This Article 
 Bibliographic References 
 Add to: 
A Network Flow Model for Load Balancing in Circuit-Switched Multicomputers
June 1993 (vol. 4 no. 6)
pp. 649-657

In multicomputers that utilize circuit switching or wormhole routing, communicationoverhead depends largely on link contention-the variation due to distance between nodesis negligible. This has a major impact on the load balancing problem. In this case there are some nodes with an excess load (sources) and other with a deficit load (sinks). Amatching of sources to sinks is required to avoid contention. The problem is madecomplex by the hardwired routing on currently available machines: The user can controlonly which nodes communicate but not how the messages are routed. Network flowmodels of message flow in the mesh and the hypercube have been developed to solvethis problem. The crucial property of these models is the correspondence betweenminimum cost flows and correctly routed messages. To solve a given load balancingproblem, a minimum cost flow algorithm is applied to the network. This permits theefficient determination of a maximum contention free matching of sources to sinks that, in turn, tells how much of the given imbalance can be eliminated without contention.

[1] G. Barneset al., "The Illiac-IV computer,"IEEE Trans. Comput., vol. C-17, pp. 746-757, 1968.
[2] S. H. Bokhari, "Communication overheads on the Intel iPSC-860 hypercube," ICASE Interim Rep. 10, May 1990, NASA CR 182055.
[3] L. Bomans and D. Roose, "Benchmarking the iPSC/2 hypercube multiprocessor,"Concurrency: Practice and Experience, vol. 1, no. 1, pp. 3-18, Sept. 1989.
[4] S. Borkar et al., "iWarp: An Integrated Solution to High Speed Parallel Computing,"Proc. Supercomputing 88, Vol. 1, CS Press, Los Alamitos, Calif., Order No. 882, pp. 330-339.
[5] R. G. Busacker and P. J. Gowen, "A procedure for determining a family of minimal-cost network flow patterns," Tech. Rep. 15, O.R.O., Johns Hopkins Univ., 1961.
[6] R. G. Busacker and T. L. Saaty,Finite Graphs and Networks. New York: McGraw-Hill, 1964.
[7] S. Chittor and R. Enbody, "Performance analysis of Symult 2010's interprocessor communication network," Tech. Rep. CPS-89-19, Dep. Comput. Sci., Michigan State Univ., 1989.
[8] Intel Corp., Touchstone-delta programmers reference manual, Aug. 1991.
[9] Intel Corp., iPSC/2 and iPSC/860 programmers reference manual, June 1990.
[10] W.J. Dally and C.L. Seitz, "Deadlock-Free Message Routing in Multiprocessor Interconnection Networks,"IEEE Trans. Computers, Vol. C-36, No. 5, May 1987, pp. 547-553.
[11] E. W. Dijkstra, "A note on two problems in connexion with graphs,"Numerische Mathematik, vol. 1, pp. 269-271, 1959.
[12] J. Edmonds and R. M. Karp, "Theoretical improvements in algorithmic efficiency for network flow problems,"J. Ass. Comput. Mach., vol. 19, no. 2, pp. 248-264, Apr. 1972.
[13] D. C. Grunwald, "Circuit switched multicomputers and heuristic load placement," Ph.D. thesis, Dep. of Comput. Sci., Univ. of Illinois at Urbana-Champaign, Sept. 1989.
[14] T. Hoshino,PAX Computer: High Speed Parallel Processing and Scientific Computing. Reading, MA: Addison-Wesley, 1989.
[15] T. Hoshino, "An Invitation to the world of PAX,"IEEE Comput. Mag., vol. 19, no. 5, pp. 68-79, May 1986.
[16] T. C. Hu,Combinatorial Algorithms. Reading, MA: Addison-Wesley, 1982.
[17] E. L. Lawler,Combinatorial Optimization: Networks and Matroids. New York: Holt, Rinehart and Winston, 1976.
[18] F. P. Preparata and J. Vuillemin, "The cube-connected cycle: A versatile network for parallel computation,"Commun. ACM, vol. 24, pp. 300-309, May 1981.
[19] C. L. Seitzet al., "The architecture and programming of the Ametek Series 2010 Multicomputer," inProc. Third Conf. Hypercube Concurrent Comput. Appl., ACM, Jan. 1988, pp. 33-37.
[20] H. Sulivan and T. R. Bashkov, "A large scale homogeneous, fully distributed parallel machine, I," inProc. 4th Symp. Comput. Arch., March 1977, pp. 105-117.
[21] R. Endre Tarjan,Data Structures and Network Algorithms, Society for Industrial and Applied Mathematics, 1983.

Index Terms:
Index Termsnetwork flow model; load balancing; circuit-switched multicomputers; wormhole routing;hypercube; minimum cost flow algorithm; contention free matching; concurrency control; hypercube networks; multiprocessing systems
S.H. Bokhari, "A Network Flow Model for Load Balancing in Circuit-Switched Multicomputers," IEEE Transactions on Parallel and Distributed Systems, vol. 4, no. 6, pp. 649-657, June 1993, doi:10.1109/71.242158
Usage of this product signifies your acceptance of the Terms of Use.