Advertisement

Mathematical Models and Computer Simulations

, Volume 8, Issue 6, pp 734–743 | Cite as

Efficiency of classical molecular dynamics algorithms on supercomputers

  • G. S. SmirnovEmail author
  • V. V. Stegailov
Article

Abstract

High performance computing hardware is developed faster than the algorithms for fundamental mathematical models such as classical molecular dynamics are adapted. A wide variety of choice makes it necessary to determine clear criteria based on the computational efficiency of a specific algorithm on a particular hardware. The LINPACK benchmark can no longer serve this purpose. In this paper, we analyze the solution time–peak performance metric based on practical considerations. In this metric, we compare different hardware (both current and obsolete) based on the example of the LAMMPS benchmark, which is widely used for atomistic simulations. It is shown that the considered metric can be used for unambiguous comparison of different combinations of CPUs, accelerators, and interconnection.

Keywords

atomistic simulation CPU architecture accelerators peak performance 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. 1.
    List of 500 Best Supercomputers. http://top500.org. Cited April 16, 2015.Google Scholar
  2. 2.
    High-Performance Computing Using FPGAsW, Ed. by W. Vanderbauwhede and K. Benkrid (Springer, New York, 2013).Google Scholar
  3. 3.
    I. Ohmura, G. Morimoto, Y. Ohno, A. Hasegawa, and M. Taiji, “MDGRAPE-4: a special-purpose computer system for molecular dynamics simulations,” Phil. Trans. R. Soc. A 372, 20130387 (2014).MathSciNetCrossRefGoogle Scholar
  4. 4.
    D. E. Shaw et al., “Anton 2: raising the bar for performance and programmability in a special-purpose molecular dynamics supercomputer,” in Proceedings of the SC14 International Conference on High Performance Computing, Networking, Storage and Analysis, New Orleans, LA, USA, November 16–21, 2014 (IEEE Press, Piscataway, 2014), pp. 41–53.Google Scholar
  5. 5.
    V. O. Podryga and S. V. Polyakov, “Molecular dynamic simulation of thermodynamic equilibrium for nickel system,” Math. Models Comput. Simul. 7, 456–466 (2015).MathSciNetCrossRefzbMATHGoogle Scholar
  6. 6.
    V. O. Podryga, S. V. Polyakov, and D. V. Puzyr’kov, “Supercomputer molecular simulation of thermodynamic equilibrium in gas-metal microsystems,” Vychisl. Metody Programmir. 16 (1), 123–138 (2015).Google Scholar
  7. 7.
    W. Eckhardt et al., “591 TFLOPS multi-trillion particles simulation on SuperMUC,” Supercomputing 7905 (2013).Google Scholar
  8. 8.
    V. V. Stegailov and G. E. Norman, “Challenges to the supercomputer development in Russia: a HPC user perspective,” Program. Sistemy: Teor. Prilozh. 5 (1), 111–152 (2014).Google Scholar
  9. 9.
    A. Yu. Kuksin, A. V. Lankin, I. V. Morozov, G. E. Norman, N. D. Orekhov, V. V. Pisarev, G. S. Smirnov, S. V. Starikov, V. V. Stegailov, and A. V. Timofeev, “Predictive modeling and simulation of properties and multiscale processes in materials science. Tasks for Exaflops-era supercomputers,” Program. Sistemy: Teor. Prilozh. 5 (1), 191–244 (2014).Google Scholar
  10. 10.
    E. M. Pestryaev, “Testing of multi-cores graphic processors by molecular dynamics algorithm,” Mat. Model. 26 (1), 69–82 (2014).zbMATHGoogle Scholar
  11. 11.
    F. Reid and I. Bethune, Optimising CP2K for the Intel Xeon Phi. http://www.praceri. eu/IMG/pdf/wp140.pdf. Cited April 16, 2015.Google Scholar
  12. 12.
    H. Jeong et al., Performance of Kepler GTX Titan GPUs and Xeon Phi System. http://arxiv.org/abs/1311.0590. Cited April 16, 2015.Google Scholar
  13. 13.
    E. Y. K. Chan, Benchmarks for Intel MIC Architecture. http://www.clustertech.com/wp-content/ uploads/2014/01/MICBenchmark.pdf. Cited April 16, 2015.Google Scholar
  14. 14.
    http://www.nvidia.ru/object/gpu-computing-facts-ru.html. Cited April 16, 2015.Google Scholar
  15. 15.
