Abstract
In this paper we present algorithms for optimizing the behaviour of a restricted class of dataflow networks with bounded buffers by appropriately scheduling the firings of the individual processes. We show how one can find schedules that lead to minimal use of buffers for networks with independent loops. We also present a class of networks for which one can minimize all the buffers simultaneously. We also show that one can maximize the overall concurrency of the computation while maintaining finiteness of the buffers. The techniques we use are largely those developed in the course of the study of the semantics of dataflow networks despite the fact that our study is of a much more intensional nature. The class of processes considered are restricted, in Petri net terminology they are like marked graphs with multiplicities. It turns out that many of the algorithms that arise in digital signal processing can be expressed in terms of these graphs and thus these results are of interest to that community.
Research supported in part by NSERC and FCAR.
Research supported in part by NSERC and FCAR.
This is a preview of subscription content, log in via an institution.
Preview
Unable to display preview. Download preview PDF.
References
S. Bhattacharya, S. Ha, and E. A. Lee. Single appearance schedules for synchronous dataflow programs. Electronics research laboratory memorandum, University of California, Berkeley, Electronics Research Laboratory, College of Engineering, University of California, Berkeley, CA, 94720, 1993.
S. Bhattacharya and E. A. Lee. Memory management for synchronous dataflow programs. Electronics research laboratory memorandum, University of California, Berkeley, Electronics Research Laboratory, College of Engineering, University of California, Berkeley, CA, 94720, 1992.
J. B. Dennis. Control Flow and Data Flow: Concepts of Distributed Programming, volume 14 of NATO ASI Series F: Computer and System Sciences, chapter Data Flow Computation, pages 345–398. Springer-Verlag, 1984.
G. R. Gao, R. Govindarajan, and P. Panangaden. Construction rules for well-behaved stream computations. ACAPS Memo 26, McGill University, ACAPS Group, 3480 rue University Room 318, McGill University, Montreal, Quebec, H3A 2A7, Canada, 1992.
G. R. Gao, R. Govindarajan, and P. Panangaden. Well behaved dataflow graphs for signal processing. In Proceedings of ICASSP92, 1992.
V. Gehlot and C. Gunter. Normal process representatives. In Proceedings of the 5th Annual IEEE Symposium on Logic in Computer Science, pages 200–207, 1990.
G. Kahn. The semantics of a simple language for parallel programming. In Information Processing 74, pages 993–998. North-Holland, 1977.
E. A. Lee. Consistency in dataflow graphs. IEEE Transactions on Parallel and Distributed Systems, 2(2):223–235, 1991.
E. A. Lee. Data parallelism in graphical signal flow representation of algorithms. Electronics Research Laboratory Memorandum UCB/ERL M92/110, University of California, Berkeley, Electronics Research Laboratory, College of Engineering, University of California, Berkeley, CA, 94720, 1992.
Q. Ning and G. R. Gao. A novel framework for register allocation for software pipelining. In Proceedings Of The 20th Annual ACM Symposium On Principles Of Programming Languages, 1993.
W. Reisig. Petri Nets, An Introduction, volume 4 of EATCS Monographs on Theoretical Computer Science. Springer-Verlag, 1985.
Author information
Authors and Affiliations
Editor information
Rights and permissions
Copyright information
© 1993 Springer-Verlag Berlin Heidelberg
About this paper
Cite this paper
Čubrić, M., Panangaden, P. (1993). Minimal memory schedules for dataflow networks. In: Best, E. (eds) CONCUR'93. CONCUR 1993. Lecture Notes in Computer Science, vol 715. Springer, Berlin, Heidelberg. https://doi.org/10.1007/3-540-57208-2_26
Download citation
DOI: https://doi.org/10.1007/3-540-57208-2_26
Published:
Publisher Name: Springer, Berlin, Heidelberg
Print ISBN: 978-3-540-57208-4
Online ISBN: 978-3-540-47968-0
eBook Packages: Springer Book Archive