Abstract
Most massively parallel architectures exhibit a large gap between hardware capacities and actual communication performance. Dynamic routing is the major cause of this loss of efficiency, because the interconnection network and the processing elements must be quite loosely coupled. The first part of this paper presents experimental analysis of communications in parallel scientific programs, showing that most communication patterns of application programs are determined at compile-time. In the second part, we sketch an execution model intended to exploit this knowledge, under very general assumptions about the underlying interconnection network, and we prove that the model gracefully degrades with the growing complexity and dynamicity of the communication patterns.
Preview
Unable to display preview. Download preview PDF.
References
D. Bailey et al. NAS parallel benchmarks results. In Supercomputing 92, pp 386–393, IEEE Comp. Soc. Press, 1992
O. Lubeck and M. Simmons. The performance realities of massively parallel processors: A case study. In Supercomputing 92, pp 551–560, IEEE Comp. Soc. Press, 1992.
R. Cypher, A. Ho, S. Konstantinidou, and P. Messina. Architectural requirements of parallel scientific applications with explicit communication. In 20th Int. Symp. on Computer Architecture, pp 2–13, 1993.
INTEL Scientific Computers. Paragon XP/S Product Overview, 1991.
C. E. Leiserson et al. The network architecture of the Connection Machine CM-5. In SPAA'92, pp 272–285, 1992
M. Lin et al. Performance Evaluations of the CM-5 Interconnection Network. In COMPCON 93, pp 189–198. IEEE Comp. Soc. Press, 1993.
F.Cappello et al. Balanced distributed memory parallel computers. In 22nd Int. Conf. on Parallel Processing, 1993.
C. Germain, F. Delaplace, and R. Carlier. A static execution model for data parallelism. LRITR-93-862. Submitted to Parallel Processing Letters.
K. Iwama, E. Miyano, and Y. Kambayashi. Routing problems on the mesh of buses. In 3th ISAAC, pp 155–162, 1992. LNCS 650.
J. Lenfant. A versatile mechanism to move data in an array processor. IEEE Trans. on Computers, 34(6): 506–522, June 1985.
G. Lev, N. Pippenger, and L. Valiant. A fast parallel algorithm for routing in permutation networks. IEEE Trans. on Computers, 30(2):93–100, February 1981.
C. Leiserson. Fat-trees: universal networks for hardware-efficient supercomputing. IEEE Trans. Computers, 34(10): 892–901, October 1985.
E. Dahl. Mapping and compiled communication on the Connection Machine system. In 5th Distributed Memory Computing Conference, 1990.
A. G. Mohamed et al. Applications Benchmark Set for Fortran-D and High Performance Fortran. TR-SCCS 327. Northeast Parallel Architecture center, Syracuse University
J. K. Prentice. A performance benchmark study of Fortran 90 compilers. Fortran Journal, 5(3), 1993.
M. Berry, G. Cybenko, and J. Larson. Scientific benchmark characterizations. Parallel Computing, (17):1173–1194, 1991.
M. Wolfe. Tiny: a loop restructuring research tool. Technical report, Oregon Graduate Institute of Science and Technology, 1992.
F. Leighton. Parallel Algorithms and Architectures. Morgan Kaufmann, 1992.
R. Hockney. Performance parameters and benchmarking of supercomputers. Parallel Computing (17)10: 1111–1130, 1991
V. Eicken et al. Active messages: a mechanism for integrated communication and computation. In 19th Int. Symp. on Computer Architecture., pp 256–266, 1992
M. Kumar. Unique design concepts in GF11 and their impact on performance. IBM Jour. Res.Develop. (36)6: 990–999, 1992
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1994 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
de Lahaut, D.G., Germain, C. (1994). Static communications in parallel scientific programs. In: Halatsis, C., Maritsas, D., Philokyprou, G., Theodoridis, S. (eds) PARLE'94 Parallel Architectures and Languages Europe. PARLE 1994. Lecture Notes in Computer Science, vol 817. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-58184-7_107
Download citation
DOI: https://doi.org/10.1007/3-540-58184-7_107
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-58184-0
Online ISBN: 978-3-540-48477-6
eBook Packages: Springer Book Archive