Parallelization of the Fast Multipole Method for Molecular Dynamics Simulations on Multicore Computers
We have parallelized the fast multipole method (FMM) on multicore computers using OpenMP programming model. The FMM is the one of the fastest approximate force calculation algorithms for molecular dynamics simulations. Its computational complexity is linear. Parallelization of FMM on multicore computers using OpenMP has been reported since the multicore processors become increasingly popular. However the number of those FMM implementations is not large. The main reason is that those FMM implementations have moderate or low parallel efficiency for high expansion orders due to sophisticated formulae of the FMM. In addition, parallel efficiency of those implementations for high expansion orders rapidly drops to 40% or lower as the number of threads increases to 8 or higher. Our FMM implementation on multicore computers using a combination approach as well as a newly developed formula and a computational procedure (A2P) solved the above issues. Test results of our FMM implementation on a multicore computer show that our parallel efficiency with 8 threads is at least 70% for moderate and high expansion orders p = 4,5,6,7. Moreover, the parallel efficiency for moderate and high expansion orders gradually drops from 96% to 70% as the number of threads increases.
Keywordsmolecular dynamics simulations fast multipole method multicore OpenMP parallelization
Unable to display preview. Download preview PDF.
- 1.Haile, M.: Molecular dynamics simulation: Elementary methods. Wiley-Interscience (1997)Google Scholar
- 2.Rappaport, D.C.: The art of molecular dynamics simulation, 2nd edn. Cambridge University Press (2004)Google Scholar
- 3.Satou, A.: Introduction to Practice of Molecular Simulation: Molecular Dynamics, Monte Carlo, Brownian Dynamics, Lattice Boltzmann and Dissipative Particle Dynamics. Elsevier (2010)Google Scholar
- 4.Griebel, M.: Numerical Simulation in Molecular Dynamics: Numerics, Algorithms, Parallelization, Applications. Springer (2010)Google Scholar
- 11.Gumerov, N.A., Duraiswami, R., Zotkin, D.N., Fantalgo, M.D.: Fast Multipole Accelerated Boundary Elements for Numerical Computation of the Head Related Transfer Function. In: IEEE International Conference on Acoustics, Speech and Signal Processing, ICASSP 2007 (2007)Google Scholar
- 12.Gumerov, N.A., Duraiswami, R.: Efficient FMM accelerated vortex methods in three dimensions via the Lamb-Helmholtz decomposition. Submitted to Journal of Computational Physics, arXiv:1201.5430Google Scholar
- 13.Pan, X.M., Pi, W.C., Sheng, X.Q.: On OpenMP parallelization of the multilevel fast multipole algorithm. Progress in Electromagnetics Research 112, 199–213 (2011)Google Scholar
- 14.Yokota, R., Barba, L.: A Tuned and Scalable Fast Multipole Method as a Preeminent Algorithm for Exascale Systems, arXiv:1106.2176v2 [cs.NA] (2011)Google Scholar
- 19.Kawai, J.M.: Pseudoparticle multipole method: A simple method to implement a high-accuracy treecode. The Astrophysical Journal 550, L143–L146 (2001)Google Scholar
- 20.Chau, N.H., Kawai, A., Ebisuzaki, T.: Implementation of fast multipole algorithm on special-purpose computer MDGRAPE-2. In: Proceedings of the 6th World Multiconference on Systemics, Cybernetics and Informatics, SCI 2002, Orlando, Colorado, USA, July 14-18, pp. 477–481 (2002)Google Scholar
- 22.Chau, N.H.: A new formulation for fast calculation of far field force in molecular dynamics simulations. Journal of Science (Mathematics-Physics) 23(1), 1–8 (2007)Google Scholar