Abstract
In general, message passing multiprocessors suffer from communication overhead between processors and shared memory multiprocessors suffer from memory contention. Also, in computer vision tasks, data I/O overhead limits performance. In particular, high level vision tasks, which are complex and require nondeterministic communication, are strongly affected by these disadvantages. This paper proposes a flexibly (tightly/loosely) coupled hypercube multiprocessor (FCHM) for high level vision to alleviate these problems. A variable address space memory scheme in which a set of adjacent memory modules can be merged into a shared memory module by a dynamically partitionable hypercube topology is proposed. The architecture is quantitatively analyzed using computational models and simulated on the Intel’s Personal SuperComputer (iPSC/I), a hypercube multiprocessor. A parallel algorithm for exhaustive search is simulated on FCHM using the iPSC/I showing significant performance improvements over that of the iPSC/I.
This is a preview of subscription content, log in via an institution to check access.
Similar content being viewed by others
References
Agrawal DP, Jain R (1982) A pipelined pseudoparallel system architecture for real-time dynamic scene analysis. IEEE Trans. Comput. C-31, pp. 952–962
Arden BW, Ginosar R (1982) MP/C: a multiprocessor/computer architecture. IEEE Trans. Comput. C-31, pp 455–473
Baru CK, Su SYW (1985) The architecture of SM3: a dynamically partitionable multicomputer system. IEEE Trans. Comput. C-35, pp 790–802
Briggs FA, Fu KS, Hwang K, Wah BW (1982) PUMPS architecture for pattern analysis and image database management. IEEE Trans. Comput. C-31, pp 969–983
Choudhary AN, Patel JH (1988) Performance of integrated image understanding benchmarks on NETRA: a parallel architecture for integrated vision systems. Proc DARPA Image Understanding Benchmark Workshop, Avon, CT
Danielsson PE (1980) The time-shared bus—a key to efficient image processing. Proc. Int. Conf. Pattern Recognition, pp 296–299
Dessimoz JD, Birk J, Kelley R, Hall J (1981) A vision system with splitting bus. Proc. 1981 IEEECS Workshop Comput. Architect. for Pattern Anal. and Image Database Management, pp 62–66
Kartashev SI, Kartashev SP (1978) Dynamic architectures: problems and solutions. Computer 11, pp 26–41
Kushner T, Wu A Y, Rosenfeld A (1981) Image processing in ZMOB. Proc. IEEECS Workshop Comput. Architect. for Pattern Anal. and Image Database Management, pp 88–95
Lee SY, Aggarwal JK (1986) Exploitation of image parallelism via the hypercube. Second Conf. Hypercube Multiprocessors, Knoxville, TN
Maresca M, Lavin MA, Li H (1988) Parallel architectures for vision. Proceedings of IEEE, vol. 76, pp 970–981
Marsan MA, Balbo G, Conte G (1986) Performance models of multiprocessor systems. MIT Press, Cambridge, MA
Sand Y, Schultz MH (1985) Topological properties of hypercubes. Dept. of Computer Science, Yale University, Research Report YALEU/DCS/RR-389
Schwartz JT (1980) Ultracomputer. ACM TOPLAS 2, pp 484–521
Seitz CL (1985) The cosmic cube. Commun. ACM, vol 28, pp 22–33
Snyder L (1985) An inquiry into the benefits of multiguage parallel computation. Proc. 1985 Int. Conf. Parallel Processing, pp 488–492
Sunwoo MH, Baroody BS, Aggarwal JK (1987) A parallel algorithm for region labeling. Proc. 1987 IEEECS Workshop Comput. Architect. for Pattern Anal. Machine Intell., Seattle, WA, pp 27–34
Sunwoo MH, Aggarwal JK (1988a) Flexibly coupled multiprocessors for image processing. Proc. Int. Conf. Parallel Processing, St Charles IL, vol. I, pp 452–461
Sunwoo MH, Aggarwal JK (1988b) A sliding memory plane array processor. Proc. the 2nd Symp. Frontiers ′88 Massively Parallel Computation, Fairfax, VA, pp 537–540
Sunwoo MH, Aggarwal JK (1988c) A vision tri-architecture (VISTA) for an integrated computer vision system. Proc. DARPA Image Understanding Benchmark Workshop, Avon, CT
Sunwoo MH (1990) VisTA: an integrated vision tri-architecture system. Ph.D. Dissertation, Dept. of Elec. Comput. Eng., University of Texas at Austin
Sunwoo MH, Aggarwal JK (1990a) Flexibly coupled multiprocessors for image processing. Journal of Parallel and Distributed Computing 10: pp 115–129
Sunwoo MH, Aggarwal JK (1990b) VisTA for a general purpose computer vision system. Proc. Int. Conf. Pattern Recognition, Atlantic City, NJ, pp 635–641
Sunwoo MH, Aggarwal JK (1990c) A sliding memory plane array processor for low level vision. Proc. 10th Int. Conf. Pattern Recognition, Atlantic City, NJ, pp 312–317
Sunwoo MH, Aggarwal JK (1991) VisTA for an image understanding architecture. In: Kumar VKP (ed) Parallel architectures and algorithms for image understanding. Academic Press, San Diego, CA, pp 121–154
Tanimoto SL (1986) Architectural issues for intermediate-level vision. In: Duff MJB (ed) Intermediate-level image processing. Academic Press, pp 3–17
Unger SH (1958) A computer oriented toward spatial problems. Proc. IRE 46: pp 1744–1750
Wallace RS, Howard MD (1987) HBA vision architecture: built and benchmarked. Proc. IEEECS Workshop Comput. Architect. for Pattern Anal. Machine Intell., Seattle, WA, pp 209–216
Weems CC, Levitan SP, Hanson AR, Riseman EM, Shu DB, and Nash JG (1989) The image understanding architecture. International Journal of Computer Vision 2: pp 251–282
Yalamanchili S, Palem KV, Davis LS, Welch AJ, and Aggarwal JK (1985) Image processing architectures: a taxonomy and survey. Progress in Pattern Recognition, vol. 2. North-Holland, Amsterdam, pp 1–37
Author information
Authors and Affiliations
Additional information
This research was supported in part by IBM corporation.
Rights and permissions
About this article
Cite this article
Sunwoo, M.H., Aggarwal, J.K. A flexibility coupled hypercube multiprocessor for high level vision. Machine Vis. Apps. 5, 127–138 (1992). https://doi.org/10.1007/BF02620311
Issue Date:
DOI: https://doi.org/10.1007/BF02620311