Abstract
We propose a class of functions called quasi-linear functions to be used as allocation functions in systolic array synthesis. We describe the special properties of such functions that make them suitable as allocation functions, and give necessary and sufficient conditions that they must satisfy. The main advantage of quasi-linear allocation functions is that they provide a unified framework for dealing with a technique called clustering which is often used as apost-processing technique to improve the efficiency of systolic arrays. We show that it is always possible to choose quasi-linear allocation functions so that the derived arrays have 100% efficiency. Our method is constructive and the allocation function can be obtained automatically. We also generalize our result to deal witharbitrary systolic arrays (outside the context of synthesis). The technique can also be extended to any (piece-wise) systolic array to derive arrays with 100% efficiency.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
Sailesh Rao, “Regular iterative algorithms and their implementations on processor array,” Ph.D. thesis, Stanford University, Information Systems Lab., Stanford, CA, October 1985.
Patrice Quinton, “The systematic design of systolic arrays,” InAutomata Networks in Computer Science, Princeton: Princeton University Press, 1987, pp. 229–260. Preliminary versions appear as IRISA Tech Reports 193 and 216, 1983.
D.I. Moldovan, “On the design of algorithms for VLSI systolic arrays,”Proceedings of the IEEE, vol. 71, 1983, pp. 113–120.
R.M. Karp, R.E. Miller, and S. Winograd, “The organization of computations for uniform recurrence equations,”Journal of the ACM, vol. 14, 1967, pp. 563–590.
Sanjay V. Rajopadhye and Richard M. Fujimoto, “Synthesizing systolic arrays from recurrence equations,”Parallel Computing, vol. 14, 1990, pp. 163–189.
Peter R. Cappello and Kenneth Steiglitz, “Unifying VLSI designs with linear transformations of space-time,”Advanced in Computing Research, vol. 2, 1984, pp. 23–65.
G.J. Ki and B.W. Wah, “Design of optimal systolic arrays,”IEEE Transactions on Computers, vol. C-35, 1985, pp. 66–77.
Jiwan Wong and Jean-Marc Delosme, “Optimization of the processor count for systolic arrays”, Technical Report YALEU-DCS-RR-697, Computer Science Dept., Yale University, May 1989.
Xiaoxiong Zhong, Sanjay V. Rajopadhye, and Ivan Wong, “Systematic generation of linear allocation functions in systolic array design,”Journal of VLSI Signal Processing, to appear.
Lothar Thiele, “On the design of piecewise regular processor arrays,” InInternational Symposium on Circuits and Systems, IEEE Press, 1989, pp. 2239–2542.
S.Y. Kung,VLSI Array Processors, Englewood Cliffs, NJ: Prentice Hall, 1988.
Jichun Bu and Ed F. Deprettere, “Converting sequential iterative algorithms to recurrent equations for automatic design of systolic arrays,” InProceedings of ICASSP, 1988, pp. 2025–2028.
P. Clauss, C. Mongenet, and G.R. Perin, “Calculus of spaceoptimal mappings of systolic algorithms on processor arrays,” In (S.Y. Kung and E. Swartzlander, eds.),International Conference on Application Specific Array Processing, Princeton, New Jersey, 1980, pp. 4–18.
K. Jainundunsing, “Parallel algorithms for solving systems of linear equations and their mapping on systolic arrays,” Ph.D. thesis, Delft University of Technology, Electrical Engineering Department, Delft, The Netherlands January 1989.
Alain Darte and Jean-Marc Delosme, “Partitioning for array processors,” Technical Report 90-23, Laboratoire de I'Informatique du Parallelisme, Ecole Normale Superieure De Lyon, France, October 1990.
Vincent Van Dongen, “Quasi-regular arrays: definition and design methodology,” In (J. McCanny, J. McWhirter, and E. Swartzlander, eds.),Systolic arrays processors, International Conference on Systolic Arrays, Englewood Cliffs, NJ: Prentice Hall, 1989, pp. 126–135.
H.T. Kung and C.E. Leiserson,Algorithms for VLSI Processor Arrays, Reading, MA: Addison-Wesley, 1980, pp. 271–292.
A. Schrijver,Theory of Integer and Linear Programming, New York: Wiley and Sons, 1988.
Utpal Benerjee,Dependence Analysis for Supercomputing, Boston: Kluwer Academic Publishers, 1988.
Sailesh Rao and Thomas Kailath, “What is a systolic algorithm,” InSPIE Proceedings, Highly Parallel Signal Processing Architectures, Los Angeles, CA, 1986, pp. 34–48.
Gunter Rote, “A systolic array algorithm for the algebraic path problem (shortest paths; matrix inversion),”Computing, vol. 34, 1985, pp. 191–219.
J. Bu, E.F. Deprettere, and P. Dewilde, “A design methodology for fixed-size systolic arrays,” In (S.Y. Kung and E. Swartzlander, eds.),International Conference on Application Specific Array Processing, Princeton, New Jersey, 1990, pp. 591–602.
Xiaoxiong Zhong and Sanjay V. Rajopadhye, “Synthesizing efficient systolic arrays,” InIEEE International Conference on Acoustics, Speech and Signal Processing, Toronto, Canada, 1991.
Author information
Authors and Affiliations
Additional information
Supported by NSF grant MIP-8802454.
Rights and permissions
About this article
Cite this article
Zhong, X., Rajopadhye, S. Quasi-Linear allocation functions for efficient array design. J VLSI Sign Process Syst Sign Image Video Technol 4, 97–110 (1992). https://doi.org/10.1007/BF00925116
Received:
Revised:
Published:
Issue Date:
DOI: https://doi.org/10.1007/BF00925116