Multi-GPU fast Fourier transforms in MATLAB (for large-scale phase-field crystal simulations)

TL;DR

This paper introduces a MATLAB framework enabling multi-GPU FFTs for large-scale phase-field crystal simulations, significantly accelerating spectral solvers beyond CPU capabilities.

Contribution

The authors develop a MATLAB-based multi-GPU FFT framework with two strategies to overcome memory limits and speed up large-scale spectral simulations.

Findings

Achieves approximately sixfold speedup for standard PFC simulations.

Attains up to sixtyfold acceleration for multiphysics extensions.

Demonstrates effective multi-GPU strategies for spectral solvers.

Abstract

We present a MATLAB-based framework for two- and three-dimensional fast Fourier transforms on multiple GPUs for large-scale numerical simulations using the pseudo-spectral Fourier method. The software implements two complementary multi-GPU strategies that overcome single-GPU memory limitations and accelerate spectral solvers. This approach is motivated by and applied to phase-field crystal (PFC) models, which are governed by tenth-order partial differential equations, require fine spatial resolution, and are typically formulated in periodic domains. Our resulting numerical framework achieves significant speedups, approximately sixfold for standard PFC simulations and up to sixtyfold for multiphysics extensions, compared to a purely CPU-based implementation running on hundreds of cores.