High Performance Lattice Boltzmann Algorithms for Fluid Flows

While the lattice Boltzmann method (LBM) has attracted much attention in the area of CFD in recent years, it has also been recognized that it is both computationally demanding and memory intensive. Extensive studies on improving the performance of LBM have been carried out. In this work, various efficient implementation algorithms of LBM are investigated in terms of computational performance and memory consumptions. More precisely, we consider four types high performance LB algorithms: efficient grid refinement, parallel, cache optimization and GPU-based algorithms.