Accelerating DEM simulations on GPUs by reducing the impact of warp divergences

TL;DR

This paper presents a GPU-based method to accelerate Discrete Element Method (DEM) simulations by analyzing and reducing warp divergences during contact detection and force calculations, improving computational efficiency.

Contribution

The paper introduces a novel strategy to mitigate warp divergences in GPU-accelerated DEM simulations, enhancing performance over previous approaches.

Findings

Reduced runtime of DEM force calculations on GPUs.

Identified key sources of warp divergence in contact detection.

Demonstrated improved GPU utilization and efficiency.

Abstract

A way to accelerate DEM calculations on the GPUs is developed. We examined how warp divergences take place in the contact detection and the force calculations taking account of the GPU architecture. Then we showed a strategy to reduce the impact of the warp divergences on the runtime of the DEM force calculations.