Heterogeneous Dynamic Load Balancing with a Scheme Based on the Laplacian Polynomial
The problem of dynamic load balancing was extensively studied in the last decade, mainly in homogeneous systems. Significant progress was achieved in the context of applications based on unstructured meshes. If the problem can be considered reasonably solved in homogeneous environments, this is not the case of the heterogeneous systems. In this paper an improved algorithm is proposed. The algorithm is useful in the context of adaptive parallel applications with irregular communication patterns. Our work has been carried out within a heterogeneous model in which different processing capacities are associated with the processors and different costs with the communication links. Experiments were conducted in a heterogeneous cluster of workstations.
KeywordsDynamic Load Unstructured Mesh Heterogeneous Model Homogeneous Environment Load Imbalance
Unable to display preview. Download preview PDF.
- 2.Diekmann, R., Frommer, A., Monien, B.: Efficient Schemes for Nearest Neighbor Load Balancing. In G. Bilardi et al. (eds.), editor, Proc. European Symp. on Algorithms (ESA’98), volume 1461 of Lecture Notes in Computer Science, pages 429–440. Springer, 1998.Google Scholar
- 3.Diekmann, R., Muthukrishnan, S., Nayakkankuppam, M. V.: Engineering Diffusive Load Balancing Algorithms Using Experiments. In G. Bilardi, A. Ferreira, R. Lueling, and J. Rolim, editors, Solving Irregulary Structured Problems in Parallel (IRREGULAR’ 97), volume 1253 of Lecture Notes in Computer Science, pages 111–122. Springer, 1997.CrossRefGoogle Scholar
- 4.Elsässer, R., Monien B., Preis, R.: Diffusive load balancing schemes on heterogeneous networks. In G. Bilardi et al. (eds.), editor, 12th ACM Symposium on Parallel Algorithms and Architectures (SPAA), Vol. 1461, pages 30–38, 2000.Google Scholar
- 6.Hu, Y. F., Blake, R. J.: Load Balancing for Unstructured Mesh Applications. To appear in Parallel and Distributed Computing Practice.Google Scholar
- 7.Hu, Y. F., Blake, R. J.: The Optimal Property of Polynomial Based Diffusion-like Algorithms in Dynamic Load Balancing. In K. D. Papailiou and D. Tsahalis and J. Périaux and D. Knörzer, eds., John Wiley & Son, Computational Dynamics’98, Chichester, 1998.Google Scholar
- 10.Jájá, J.: An Introduction to Parallel Algorithms. Addison-Wesley, 1992.Google Scholar
- 11.Karypis, G., Kumar., V.: Parallel Multilevel k-Way Partitioning Scheme for Irregular Graphs. Technical Report 96-036, Department of Computer Science and Engineering, University of Minnesota, 1996.Google Scholar
- 12.Kelmans, A., Pak, I., Postnikov, A.: Tree and Forest Volumes of Graphs. DIMACS Technical Report, 2000–03, January 2000.Google Scholar
- 13.Rotaru, T., Nägeli, H.-H.: The Generalized Diffusion Algorithm. Techn. Rep. RT-2000/06-1, Institut d’Informatique, Université de Neuchâtel, June 2000.Google Scholar
- 14.Rotaru, T., Nägeli, H.-H.: Minimal Flow Generated by Heterogeneous Diffusion Schemes. In Interntional Conference On Parallel and Distributed Computing and Systems, Anaheim, USA, August 21–24 2001.Google Scholar
- 15.Willebeeck-LeMair, M.H., Reeves, A.P.: Strategies For Dynamic Load Balancing on Highly Parallel Computers. IEEE Transactions on Parallel and Distributed Systems, 4(9):1305–1336, 1993.Google Scholar
- 16.C. Xu and F. Lau.: Load Balancing in Parallel Computers Theory and Practice. The Kluwer International Series in Engineering and Computer Science. Kluwer Academic Publishers, 1997.Google Scholar