Parallel Performance of Algebraic Multigrid Domain Decomposition (AMG-DD)

TL;DR

This paper introduces AMG-DD, a low-communication algebraic multigrid domain decomposition method that improves parallel scalability and reduces communication costs in large-scale linear system solutions.

Contribution

The paper presents a novel AMG-DD algorithm that trades additional computation for reduced communication, enhancing parallel performance on modern high-performance computing systems.

Findings

AMG-DD achieves better accuracy per communication cost than standard AMG.

AMG-DD demonstrates significant speedup over AMG on GPU clusters.

Numerical results confirm improved scalability and efficiency of AMG-DD.

Abstract

Algebraic multigrid (AMG) is a widely used scalable solver and preconditioner for large-scale linear systems resulting from the discretization of a wide class of elliptic PDEs. While AMG has optimal computational complexity, the cost of communication has become a significant bottleneck that limits its scalability as processor counts continue to grow on modern machines. This paper examines the design, implementation, and parallel performance of a novel algorithm, Algebraic Multigrid Domain Decomposition (AMG-DD), designed specifically to limit communication. The goal of AMG-DD is to provide a low-communication alternative to standard AMG V-cycles by trading some additional computational overhead for a significant reduction in communication cost. Numerical results show that AMG-DD achieves superior accuracy per communication cost compared to AMG, and speedup over AMG is demonstrated on a large GPU cluster.