Parallel spatiotemporally adaptive DEM-based snow simulation

Abstract

This paper applies spatial and temporal adaptivity to an existing discrete element method (DEM) based snow simulation on the GPU. For spatial adaptivity, visually significant spatial regions are identified and simulated at varying resolutions. To this end, we propose efficient splitting and merging to generate adaptive resolution while maintaining the simulation stability. We obtain further optimization by skipping computation on temporally inactive regions. In agreement with the base solver, our method also operates almost entirely on the GPU, which includes operations like activity determination, merging, and splitting of the particles. We demonstrate that a speedup of three times or more of the original non-adaptive simulation can be achieved on scenes containing about 3 million particles. We also discuss the advantages and drawbacks of our spatiotemporal optimization in different simulation scenarios.