Advertisement

A heuristic algorithm for dynamic task allocation in highly parallel systems

  • Hans-Ulrich Heiss
  • Rainer Wiesenfarth
Conference paper
Part of the Lecture Notes in Computer Science book series (LNCS, volume 591)

Abstract

We consider the task allocation problem for a homogeneous, multiprogrammed MIMD multicomputer system. Programs arrive in a Poisson stream, and are given as so-called phase graphs. Each phase is described by a task interaction graph. We propose an O(n2) heuristic allocation algorithm where n is the number of tasks. The algorithm has two parts, the first independent of, the second dependent on the topology of the multicomputer. The first part performs a linear hierarchic clustering of the tasks which is used by the second part to map clusters of suitable size onto free parts of the processor graph. The algorithm is evaluated by simulation for a binary tree topology.

Keywards

task allocation mapping problem highly parallel system distributed system 

Preview

Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.

References

  1. [ABPV88]
    Antonelli,S.; Baiardi,F.; Pelagatti,S.; Vanneschi,M.: A Static Approach to Process Mapping in Massively Parallel Systems. Parallel Processing (1988) pp.319–332.Google Scholar
  2. [BP89]
    Baxter,J.; Patel,J.H.: The LAST Algorithm: A Heuristic-Based Static Task Allocation Algorithm. Proc. 1989 Int. Conf. on Parallel Processing, pp.II-217–II-222.Google Scholar
  3. [Bok88]
    Bokhari, S.H.: Partitioning Problems in Parallel, Pipelined, and Distributed Computing. IEEE TOC Vol.37, 1 (Jan. 1988), pp. 48–57.Google Scholar
  4. [BK90]
    Boillat, J.E.; Kropf,P.G.: A Fas Distributed Mapping Algorithm. CONPAR 90, pp.405–416.Google Scholar
  5. [CCL89]
    Chern,M.-S.; Chen,G.H.; Liu,P.: An LC Branch-and-Bound Algorithm for the Module Assignment Problem. Information Processing Letters 32 (1989) pp.61–71.MathSciNetGoogle Scholar
  6. [Cof76]
    Coffman,E.G.: Computer and Job-Shop-Scheduling Theory. John Wiley & Sons, New York, 1976.Google Scholar
  7. [DCDP90]
    Dussa,K.; Carlson,B.; Dowdy,L.; Park,K.-H.: Dynamic Partitioning in a Transputer Environment. ACM SIGMETRICS Conf. 1990, pp.203–213.Google Scholar
  8. [ERS90]
    Ercal,F.; Ramanujam,J.; Sadayappan,P.: Task Allocation onto a Hypercube by Recursive Bipartitioning. Journal of Parallel and Distributed Computing 10 (1990) pp.35–44.Google Scholar
  9. [GJ79]
    Garey,M.R.; Johnson,D.S.: Computers and Intractability: A Guide to the Theory of NP-Completeness. Freeman, San Francisco (1979).Google Scholar
  10. [GIB90]
    Gulati, S.; Iyengar,S.S.; Barhen,J.: The Pebble Crunching Model for Faulttolerant Load Balancing in Hypercube Ensembles. The Computer Journal 33,3 (1990) pp. 204–214.Google Scholar
  11. [Hei91]
    Heiss, H.-U.: Classification of Task Assignment Problems in Parallel Systems. Internal Report No. 7/91, Faculty for Informatics, University of Karlsruhe, June 1991.Google Scholar
  12. [KDL89]
    Kim,J.; Das.C.R.; Lin,W.: A Processor Allocation Scheme for Hypercube Computers. Proc. 1989 Int. Conf. on Parallel Processing, pp.II-231–II-238.Google Scholar
  13. [KM88]
    Krämer,O.; Mühlenbein,H.: Mapping Strategies in Message-Based Multiprocessor Systems. Parallel Computing 9 (1988/89), pp. 213–225.Google Scholar
  14. [Lo88]
    Lo, V.M.: Heuristic Algorithms for Task Assignment in Distributed Systems. IEEE TOC Vol.37, No.11 (Nov. 1988) pp. 1384–1397.Google Scholar
  15. [SER90]
    Sadayappan,P.; Ercal,F.; Ramanujam,J.: Cluster partitioning approaches to mapping parallel programs onto a hypercube. Parallel Computing, Vol.13 (1990), pp. 1–16.CrossRefGoogle Scholar
  16. [Sev89]
    Sevcik,K.: Characterizations of Parallelisms in Applications and Their Use in Scheduling. ACM SIGMETRICS Conf., Berkeley, 1989.Google Scholar
  17. [Wie90]
    Wiesenfarth, R.: Design and Analysis of a Heuristic Algorithm for Dynamic Process Allocation in Highly Parallel Systems. Diploma thesis, University of Karlsruhe, Faculty for Informatics, 1990 (in German)Google Scholar
  18. [Wil90]
    Williams,S.A.: Programming Models for Parallel Systems. John Wiley & Sons, Chichester, 1990.Google Scholar

Copyright information

© Springer-Verlag Berlin Heidelberg 1992

Authors and Affiliations

  • Hans-Ulrich Heiss
    • 1
  • Rainer Wiesenfarth
    • 1
  1. 1.Faculty for InformaticsUniversity of KarlsruheKarlsruhe 1Germany

Personalised recommendations