LBFAST: A Lightweight Moment-Represented Lattice Boltzmann Solver for Multi-GPU Architectures
Abstract
We present LBFAST, a GPU-oriented lattice Boltzmann solver based on a lightweight moment-represented formulation, in which post-collision populations are reconstructed on the fly from a reduced set of moments rather than stored explicitly. This approach significantly lowers the memory footprint, enabling large three-dimensional simulations within the constraints of modern accelerator architectures, where VRAM capacity and bandwidth are critical resources. The method is assessed through standard single- and two-component benchmarks demonstrating good accuracy and stability. Extensive scaling experiments on multi-GPU systems show near-ideal weak scaling up to 512 GPUs and sustained performance across different velocity sets. The combination of reduced memory usage, competitive throughput, and stable energy efficiency makes the proposed formulation a practical route for large-scale lattice Boltzmann simulations on current and emerging HPC platforms.
Versions
Related
- A non-uniform local reference frame method for the simulation of rotating geometries with non-conformal grids on the hybrid recursive regularized lattice Boltzmann method
- On the hydrodynamic behaviour of the immersed boundary-lattice Boltzmann method for wetting problems (a)
- Vectorial lattice Boltzmann solver for compressible inviscid flows with generic equation of state
- Residual Carleman lift for nonlinear lattice Boltzmann simulations
- Bridging the Lattice Boltzmann Method and Moment Methods
BibTeX
@misc{lauricella2026lbfast,
title = {{LBFAST: A Lightweight Moment-Represented Lattice Boltzmann Solver for Multi-GPU Architectures}},
author = {Marco Lauricella and Andrea Montessori and Giorgio Amati and Filippo Spiga and Adriano Tiribocchi and Massimo Bernaschi and Sauro Succi},
howpublished = {arXiv},
year = {2026},
month = oct,
doi = {10.48550/arxiv.2609.09160},
eprint = {2609.09160},
archivePrefix = {arXiv},
url = {https://arxiv.org/abs/2609.09160},
}
RIS
TY - UNPB
TI - LBFAST: A Lightweight Moment-Represented Lattice Boltzmann Solver for Multi-GPU Architectures
AU - Marco Lauricella
AU - Andrea Montessori
AU - Giorgio Amati
AU - Filippo Spiga
AU - Adriano Tiribocchi
AU - Massimo Bernaschi
AU - Sauro Succi
PY - 2026
DA - 2026/10/01
PB - arXiv
DO - 10.48550/arxiv.2609.09160
UR - https://arxiv.org/abs/2609.09160
AB - We present LBFAST, a GPU-oriented lattice Boltzmann solver based on a lightweight moment-represented formulation, in which post-collision populations are reconstructed on the fly from a reduced set of moments rather than stored explicitly. This approach significantly lowers the memory footprint, enabling large three-dimensional simulations within the constraints of modern accelerator architectures, where VRAM capacity and bandwidth are critical resources. The method is assessed through standard single- and two-component benchmarks demonstrating good accuracy and stability. Extensive scaling experiments on multi-GPU systems show near-ideal weak scaling up to 512 GPUs and sustained performance across different velocity sets. The combination of reduced memory usage, competitive throughput, and stable energy efficiency makes the proposed formulation a practical route for large-scale lattice Boltzmann simulations on current and emerging HPC platforms.
ER -