Subdomain deflation combined with local AMG: a case study using AMGCL library

The final publication is available at Springer via http://dx.doi.org/10.1134/S1995080220040071The paper proposes a combination of the subdomain deflation method and local algebraic multigrid as a scalable distributed memory preconditioner that is able to solve large linear systems of equations. The implementation of the algorithm is made available for the community as part of an open source AMGCL library. The solution targets both homogeneous (CPU-only) and heterogeneous (CPU/GPU) systems, employing hybrid MPI/OpenMP approach in the former and a combination of MPI, OpenMP, and CUDA in the latter cases. The use of OpenMP minimizes the number of MPI processes, thus reducing the communication overhead of the deflation method and improving both weak and strong scalability of the preconditioner. The examples of scalar (single degree of freedom per grid node), Poisson-like, systems as well as non-scalar problems, stemming out of the discretization of the Navier-Stokes equations, are considered in order to estimate performance of the implemented algorithm. A comparison with a traditional global AMG preconditioner based on a well-established Trilinos ML package is provided.Contribution of Dr. Demidov was funded by the state assignment to the Joint Supercomputer Center of theRussian Academy of Sciences for Scientific Research and Russian Foundation for Basic Research, grant no. 18-07-00964. Dr. Rossi acknowledges the financial support to CIMNE via the CERCA Programme/Generalitat de Catalunya and the support of the ExaQUte FetHPC, project GA 800898. The authors thankfully acknowledge the support of the PRACE program (project 2010PA4058), in providing access to the MareNostrum 4 and PizDaint clusters. Without such resources the testing would not have been possible. The help of Prof. Labarta of the POP Center of Excellence in improving the NUMA scalability of the solver is also gratefully acknowledged.Peer ReviewedPostprint (author's final draft