Parallelization and performance tuning of molecular dynamics code with OpenMP

Abstract

An OpenMP approach was proposed to parallelize the sequential molecular dynamics (MD) code on shared memory machines. When a code is converted from the sequential form to the parallel form, data dependence is a main problem. A traditional sequential molecular dynamics code is anatomized to find the data dependence segments in it, and the two different methods, i. e., recover method and backward mapping method were used to eliminate those data dependencies in order to realize the parallelization of this sequential MD code. The performance of the parallelized MD code was analyzed by using some performance analysis tools. The results of the test show that the computing size of this code increases sharply form 1 million atoms before parallelization to 20 million atoms after parallelization, and the wall clock during computing is reduced largely. Some hot-spots in this code are found and optimized by improved algorithm. The efficiency of parallel computing is 30% higher than that of before, and the calculation time is saved and larger scale calculation problems are solved.

Keywords

system analysis / molecular dynamics / parallel computing / performance tuning / OpenMP

References

[1] PetersenW P, ArbenzPIntroduction to parallel computing[M], 2004, Boston, Addison-Wesley

[2] LyubartsevA P, LaaksonenA M. DynaMix-a scalable portable parallel MD simulation package for arbitrary molecular mixtures[J]. Comput Phys Commun, 2000, 128(3): 565-589

[3] CouturierR, ChipotC. Parallel molecular dynamics using openMP on a shared memory machine[J]. Comput Phys Commun, 2000, 124(1): 49-59

[4] GoedeckerS. Optimization and parallelization of a force field for silicon using openMP[J]. Comput Phys Commun, 2002, 148(1): 124-135

[5] PlimtonS. Fast parallel algorithms for short-range molecular dynamics[J]. J Comp Phys, 1995, 117(1): 1-19

[6] XiaoShi-fang, DengHui-qiu, ChenShu-guang, et al. . Molecular dynamics simulation of microstructures and mechanical properties of nanocrystalline Ni [J]. Chinese Journal of Nonferrous Metals, 2004, 14(7): 1178-1182(in Chinese)

[7] Hu W, Deng H, Yuan X, et al. Point-defect Properties in HCP rare earth metals with analytic modified embedded atom potentials[J], Euro Phys J B, 2003 (34): 429–440.

[8] HuW, FukumotoM. The application of the analytic embedded atom potentials to alkali metals[J]. Modeling Simu Mater Sci Eng, 2002, 10(6): 707-726

[9] PlimptonS, HendricksonB. A new parallel method for melecular dynamics simulation of macromolecular systerns [J]. Journal of Compaltational Chemistry, 1996, 17(3): 326-337

[10] Steven J, Bruce A. Parallel molecular dynamics with the embedded atom method[C]// MRS Proceedings 291. Pittsburgh: PA, 1993: 37–42.