A new GPU implementation for lattice-Boltzmann simulations on sparse geometries

TL;DR

This paper presents a GPU-optimized implementation of the lattice Boltzmann method tailored for sparse 3D geometries, achieving high memory bandwidth utilization and superior performance compared to existing solutions.

Contribution

It introduces a novel data layout and tiling strategy that significantly improves memory efficiency and performance for sparse LBM simulations on GPUs.

Findings

Achieves over 70% of GPU memory bandwidth utilization.

Performs best on sparse geometries with good spatial locality.

Outperforms other GPU LBM implementations in benchmarks.

Abstract

We describe a high-performance implementation of the lattice Boltzmann method (LBM) for sparse 3D geometries on graphic processors (GPU). The main contribution of this work is a data layout that allows to minimise the number of redundant memory transactions during the propagation step of LBM. We show that by using a uniform mesh of small three-dimensional tiles and a careful data placement it is possible to utilise more than 70% of maximum theoretical GPU memory bandwidth for D3Q19 lattice and double precision numbers. The performance of our implementation is thoroughly examined and compared with other GPU implementations of LBM. The proposed method performs the best for sparse geometries with good spatial locality.