Atredia: A Mapping Environment for Dynamic Tree-Structured Problems

  • Angela Sodan


Problems with dynamic tree-structured behavior are usually highly irregular with respect to the shape of potential processes. Such problems require a special mapping on to a parallel machine, with appropriate partitioning of the tree as well as dynamic load balancing. Atredia is an environment providing several tools for mapping dynamic tree-structured problems. These include a granularity controller (for partitioning), a load balancer, a scheduler, and a profiler. Innovative features of Atredia are its support of selection from a set of given granularity-control and load-balancing strategies and their parameterization according to the characteristics of the respective application, the fact that it uses explicit granularity control at all, and its use of a systematic and formalized approach. The latter is realized by performing classifications and calculations based on a model of the application and the machine, and by obtaining dynamic behavior characteristics of the specific application via profiling. One of Atredia’s main aims is applicability to large real-life problems.


Load Balance Critical Path Mapping Environment Task Creation Dynamic Load Balance 


Unable to display preview. Download preview PDF.

Unable to display preview. Download preview PDF.


  1. [1]
    Ishfaq Ahmad. A Semi Distributed Task Allocation Strategy for Large Hybercube Supercomputers, IEEE Supercomputing 1990.Google Scholar
  2. [2]
    Robert S. Boyer and J. Strother Moore: A Computational Logic. Academic Press, New York, 1988.Google Scholar
  3. [3]
    Heiko Bock. Konzeption und Implementierung eines Profilers zur Gewinnung von symbolischen Anwendungen. Diploma thesis, Technical University Berlin, 1994.Google Scholar
  4. [4]
    Christophe Coroyer and Zhen Liu. Effectiveness of Heuristics and Simulated Annealing for the Scheduling of Concurrent Tasks: An Empirical Comparison. Proc. PARLE’93, Parallel Architectures and Languages Europe, Springer-Verlag, 1993.Google Scholar
  5. [5]
    Masakazu Furuichi, Kazuo Taki, Nobuyuki Ichiyoshi. A Multi-Level Load Balancing Scheme for OR-Parallel Exhaustive Search Programs on the Multi-PSI. PPOPP 1990.Google Scholar
  6. [6]
    Franz Incorporated. Allegro Composer, Franz Inc., 1990.Google Scholar
  7. [7]
    Richard P. Gabriel. Performance and Evaluation of Lisp Systems. MIT Press, 1985.Google Scholar
  8. [8]
    Wolfgang K. Giloi. From SUPRENUM to MANNA and META - Parallel Computer Development at GMD FIRST. Proc. 1994 Mannheim Super-computing Seminar, Sauer-Verlag, Munich 1994.Google Scholar
  9. [9]
    Peter Kabat. Parallelisierung des Boyer-Moore Theorembeweisers. Bachelor thesis, Technical University Berlin, 1994.Google Scholar
  10. [10]
    Vipin Kumar and Anshul Gupta. Analyzing Scalability of Parallel Algorithms and Architectures, Journal of Parallel and Distr. Computing, 1994.Google Scholar
  11. [11]
    Vipin Kumar and Anshul Gupta. Scalable Load Balancing Techniques for Parallel Computers, Journal of Parallel and Distributed Computing, 1994.Google Scholar
  12. [12]
    V. Kumar, A. Grama, A. Gupta, and G. Karypis. Introduction to Parallel Computing - Design and Analysis of Algorithms. Benjamin/Cummings Publ. Company, 1994.Google Scholar
  13. [13]
    L.V. Kale and S. Krishnan. CHARM++: A Portable Concurrent Object Oriented System Based on C++. OOPSLA93.Google Scholar
  14. [14]
    L.V. Kale, B. Ramkumar, V. Saletore, and A.B. Sinha. Prioritization in Parallel Symbolic Computing. In Halstead and Ito (eds.), Proc. US/Japan Workshop on Parallel Symbolic Computing: Languages, Systems, and Applications, Oct. 1992, Springer-Verlag, 1993.Google Scholar
  15. [15]
    Wolfgang Küchlin, Universität Tübingen, private communication, Sept. 1994.Google Scholar
  16. [16]
    Eric Mohr, David A. Kranz, and Robert H. Halstead. Lazy Task Creation: A Technique for Increasing the Granularity of Parallel Programs. Proceedings ACM Conference on Lisp and Functional Programming, 1990.Google Scholar
  17. [17]
    Brian Reistad and David K. Gifford. Static Dependent Costs for Estimating Execution Time. ACM Conf. on Lisp and Functional Programming, June 1994.Google Scholar
  18. [18]
    A. Reinefeld and V. Schnecke. Work-Load Balancing in Highly Parallel Depth-First Search. Proc. SHPCC’94, Knoxville/Tennessee, May 1994.Google Scholar
  19. [19]
    Vivek Sarkar. Partitioning and Scheduling Parallel Programs for Multiprocessors. MIT Press, 1989.Google Scholar
  20. [20]
    Angela Sodan and Hua Bi. A Semi-Automatic Approach for Paralleli-zing Symbolic Processing Programs First Int. Symp. on Parallel Symbolic Comp., Linz/ Austria, Sept. 1994.Google Scholar
  21. [21]
    Amitabh B. Sinha and Laxmikant Kale. A Load Balancing Strategy for Prioritized Execution of Tasks. Internat. Parallel Processing Symposium, Los Angeles/CA, April 1993.Google Scholar
  22. [22]
    Amitabh B. Sinha and Kaxmikant V. Kale. A framework for intelligent performance feedback. ICPP-94.Google Scholar
  23. [23]
    Angela Sodan. Parallelisierung von Lisp - Inwieweit bieten deklarative Sprachmittel Vorteile? Workshop on “Entwicklung, Test und Wartung deklarativer KI-Programme” at 18th German Ann. Conf. on AI, Saarbrücken, Sept. 1994, Springer Press (short version), GMD-Berichte (long version).Google Scholar

Copyright information

© Springer Science+Business Media Dordrecht 1995

Authors and Affiliations

  • Angela Sodan
    • 1
  1. 1.GMD Institute for Computer Architecture and Software TechnologyBerlinGermany

Personalised recommendations