6th International Conference on Discrete Element Methods and Related Techniques (DEM 6)
AUG 05-06, 2013
In this paper a parallel computing algorithm for discrete element systems is presented. The discrete model is consisted of finite elements and contacts among the elements. The algorithm is realized using C++ and CUDA and was optimized for NVIDIA GPUs. As a result, the performance of the GPU code is 43 times faster than the sequential code on CPU. The parallel algorithm, the optimism strategy, and the test results are discussed.
Ma ZS,Feng C,Liu TP,et al. An optimized algorithm for discrete element system analysis using CUDA[C]Proceedings of the 6th International Conference on Discrete Element Methods and Related Techniques,2013:170-173.