    S. J. Pennycook et al., “Exploring SIMD for molecular dynamics, using Intel Xeon processors and Intel Xeon Phi coprocessors,” in Proceedings of the 27th International Parallel and Distributed Processing Symposium IPDPS- 13, Cambridge, Boston, MA, May 20–24, 2013.Google Scholar
  16. 16.
    http://www.nvidia.com/object/io_1258360868914.html. Cited April 16, 2015.Google Scholar
  17. 17.
    http://www.anandtech.com/show/7521/nvidia-launches-tesla-k40. Cited April 16, 2015.Google Scholar
  18. 18.
    http://ark.intel.com/products/75800/Intel-Xeon-Phi-Coprocessor-7120X-16GB-1_238-GHz-61-core. Cited April 16, 2015.Google Scholar
  19. 19.
    J. D. McCalpin, “Memory bandwidth and machine balance in current high performance computers,” IEEE Comput. Soc. Tech. Comm. Comput. Archit. Newslett., 19–25 (1995).Google Scholar
  20. 20.
    http://www.cs.virginia.edu/stream/. Cited April 16, 2015.Google Scholar
  21. 21.
    https://devtalk.nvidia.com/default/topic/381934/stream-benchmark/. Cited April 16, 2015.Google Scholar
  22. 22.
    http://devblogs.nvidia.com/parallelforall/optimizing-high-performance-conjugate-gradient-benchmarkgpus/. Cited April 16, 2015.Google Scholar
  23. 23.
    http://www.cs.virginia.edu/stream/stream_mail/2013/0002.html. Cited April 16, 2015.Google Scholar
  24. 24.
    LAMMPS Software. http://lammps.sandia.gov. Cited April 16, 2015.Google Scholar
  25. 25.
    S. Plimpton, “Fast parallel algorithms for short-range molecular dynamics,” J. Comput. Phys. 117, 1–19 (1995).CrossRefzbMATHGoogle Scholar
  26. 26.
    W. M. Brown et al., “Implementing molecular dynamics on hybrid high performance computers-short range forces,” Comput. Phys. Commun. 182, 898–911 (2011).CrossRefzbMATHGoogle Scholar
  27. 27.
    E. H. Carter, C. R. Trott, and D. Sunderland, “Kokkos: enabling manycore performance portability through polymorphic memory access patterns,” J. Parallel Distrib. Comput. 74, 3202–3216 (2013).CrossRefGoogle Scholar
  28. 28.
    I. V. Morozov, A. M. Kazennov, R. G. Bystryi, G. E. Norman, V. V. Pisarev, and V. V. Stegailov, “Molecular dynamics simulations of the relaxation processes in the condensed matter on GPUs,” Comput. Phys. Commun. 182, 1974–1978 (2011).CrossRefGoogle Scholar
  29. 29.
    W. M. Brown, A. Kohlmeyer, S. J. Plimpton, and A. N. Tharrington, “Implementing molecular dynamics on hybrid high performance computers–particle-particle particle-mesh,” Comput. Phys. Commun. 183, 449–459 (2012).CrossRefGoogle Scholar
  30. 30.
    W. M. Brown and M. Yamada, “Implementing molecular dynamics on hybrid high performance computers–three-body potentials,” Comput. Phys. Commun. 184, 2785–2793 (2013).CrossRefGoogle Scholar
  31. 31.
    C. Begau and G. Sutmann, “Adaptive dynamic load-balancing with irregular domain decomposition for particle simulations,” Comput. Phys. Commun. 190, 51–61 (2015).CrossRefGoogle Scholar
  32. 32.
    Vl. V. Voevodin, S. A. Zhumatii, S. I. Sobolev, A. S. Antonov, P. A. Bryzgalov, D. A. Nikitenko, K. S. Stefanov, and Vad. V. Voevodin, “Practics of ‘Lomonosov’ supercomputer,” Otkryt. Sistemy 7, 36–39 (2012).Google Scholar
  33. 33.
    http://lammps.sandia.gov/bench/lj_one.html. Cited April 16, 2015.Google Scholar
  34. 34.
    Y. Sun, G. Zheng, C. Mei, E. J. Bohm, J. C. Phillips, L. V. Kalé, and T. R. Jones, “Optimizing fine-grained communication in a biomolecular simulation application on Cray XK6,” in Proceedings of the ACM/IEEE International Conference on High Performance Computing, Networking, Storage and Analysis SC12, Salt Lake City, Utah, Nov. 10–16, 2012.Google Scholar
  35. 35.
    S. Kumar, Y. Sun, and L. V. Kalé, “Acceleration of an asynchronous message driven programming paradigm on IBM Blue Gene/Q,” in Proceedings of the 27th International Parallel and Distributed Processing Symposium IPDPS-13, Cambridge, Boston, MA, May 20–24, 2013.Google Scholar

Copyright information

© Pleiades Publishing, Ltd. 2016

Authors and Affiliations

  1. 1.Joint Institute for High TemperaturesRussian Academy of SciencesMoscowRussia
  2. 2.Moscow Institute of Physics and TechnologyMoscowRussia

Personalised recommendations