Scalability and Locality of Extrapolation Methods for Distributed-Memory Architectures

  • Matthias Korch
  • Thomas Rauber
  • Carsten Scholtes
Part of the Lecture Notes in Computer Science book series (LNCS, volume 6272)


The numerical simulation of systems of ordinary differential equations (ODEs), which arise from the mathematical modeling of time-dependent processes, can be highly computationally intensive. Thus, efficient parallel solution methods are desirable. This paper considers the parallel solution of systems of ODEs by explicit extrapolation methods. We analyze and compare the scalability of several implementation variants for distributed-memory architectures which make use of different load balancing strategies and different loop structures. By exploiting the special structure of a large class of ODE systems, the communication costs can be reduced considerably. Further, by processing the micro-steps using a pipeline-like loop structure, the locality of memory references can be increased and a better utilization of the cache hierarchy can be achieved. Runtime experiments on modern parallel computer systems show that the optimized implementations can deliver a high scalability.


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. 1.
    Bulirsch, R., Stoer, J.: Numerical treatment of ordinary differential equations by extrapolation methods. Numer. Math. 8, 1–13 (1966)MathSciNetCrossRefzbMATHGoogle Scholar
  2. 2.
    Burrage, K.: Parallel and Sequential Methods for Ordinary Differential Equations. Oxford University Press, New York (1995)zbMATHGoogle Scholar
  3. 3.
    Deuflhard, P.: Order and stepsize control in extrapolation methods. Numer. Math. 41, 399–422 (1983)MathSciNetCrossRefzbMATHGoogle Scholar
  4. 4.
    Deuflhard, P.: Recent progress in extrapolation methods for ordinary differential equations. SIAM Rev. 27, 505–535 (1985)MathSciNetCrossRefzbMATHGoogle Scholar
  5. 5.
    Ehrig, R., Nowak, U., Deuflhard, P.: Massively parallel linearly-implicit extrapolation algorithms as a powerful tool in process simulation. In: Parallel Computing: Fundamentals, Applications and New Directions, pp. 517–524. Elsevier, Amsterdam (1998)CrossRefGoogle Scholar
  6. 6.
    Gragg, W.B.: On extrapolation algorithms for ordinary initial value problems. SIAM J. Numer. Anal. 2, 384–404 (1965)MathSciNetzbMATHGoogle Scholar
  7. 7.
    van der Houwen, P.J., Sommeijer, B.P.: Parallel iteration of high-order Runge–Kutta methods with stepsize control. J. Comput. Appl. Math. 29, 111–127 (1990)MathSciNetCrossRefzbMATHGoogle Scholar
  8. 8.
    Lustman, L., Neta, B., Gragg, W.: Solution of ordinary differential initial value problems on an Intel Hypercube. Computer Math. Appl. 23(10), 65–72 (1992)CrossRefzbMATHGoogle Scholar
  9. 9.
    Nørsett, S.P., Simonsen, H.H.: Aspects of parallel Runge–Kutta methods. In: Numerical Methods for Ordinary Differential Equations. LNM, vol. 1386, pp. 103–117 (1989)Google Scholar
  10. 10.
    Orozco, D., Gao, G.: Mapping the FDTD application to many-core chip architectures. In: Int. Conf. on Parallel Processing (ICPP-2009). IEEE, Los Alamitos (2009)Google Scholar
  11. 11.
    Rauber, T., Rünger, G.: Load balancing schemes for extrapolation methods. Concurrency: Pract. Ex. 9(3), 181–202 (1997)CrossRefGoogle Scholar
  12. 12.
    Schmitt, B.A., Weiner, R., Jebens, S.: Parameter optimization for explicit parallel peer two-step methods. Appl. Numer. Math. 59, 769–782 (2008)MathSciNetCrossRefzbMATHGoogle Scholar
  13. 13.
    Snir, M., Otto, S.W., Huss-Lederman, S., Walker, D.W., Dongarra, J.: MPI the complete reference, 2nd edn. MIT Press, Cambridge (1998)Google Scholar
  14. 14.
    van der Houwen, P.J., Sommeijer, B.P.: Parallel ODE solvers. In: ACM Int. Conf. on Supercomputing, pp. 71–81 (1990)Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 2010

Authors and Affiliations

  • Matthias Korch
    • 1
  • Thomas Rauber
    • 1
  • Carsten Scholtes
    • 1
  1. 1.Department of Computer ScienceUniversity of BayreuthGermany

Personalised recommendations