Dynamic load balancing in the PICADOR plasma simulation code

Authors

  • S.I. Bastrakov
  • I.B. Meyerov
  • I.A. Surmin
  • A.S. Malyshev
  • M.A. Shiryaev
  • E.S. Efimenko
  • A.A. Gonoskov

Keywords:

load balancing
plasma physics
particle-in-cell method
high-performance computing PDF (in Russian) (334KB) PDF. zip (in Russian) (295KB)

Abstract

The load balancing problem for the particle-in-cell plasma simulation on cluster systems is considered. A dynamic load balancing scheme based on the rectilinear partitioning is proposed. An efficient imbalance estimation and the implementation of repartitioning are discussed. Experimental data show that, for significantly imbalanced problems, our implementation is at least two times more efficient compared to the uniform partitioning. The load balancing overhead is less than 1% of the total computational time. This work was prepared in the UNN-Intel ITLab supported by the Federal Target Program «Scientific and educational staff of innovative Russia» (contract N 14.B37.21.0393) and by the Grant Council of the President of the Russian Federation (project code MX–1960.2012.9). The paper is recommended for publication by the Program Committee of the International Scientific Conference «Scientific Service in Internet: All Incarnations of Parallelism» (http://agora.guru.ru/abrau2013).

New computational technologies

Downloads

Published

2013-09-26

Issue

Vol. 14 (2013): Issue 3.

Section

Section 2. Programming

Author Biographies

S.I. Bastrakov

Lobachevsky State University of Nizhni Novgorod
• Assistant

I.B. Meyerov

Lobachevsky State University of Nizhni Novgorod
• Deputy Head of Department

I.A. Surmin

Lobachevsky State University of Nizhni Novgorod
• Programmer

A.S. Malyshev

Lobachevsky State University of Nizhni Novgorod
• Student

M.A. Shiryaev

Lobachevsky State University of Nizhni Novgorod
• Student

E.S. Efimenko

Institute of Applied Physics of RAS
• Junior Researcher

A.A. Gonoskov

Institute of Applied Physics of RAS
• Junior Researcher

References

  1. Бэдсел Ч., Ленгдон А. Физика плазмы и численное моделирование. М.: Энергоатомиздат, 1989.
  2. Pukhov A. Three-dimensional electromagnetic relativistic Particle-In-Cell code VLPL // J. of Plasma Physics. 1999. 61, N 3. 425-433.
  3. Fonseca R.A., et al. OSIRIS: a three-dimensional, fully relativistic Particle-In-Cell code for modeling plasma based accelerators // Lecture Notes in Computational Science. Vol. 2331. Berlin: Springer, 2002. 342-351.
  4. Burau H., Widera R., Honig W., et al. PIConGPU: a fully relativistic Particle-In-Cell code for a GPU cluster // IEEE Trans. on Plasma Science. 2010. 33, N 10. 2831-2839.
  5. Korzhimanov A., Gonoskov A. ELMIS - a fully parallel Fourier-based multidimensional PIC code for laser-plasma interaction simulations // Poc. of the 22-nd Intl. Conf. on Numerical Simulation of Plasmas (ICNSP 2011). September 7-9, 2011. Long Branch, New Jersey, USA (http://icnsp2011/pppl.gov/).
  6. Bastrakov S., Donchenko R., Gonoskov A., Efimenko E., Malyshev A., Meyerov I., Surmin I. Particle-In-Cell plasma simulation on heterogeneous cluster systems // J. of Computational Science. 2012. 3, N 6. 474-479.
  7. Bastrakov S., Meyerov I., Gergel V., et al. High performance computing in biomedical applications // Procedia Computer Science. 2013. 18. 10-19.
  8. Liewer P.C., Decyk V.K. A general concurrent algorithm for plasma particle-in-cell codes // J. Computational Physics. 1989. 85, N 2. 302-322.
  9. Fox G.C. A review of automatic load balancing and decomposition methods for the Hypercube // Numerical Algorithms for Modern Parallel Computer Architectures. The IMA Volumes in Mathematics and Its Applications / Ed. by M. Schultz. New York: Springer, 1988. Vol. 1. 63-76.
  10. Barnes J., Hutt P. A hierarchical O(N logN) force calculation algorithm // Nature. 1986. 324. 446-449.
  11. Seidel D.B., Plimpton S.J., Pasik M.F., et al. Dynamic load balancing for a parallel electromagnetic particle-in-cell code // Proc. IEEE Intl. Pulsed Power Conf. Las Vegas: IEEE Press, 2001. 1000-1003.
  12. Nakashima H., Miyake Y., Usui H., Omura Y. OhHelp: a scalable domain-decomposing dynamic load balancing for particle-in-cell simulations // Proc. of the 23rd Intl. Conf. on Supercomputing (ICS-2009). New York: ACM Press, 2009. 90-99.
  13. Nicol D.N. Rectilinear partitioning of irregular data parallel computations // J. of Parallel and Distributed Computing. 1994. 23, N 2. 119-134.
  14. Taflove A. Computational electrodynamics: the finite-difference time-domain method. London: Artech House, 1995.
  15. Бастраков С.И., Гоносков А.А., Донченко Р.В., Ефименко Е.С., Малышев А.С., Мееров И.Б. Исследование и поиск наиболее эффективных подходов к параллельному моделированию плазмы методом частиц в ячейках на кластерных системах // Тр. Междунар. конф. «Параллельные вычислительные технологии, 2011» (ПаВТ-2011). Челябинск: Издательский центр ЮУрГУ, 2011. 411-417.