Communication costs have become a performance bottleneck in many applications, and are a big issue for high performance computing on massively parallel machines. This paper proposes a halo exchange method for unstructured sparse matrix vector products within the algebraic multigrid method, and evaluate it on a supercomputer with mesh/torus networks. In our numerical tests with a Poisson problem, the proposed method accelerates the linear solver more than 14 times with 23040 cores.
Akihiro FUJII
Kogakuin University
Osni MARQUES
Lawrence Berkeley National Laboratory
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Copy
Akihiro FUJII, Osni MARQUES, "Axis Communication Method for Algebraic Multigrid Solver" in IEICE TRANSACTIONS on Information,
vol. E97-D, no. 11, pp. 2955-2958, November 2014, doi: 10.1587/transinf.2014EDL8052.
Abstract: Communication costs have become a performance bottleneck in many applications, and are a big issue for high performance computing on massively parallel machines. This paper proposes a halo exchange method for unstructured sparse matrix vector products within the algebraic multigrid method, and evaluate it on a supercomputer with mesh/torus networks. In our numerical tests with a Poisson problem, the proposed method accelerates the linear solver more than 14 times with 23040 cores.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.2014EDL8052/_p
Copy
@ARTICLE{e97-d_11_2955,
author={Akihiro FUJII, Osni MARQUES, },
journal={IEICE TRANSACTIONS on Information},
title={Axis Communication Method for Algebraic Multigrid Solver},
year={2014},
volume={E97-D},
number={11},
pages={2955-2958},
abstract={Communication costs have become a performance bottleneck in many applications, and are a big issue for high performance computing on massively parallel machines. This paper proposes a halo exchange method for unstructured sparse matrix vector products within the algebraic multigrid method, and evaluate it on a supercomputer with mesh/torus networks. In our numerical tests with a Poisson problem, the proposed method accelerates the linear solver more than 14 times with 23040 cores.},
keywords={},
doi={10.1587/transinf.2014EDL8052},
ISSN={1745-1361},
month={November},}
Copy
TY - JOUR
TI - Axis Communication Method for Algebraic Multigrid Solver
T2 - IEICE TRANSACTIONS on Information
SP - 2955
EP - 2958
AU - Akihiro FUJII
AU - Osni MARQUES
PY - 2014
DO - 10.1587/transinf.2014EDL8052
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E97-D
IS - 11
JA - IEICE TRANSACTIONS on Information
Y1 - November 2014
AB - Communication costs have become a performance bottleneck in many applications, and are a big issue for high performance computing on massively parallel machines. This paper proposes a halo exchange method for unstructured sparse matrix vector products within the algebraic multigrid method, and evaluate it on a supercomputer with mesh/torus networks. In our numerical tests with a Poisson problem, the proposed method accelerates the linear solver more than 14 times with 23040 cores.
ER -