Accelerating Aspherix: Performance-Portable GPU DEM Modeling

Aspherix is an advanced Discrete Element Method (DEM) simulation platform for modeling granular materials, developed as the commercial successor to the open-source LIGGGHTS code, itself based on the LAMMPS molecular dynamics engine. Building on these robust foundations, Aspherix delivers significant improvements in computational performance, physical modeling capabilities, and simulation flexibility. Key features include a modern graphical user interface (GUI) and an intuitive, natural-language-inspired command syntax, both designed to streamline simulation setup and lower the barrier to entry for new users.

This work focuses on a performance-portable GPU implementation of Aspherix, engineered to leverage the computational power of diverse hardware platforms – including CPUs and GPUs from NVIDIA, AMD, and Intel. Porting the software to GPUs required a fundamental redesign of memory management and parallelization strategies to address the unique challenges of heterogeneous architectures. We detail some solutions developed to optimize memory usage and parallel execution, ensuring efficient scaling across single and multi-node clusters.

To evaluate Aspherix’s performance, we compare its CPU and GPU implementations against a recent benchmark study of nine leading open-source DEM frameworks [1], assessing both simulation accuracy and computational efficiency on hybrid CPU-GPU systems. Our findings demonstrate Aspherix’s superior scalability and robust performance across a wide range of hardware configurations. Ongoing work focuses on further enhancing computational efficiency and expanding support for emerging simulation scenarios.