The Performance of Different Communication Mechanisms and Algorithms Used for Parallelization of Molecular Dynamics Code
Communication performance appears to have the most important influence on parallelization efficiency of large scientific applications. Different communication algorithms and communication mechanisms were used in parallelization of molecular dynamics code. In is shown that in the case of fast communication hardware well scaling algorithm must be used. Presented data shows that MD code can be also run efficiently on the pentium cluster but low latency communication mechanism must be used.
KeywordsMolecular Dynamic Communication Mechanism Communication Library Communication Algorithm Molecular Dynamic Code
Unable to display preview. Download preview PDF.
- 3.T. Clark, R. von Hanxleden, J. A. McCammon, and L. R. Scott. Parallelizing molecular dynamics using spatial decomposition. In Scalable High Performance Computing Conference, pages 95–102, Knoxville, TN, May 1994. IEEE Computer Society. Available via anonymous ftp from softlib.rice.eduaspub/CRPC-TRs/reports/CRPC-TR93356-SGoogle Scholar
- 4.S. Plimpton. Fast parallel algorithms for short-range molecular dynamics. Journal of Computational Physics, 117:1–19, March 1995.Google Scholar
- 5.T. Clark, R. von Hanxleden, K. Kennedy, C. Koelbel, and L. R. Scott. Evaluating parallel languages for molecular dynamics computations. In Scalable High Performance Computing Conference, Williamsburg, VA, 1992. IEEE. Available via anonymous ftp from oftlib.rice.edu as pub/CRPC-TRs/reports/CRPC-TR992202-S.Google Scholar
- 6.R. von Hanxleden. Compiler Support for Machine-Independent Parallelization of Irregular Problems. PhD thesis, Rice University, December 1994. Available via anonymous ftp from softlib.rice.edu as pub/CRPC-TRs/reports/CRPC-TR94495-S.Google Scholar
- 7.P. Bała, T. W. Clark, and L. R. Scott. Application of Pfortran and Co-Array fortran in the parallelization of the GROMOS96 molecular dynamics module. In M. Bubak, J. Moűciñski, and Marian Noga, editors, SGI Users’ Conference-Conference Processdings, pages 194–204. Academic Computer Centre CYFRONET AGH, 2000.Google Scholar
- 8.P. Bała, T. W. Clark, and L. R. Scott. Application of Pfortran and Co-Array fortran in the parallelization of the GROMOS96 molecular dynamics module. Su-percomputing Journal 2001 (in press)Google Scholar
- 9.W. Van Gunsteren and H. J. C. Berendsen. GROMOS (Groningen Molecular Simulation Computer Program Package). Biomos, Laboratory of Physical Chemistry, ETH Zentrum, Zurich, 1996.Google Scholar
- 11.M. P. Allen and D. J. Tildesley. Computer simulations of liquids. Clarendon, Oxford, 1987.Google Scholar
- 14.J. W. Eastwood R. W. Hockney. Computer Simulation Using Particles. Cambridge University Press, Cambridge, 1987.Google Scholar
- 15.USA NERSC, Lawrance Research Laboratory. MVICH-MPI for virtual interface architecture. http://www.nersc.gov/research/FTG/mvich, 2000.
- 16.USA NERSC, Lawrance Research Laboratory. M-VIA: A high performance modular via for linux. http://www.nersc.gov/research/FTG/via, 2000.
- 17.R. W. Numrich. F—: a parallel extension to Cray fortran. Scientific Programming, 6(3):275–84, 1997.Google Scholar
- 18.R. W. Numrich, J. Reid, and K. Kim. Writing a multigrid solver using Co-Array fortran. In B. Kågström, J. Dongarra, E. Elmroth, and J. Waśniewski, editors, Recent Advances in Applied Parallel Computing, Lecture Notes in Computer Science 1541, pages 390–399. Springer-Verlag Berlin, 1998.Google Scholar