Abstract
It is well known that communication time is one of the most limiting factors for obtaining high performance programs for parallel architectures.
In our parallelization approach, we first express communications using the get primitive since it is natural and easy to generate automatically. This solution is unfortunately inefficient. This paper addresses the transformation of the get communication schema into a less-costly one. Two tracks are explored: 1. transforming gets into sends, where diffusion is detected and generated. 2. detection of shift communications which are efficiently implemented due to their regular character. We also discuss the combination of these two optimizations and present some experimental results.
Preview
Unable to display preview. Download preview PDF.
References
Luc Bougé: The Data-Parallel Programming Model: a Semantic Perspective. Tech. Report No LIP-IMAG 92–45 (1992)
Jingle Li and Marina Chen: Compiling Communication-Efficient Programs for Massively Parallel Machines. IEEE Trans. on Parallel and Distributed Systems 2 (1991) 361–376
P. Quinton: Automatic Synthesis of Systolic Arrays from Uniform Recurrence Equations. IEEE, Int. Symp. on Computer Architecture, Ann Arbor (1984) 208–214
Paul Feautrier: Dataflow Analysis of Scalar and Array References. Int. J. of Parallel Programming, 20(1) (1991) 23–53
Dror E. Maydan and Saman P. Amarasinghe and Monica S. Lam: Array Dataflow Analysis and its Use in Array Privatization. Proc. of ACM Conf. on Principles of Programming Languages (1993) 2–15
Paul Feautrier: Some Efficient Solutions to the Affine Scheduling Problem, I, One Dimensional Time. Int. J. of Parallel Programming 21(5) (1992) 313–348
Paul Feautrier: Toward Automatic Partitioning of Arrays on Distributed Memory Computers. ACM Int. Conf. on Supercomputing, Tokyo (1993) 175–184
Paul Feautrier: Array Expansion. ACM Int. Conf. on Supercomputing, St Malo (1988) 429–441
M. R. Werth and P. Feautrier: A Systematic Approach of Program Transformation. Procs of the Int. Workshop on Compiler for Parallel Computers, Paris (1990).
M. R. Werth and P. Feautrier: On Parallel Program Generation for Massively Parallel Architectures. High Performance Computing II (1991) M. Durand and F. El Dabaghi editors, Elsevier Science Publisher
M. R. Werth and J. Zahorjan and P. Feautrier: Using Compile-Time Conditional Analysis to Improve the Performance of Compiler Parallelized Programs. Proc. Int. Conference on Massively Parallel Processing, Applications and Development. (1994) Elsevier Science Publisher
Gilbert Strang: Linear Algebra and its Applications. Harcourt Brace Jovanovitch, New York (1988)
Antal E. Fekete: Real Linear Algebra. Marcel Decker Inc., New York (1988)
Alain Mérigot: Associative Nets: A New Parallel Computing Model. Tech. Repport 92-02 IUF-Université Paris Sud (1992)
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1994 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Werth, M.R., Feautrier, P. (1994). Optimizing communications by using compile time analysis. In: Buchberger, B., Volkert, J. (eds) Parallel Processing: CONPAR 94 — VAPP VI. VAPP CONPAR 1994 1994. Lecture Notes in Computer Science, vol 854. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-58430-7_60
Download citation
DOI: https://doi.org/10.1007/3-540-58430-7_60
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-58430-8
Online ISBN: 978-3-540-48789-0
eBook Packages: Springer Book Archive