
Research affiliation: Kiel University > Kiel Marine Science
OceanRep > The Future Ocean - Cluster of Excellence
OceanRep > The Future Ocean - Cluster of Excellence > FO-R11
Kiel University
Refereed: Yes
Open Access Journal?: No
DOI etc.: 10.1016/j.jocs.2019.02.004
ISSN: 1877-7503
Projects: Future Ocean
Date Deposited: 01 Aug 2019 10:52
Last Modified: 02 Jan 2020 12:27
URI: http://oceanrep.geomar.de/id/eprint/47312

Actions (login required)

View Item View Item