Towards Scalable Parallel Numerical Algorithms and Dynamic Load Balancing Strategies
Todays most powerful supercomputers utilize thousands of processing elements to gain an overwhelming performance. This development generates an urgent demand for software that can exploit this massive potential for parallelism. Our working group searches for new algorithms and data structures that can make efficient use of the resources provided by modern parallel computer systems. Currently, we focus on three fields, namely parallel solution methods for ordinary differential equations, task-parallel realizations of numerical algorithms, and dynamic load balancing of irregular applications. In this paper, we present an overview of our recent research related to our project on the HLRB II.
Unable to display preview. Download preview PDF.
- 1.S.Y. Borkar, P. Dubey, K.C. Kahn, D.J. Kuck, H. Mulder, S.S. Pawlowski, J.R. Rattner, Platform 2015: Intel processor and platform evolution for the next decade. Technology@Intel Magazine (2005) Google Scholar
- 2.H. Brunst, D. Kranzlmüller, W.E. Nagel, Tools for scalable parallel program analysis—Vampir VNG and DeWiz, in DAPSYS, ed. by Z. Juhasz, P. Kacsuk, D. Kranzlmüller. Kluwer International Series in Engineering and Computer Science, vol. 777 (Springer, New York, 2004), pp. 93–102 Google Scholar
- 4.N. Faroughi, Multi-cache profiling of parallel processing programs using simics, in Proceedings of the PDPTA, ed. by H.R. Arabnia (CSREA Press, 2006), pp. 499–505 Google Scholar
- 5.R. Hoffmann, M. Korch, T. Rauber, Performance evaluation of task pools based on hardware synchronization, in SC’2004 Conference CD (IEEE/ACM SIGARCH, Pittsburgh, 2004) Google Scholar
- 6.S. Hunold, T. Rauber, Automatic tuning of PDGEMM towards optimal performance, in Proceedings of the Euro-Par Conference 2005, Lisbon, Portugal (Springer, New York, 2005) Google Scholar
- 7.S. Hunold, T. Rauber, G. Rünger, Multilevel hierarchical matrix multiplication on clusters, in Proceedings of the 18th Annual ACM International Conference on Supercomputing, ICS’04, pp. 136–145 (2004) Google Scholar
- 9.M. Korch, T. Rauber, Simulation-based analysis of parallel Runge–Kutta solvers, in Applied Parallel Computing: State of the Art in Scientific Computing. 7th International Workshop, PARA 2004, Revised Selected Papers, Lyngby, Denmark, June 2004. Lecture Notes in Computer Science, vol. 3732 (Springer, Berlin, 2006), pp. 1105–1114 Google Scholar
- 10.A. Malony, S.S. Shende, A. Morris, Phase-based parallel performance profiling, in Proceedings of the PARCO, ed. by G.R. Joubert, W.E. Nagel, F.J. Peters, O.G. Plata, P. Tirado, E.L. Zapata. John von Neumann Institute for Computing Series, vol. 33 (Central Institute for Applied Mathematics, Jülich, 2005), pp. 203–210 Google Scholar
- 11.H. Meuer, E. Strohmaier, J. Dongarra, H.D. Simon, Top500 supercomputer sites. URL http://www.top500.org/
- 12.T. Rauber, G. Rünger, Improving locality for ODE solvers by program transformations. Sci. Program. 12(3), 133–154 (2004) Google Scholar
- 13.S.C. Woo, M. Ohara, E. Torrie, J.P. Singh, A. Gupta, The SPLASH-2 programs: Characterization and methodological considerations, in Proceedings of the 22nd International Symposium on Computer Architecture, Santa Margherita Ligure, Italy (1995), pp. 24–36 Google Scholar