Subdomain Deflation Combined with Local AMG: a Case Study Using AMGCL
Library
D. Demidov, and R. Rossi. (2017)cite arxiv:1710.03940Comment: 21 pages, 7 figures.
Abstract
The 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, 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.
Description
Subdomain Deflation Combined with Local AMG: a Case Study Using AMGCL Library
%0 Generic
%1 demidov2017subdomain
%A Demidov, Denis
%A Rossi, Riccardo
%D 2017
%K 65n55-pdes-bvps-multigrid-methods-domain-decomposition 65-04-numerical-analysis-software-source-code
%T Subdomain Deflation Combined with Local AMG: a Case Study Using AMGCL
Library
%U http://arxiv.org/abs/1710.03940
%X The 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, 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.
@misc{demidov2017subdomain,
abstract = {The 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, 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.},
added-at = {2019-12-07T07:46:01.000+0100},
author = {Demidov, Denis and Rossi, Riccardo},
biburl = {https://www.bibsonomy.org/bibtex/2fd96d63fae2fb2648dd4a7b872e28921/gdmcbain},
description = {Subdomain Deflation Combined with Local AMG: a Case Study Using AMGCL Library},
interhash = {e9ff29eb8376c3f0aa8b2e8732159009},
intrahash = {fd96d63fae2fb2648dd4a7b872e28921},
keywords = {65n55-pdes-bvps-multigrid-methods-domain-decomposition 65-04-numerical-analysis-software-source-code},
note = {cite arxiv:1710.03940Comment: 21 pages, 7 figures},
timestamp = {2020-08-24T06:59:39.000+0200},
title = {Subdomain Deflation Combined with Local AMG: a Case Study Using AMGCL
Library},
url = {http://arxiv.org/abs/1710.03940},
year = 2017
}