Abstract
Described are some initial results of a joint research project involving engineering and computer science. The overall objective is to apply advanced computing technology effectively for the solution of engineering problems. Based on earlier work on the automatic derivation and generation of numeric code for finite element analysis, we are conducting research into the mapping of finite element computations on parallel architectures. Software is being developed to automatically derive and generate parallel code that can be used with existing sequential code to improve speed. We are developing techniques to derive parallel procedures, based on high-level user input, to exploit parallel computer architectures. An experimental software system called P-FINGER is under development to derive key finite element routines for the Warp systolic array computer. A separate parallel code generation package, is used to render the symbolically derived parallel procedures into code for the Warp parallel computer. The generated parallel routines execute on the parallel processor under the control of a Fortran-based finite element analysis package which runs on a general purpose host computer. This approach gains speed through parallelism and enables engineers and scientists who are not computer experts to take better advantage of modern parallel computers. The techniques developed can be applied not only to finite element analysis but in many other problem areas. Examples run on the Warp machine are presented with timing comparisons to the traditional sequential approach.
Work reported herein has been supported in part by the National Science Foundation
Chapter PDF
References
Annaratone, M., Arnould, E., Gross, T., Kung, H. T., Lam, M. S., Menzilcioglu, O., and Webb, J. A., “The Warp Machine: Architecture, Implementation and Performance”, IEEE Trans. on Computers, Dec. 1987, vol. C-36, no. 12, pp. 1523–1538.
Bruegge B., Warp Programming Environment: User Manual, Department of Computer Science, Carnegie Mellon University, Pittusburgh, PA, USA.
Chang, T. Y., “NFAP — A Nonlinear Finite Element Program, Vol. 2 — Technical Report,” College of Engineering, University of Akron, Akron, Ohio, USA. 1980.
COSMIC NASTRAN USER'S Manual. Computer Services, University of Georgia, USA. 1985.
Gates, B. L., “A Numerical Code Generation Facility for REDUCE”, Proceedings, ACM SYMSAC'86, Waterloo Ontario, July 1986, pp. 94–99.
Gross, T. and Lam, M., “A Brief Description of W2”, Department of Department of Computer Science, Carnegie Mellon University, Pittsburgh, PA, USA.
van Hulzen, J. A., “Code optimization of multivariate polynomial schemes: a pragmatic approach. Proceedings”, EUROCAL 1983, Springer Lec. Notes Comp. Sci. 162,268–300.
Kung, H. T., “The Warp Computer:A Cost Effective Solution to Supercomputing,” Department of Computer Science, Carnegie Mellon University, Pittusburgh, PA, July 1987.
Navarro, J. J., Llaberia, M. J. and Valero M., “Partitioning: An essential step in Mapping Algorithms Into Systolic Array Processors”, Computer, July 1987, pp. 77–88.
Quinn, M. J., Designing Efficient Algorithms for Parallel Computers, McGraw-Hill Book Co., pp. 51–79.
STARAN-E Reference Manual, Department of Mathematical Sciences, Kent State University, Kent, OH, USA.
VAX UNIX Macsyma, Reference Manual, Version 11, Symbolics, Inc. 1985.
Wang, P. S., “FINGER: A Symbolic System for Automatic Generation of Numerical Programs in Finite Element Analysis”, Journal of Symbolic Computation, Vol. 2, 1986, pp. 305–316.
Zienkiewicz, O. C., The Finite Element Method in Engineering Science, McGraw-Hill, London, pp. 129–153.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1989 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Sharma, N., Wang, P.S. (1989). Symbolic derivation and automatic generation of parallel routines for finite element analysis. In: Gianni, P. (eds) Symbolic and Algebraic Computation. ISSAC 1988. Lecture Notes in Computer Science, vol 358. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-51084-2_4
Download citation
DOI: https://doi.org/10.1007/3-540-51084-2_4
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-51084-0
Online ISBN: 978-3-540-46153-1
eBook Packages: Springer Book Archive