Parallel <scp>CPU–GPU</scp> computing technique for discrete element method

The efficiency of the simulations with discrete element method (DEM) is significantly improved using a novel computational strategy. new developed focus on platforms equipped multi-core central processing units (CPU) and general-purpose graphics (GPU). DEM calculations are performed in parallel CPU GPU pre-calculated Verlet lists posteriori analysis their consistency. operations related to search for possible contacts CPU, whereas interactions, integration motion, executed GPU. Performance done various types tasks has shown that allows decrease average time utilize available resources more efficiently compared sequential CPU–GPU execution mode. Furthermore, due efficient calculations, overall energy requirement proposed strategy does not exceed demand conventional computations.