Abstract
The paper deals with parallel implementation policies for D.J. Evans S_AGE (Single Alternating Group Explicit) unconditionally stable methods concerning the solution of parabolic problems in two space dimensions.
Space and time mixed parallelisms are investigated with reference to distributed memory computing systems in order to achieve the best relative speedup by means of the best processing elements distribution.
Particular care has been given to startup period minimization together with suitable hypotheses of domain partitioning and of calculation procedural flow.
Some communication strategies are also proposed in terms of data transfer logic, linking resources and their concurrency capabilities.
On these bases, taking into account different parallel system features and different computation policies, a general analytical expression is built allowing to pre-evaluate the performances of different implementations of the S_AGE method.
The speedup evaluations are presented with reference to the parameters determining the actual working environment of the system. They are: the number of time steps, the sampling degree of the domain, the total number of available processors, the distribution of the processors over space and time, the communication capabilities of the processors, the ratio between computation and communication times.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Preview
Unable to display preview. Download preview PDF.
References
Bekakos, M.P. and Evans, D.J. Relative Performance Comparisons for the Group Explicit Class of Methods on MIMD, SIMD, and Pipelined Vector Computers, Parallel Computing, Vol.10, pp. 357–364,1989.
Brochard, L. Efficiency of Some Parallel Numerical Algorithms on Distributed Systems, Parallel Computing, Vol. 12, pp. 21–44,1989.
Burgers, J.M. Adv. Appl. Mach - Vol.1, pp.171,1948.
Carmignani, M. and Genco, A. Implementative Solutions for D.J.Evans Parallel Algorithms (Ed. Laforenza, D. and Perego, R.), pp. 321–326, Proceedings of the International Workshop on Supercomputing Tools for Science and Engineering, Pisa, Italy, 1989, Franco Angell, Milano, Italy, 1990.
Carmignani, M., Genco, A. and Pecorella, G. Speedup Evaluations for the GEC Method using Space and Time Single or Mixed Parallelism (Ed. Messina, P. and Murli, A.), Proceedings of the Int. Conf. Parallel Computing: Achievements, Problems and Prospects, Capri, Italy, 1990, Elsevier, Amsterdam, 1991, in press.
Evans, D.J. and Abdullah, A.R.B. Group Explicit Methods for Parabolic Equations, Int. Jour. Comp. Math., Vol.14, pp.73–105, 1983.
Evans, D.J. and Abdullah, A.R.B. A New Explicit Method for the Solution of Su/St =S2u/óx2 + 62u/Sy2 Int. Jour. Comp. Math., Vol. 14, pp. 325–354, 1983.
Evans, D.J. New Parallel Algorithms for Partial Differential Equations, (Ed. Feilmeier, M., Hebert, J. and Scendel, U.), pp. 1–56, Proceedings of the International Conference on Parallel Computing 83, Berlin, Germany, 1983, Elsevier Science Publishers B.V., North Holland, 1984.
Evans, DJ. Parallel Algorithm Design (Ed. Evans, D.J. and Sutti, C.), pp. 1 to 24, Proceedings of the Int. Meet. on Parallel Computing, Verona, Italy, 1988, Adam Hilger, Bristol and Philadelfia, 1989.
Carling, A. Parallel Processing,the Transputer and Occam, Sigma Press, Wilmslow and Cheshire, U.K., 1988.
INMOS Transputer Reference Manual, Prentice Hall, New York, London, Toronto, Sydney and Tokyo, 1988.
Kronsjö, L. Computational Complexity of Sequential and Parallel Algoritms, John Wiley & Sons, Chichester and New York, 1987.
Saul’yev, V.K. Integration of Equations of Parabolic Type by the Method of Nets, Macmillan, New York,1964.
Carmignani, M., Masnata, A., Ruisi, V. and Tortorici A. A Mathematical Model of Thermal Cycles in Welding (Ed. Lewis, R.W. and Morgan, K.) pp. 1–154 to 1–163, Proceedings of the 5th Int. Conf. on Numerical Methods in Thermal Problems, Montreal, Canada, 1987, Pineridge Press, Swansea, U.K., 1987.
Masnata, A. and Micari, F. In Process Control of Weld Penetration by a Numerical Method (Ed. Lewis, R.W. and Morgan, K.), pp. 2–1412 to 2–1421, Proceedings of the 6th Int. Conf. on Numerical Methods in Thermal Problems, Swansea, U.K, 1989, Pineridge Press, Swansea, U.K., 1989.
Author information
Authors and Affiliations
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 1991 Computational Mechanics Publications
About this chapter
Cite this chapter
Carmignani, M., Genco, A., Pecorella, G. (1991). Mixed Parallelism for Single Alternating Group Explicit Method. In: Brebbia, C.A., Peters, A., Howard, D. (eds) Applications of Supercomputers in Engineering II. Springer, Dordrecht. https://doi.org/10.1007/978-94-011-3660-0_3
Download citation
DOI: https://doi.org/10.1007/978-94-011-3660-0_3
Publisher Name: Springer, Dordrecht
Print ISBN: 978-1-85166-695-9
Online ISBN: 978-94-011-3660-0
eBook Packages: Springer Book Archive