Abstract
We propose a method based on geometrical tools to map problems onto regular and synchronous processor arrays. The problems we consider are defined by systems of affine recurrence equations (SARE). From such a problem specification we extract the data dependencies in terms of two classes of vectors: the utilization vectors and the dependence vectors. We use these vectors to express constraints on the timing or the allocation functions. We differentiate two classes of constraints. The causal ones are intrinsic timing constraints induced by the system of equations defining the problem. A given choice of target architecture may impose new constraints on the timing or the allocation. We call them the architecture-related constraints. We use these constraints to determine first an affine timing function and next an allocation by projection. We finally illustrate the method with three examples: the matrix multiplication, the recursive convolution and the LLt Cholesky factorization.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
Blankinship, W.A.: A new version of the Euclidean algorithm. Am. Math. Mon.,7, 742–745 (1963)
Choo, Y., Chen, M.C.: A theory of program optimization. TR-608, University of Yale, 1988
Clauss, Ph.: Synthèse d'algorithmes systoliques et implantation optimale en place sur réseaux de processeurs synchrones. PhD Thesis, University of Franche-Comté, France, 1990
Clauss, Ph., Perrin, G.R.: Synthesis of process arrays. CONPAR'88. Manchester, G.B., 1988
Clauss, Ph., Mongenet, C., Perrin, G.R.: Synthesis of efficient systolic arrays for matrix algebra problems. Int. Symp. on High Performance Computer Systems, Paris. Amsterdam: North-Holland 1987
Clauss, Ph., Mongenet, C., Perrin, G.R.: Calculus of space-optimal mappings of systolic algorithms on processor arrays. IEEE Int. Conf. on Application-Specific Array Processors, ASAP'90, Princeton, USA, pp. 4–18, 1990
Clauss, Ph., Mongenet, C., Perrin, G.R.: Synthesis of size-optimal toroïdal arrays for the Algebraic Path Problem: a new contribution. Parallel Comput.18, 185–194 (1992)
Darte, A., Robert, Y.: Scheduling uniform loop nests. Research report 92-10, LIP, ENS Lyon, 1992
Delosme, J.M., Ipsen, I.C.F.: An illustration of a methodology for the construction of efficient systolic architectures in VLSI. Sd. Int. Symposium on VSLI Technology, Systems and Applications, Taipei, Taiwan, R.O.C., pp. 268–273, 1985
Delosme, J.M.: A parallel algorithm for the algebraic path problem. In: Cosnard, M., et al. (eds.) Int. Workshop on Parallel and Distributed Algorithms. Amsterdam: Elsevier 1988
Fortes, J.A.B., Fu, K.S., Wah, B.W.: Systematic approaches to the design of algorithmically specified systolic arrays. Int. Conf. on Acoustics, Speech and Signal Processing, 1985
Irigoin, F., Triolet, R.: Dependence approximation and global parallel code generation for nested loops. In: Cosnard, M., et al. (eds.) Int. Workshop on Parallel and Distributed Algorithms, pp. 297–308. Amsterdam: Elsevier 1989
Karp, R.M., Miller, R.E., Winograd, S.: The organization of computations for uniform recurrence equations. J.ACM14 (3), 563–590 (1967)
Kung, H.T. Leiserson, C.E.: Systolic arrays of VLSI. Sparse Matrix Proc., SIAM, pp. 245–282, 1978
Kung, H.T.: Why systolic architectures? Computer15 (1), 37–46 (1982)
Kung, S.Y., Lo, S.C., Lewis, P.S.: Optimal systolic design for the transitive closure and the shortest path problem. IEEE Trans. Comput.C-36 (5), 603–614 (1987)
Moldovan, D.I.: On the design of algorithms for VLSI systolic arrays. Proc IEEE Conf.71 (1), 113–120 (1983)
Moldovan, D.I., Fortes, J.A.B.: Partitioning and mapping algorithms into fixed size systolic arrays. IEEE Trans. Comput.35 (1), 1–12 (1986)
Mongenet, C.: Une méthode de conception d'algorithmes systoliques, résultats théoriques et réalisation. PhD. Thesis, National Polytechnic Institute, Nancy, 1985
Mongenet, C.: Affine timings for systems of affine recurrence equations. Parallel architectures and languages europe, PARLE 91 (Lect. Notes Comput. Sci., vol. 505, pp. 236–251) Berlin, Heidelberg, New York: Springer 1991
Mongenet, C., Clauss, Ph., Perrin, G.R.: Geometrical coding to compile affine recurrence equations on regular arrays. IEEE Int. Parallel Processing Symposium, Anaheim, California, 1991
Mongenet, C., Perrin, G.R.: Synthesis of systolic arrays for inductive problems. Parallel architectures and languages Europe, PARLE 87 (Lect. Notes Comput. Sci., vol. 259, pp. 260–277) Berlin, Heidelberg, New York: Springer 1987
Perrin, G.R., Clauss, Ph., Damy, S.: Mapping programs on regular distributed architectures. Hypercube and distributed computers. Amsterdam: Elsevier 1989
Quinton, P.: Automatic synthesis of systolic arrays from uniform recurrence equations. Proc. IEEE 11th Int. Symp. on Computer Architecture, Ann Arbor, MI, USA, pp. 208–214, 1984
Quinton, P., Van Dongen, V.: The mapping of linear recurrence equations on regular arrays. J. VLSI Signal Process.1, 95–113 (1989)
Rajopadhye, S.: Synthesizing systolic arrays with control signals from recurrence equations. Distrib. Comput.3, 88–105 (1989)
Rao, S.K.: Regular iterative algorithms and their implementation on processor arrays. Ph.D. Thesis, Information Systems Lab., Stanford University, 1985
Robert, Y., Trystram, D.: Systolic solution of the algebraic problem. Int. Workshop on Systolic Arrays, Oxford, Adam-Hilger pp. 171–180, 1987
Rote, G.: A systolic array algorithm for the algebraic path problem. Computing34, 191–219 (1985)
Roychowdhury, V., Thiele, L., Rao, S.K., Kailath, T.: On the localization of algorithms for VLSI processor arrays. IEEE, VLSI Signal Processing III, Monterey, CA, 1988
Saouter, Y., Quinton, P.: Computability of recurrence equations. TR-1203, IRISA, Rennes, 1990
Schrijver, A.: Theory of linear and integer programming. New York: Wiley 1986
Wong, Y., Delosme, J.M.: Optimization of processor count for systolic arrays. Tech. Research Report, Yale University, Computer Science Department, 1989
Yaacobi, Y., Cappello, P.R.: Converting affine recurrence equations to quasi-uniform recurrence equations. Third Int. Workshop on Parallel Computation and VLSI Theory, pp. 373–382, 1988
Yaacobi, Y., Cappello, P.R.: Scheduling a system of affine recurrence equations onto a systolic array. Int. Conf. on Systolic Arrays, San Diego, USA, pp. 373–382, 1988
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Mongenet, C., Clauss, P. & Perrin, GR. Geometrical tools to map systems of affine recurrence equations on regular arrays. Acta Informatica 31, 137–160 (1994). https://doi.org/10.1007/BF01192158
Received:
Issue Date:
DOI: https://doi.org/10.1007/BF01192158