Domain decomposition and locality optimization for large-scale lattice Boltzmann simulations

TL;DR

This paper introduces a scalable parallel domain decomposition algorithm for large-scale lattice Boltzmann simulations that improves load balancing, data locality, and performance compared to standard graph partitioning tools.

Contribution

It presents a simple, distributed partitioning method tailored for lattice Boltzmann simulations, enhancing scalability and efficiency over existing tools.

Findings

Achieves comparable or higher performance than METIS and PT-SCOTCH.

Ensures load balance and good data locality in complex geometries.

Scalable for a large number of processes.

Abstract

We present a simple, parallel and distributed algorithm for setting up and partitioning a sparse representation of a regular discretized simulation domain. This method is scalable for a large number of processes even for complex geometries and ensures load balance between the domains, reasonable communication interfaces, and good data locality within the domain. Applying this scheme to a list-based lattice Boltzmann flow solver can achieve similar or even higher flow solver performance than widely used standard graph partition based tools such as METIS and PT-SCOTCH.