A Parallel Genetic Algorithm for Task Mapping on Parallel Machines

In parallel processing systems, a fundamental consideration is the maximization of system performance through task mapping. A good allocation strategy may improve resource utilization and increase significantly the throughput of the system. We demonstrate how to map the tasks among the processors to meet performance criteria, such as minimizing execution time or communication delays. We review the Local Neighborhhod Search (LNS) strategy for the mapping problem. We base our approach on LNS since it was shown that this method outperforms a large number of heuristic-based algorithms. We call our mapping algorithm, that is based on LNS, Genetic Local Neighborhood Search (GLNS), and its parallel version, Parallel Genetic Local Neighborhood Search (P-GLNS). We implemented and compared all three of these mapping strategies. The experimental results demonstrate that 1) GLNS algorithm was better performance than LNS and, 2) The P-GLNS algorithm achieves near linear speedup.

[1]  George Cybenko,et al.  Dynamic Load Balancing for Distributed Memory Multiprocessors , 1989, J. Parallel Distributed Comput..

[2]  Chien-Chung Shen,et al.  A Graph Matching Approach to Optimal Task Assignment in Distributed Computing Systems Using a Minimax Criterion , 1985, IEEE Trans. Computers.

[3]  Arndt Bode.  Load balancing in distributed memory multiprocessors , 1991, [1991] Proceedings, Advanced Computer Technology, Reliable Systems and Applications.

[4]  Kai Hwang,et al.  Optimal load balancing strategies for a multiple processor system , 1981 .

[5]  G. Horton.  A Multi-Level Diffusion Method for Dynamic Load Balancing , 1993, Parallel Comput..

[6]  El-Ghazali Talbi,et al.  A parallel genetic algorithm for the graph partitioning problem , 1991, ICS '91.

[7]  Reinhard Männer,et al.  Parallel Problem Solving from Nature — PPSN III , 1994, Lecture Notes in Computer Science.

[8]  Francisco Sandoval Hernández,et al.  Genetic Algorithms on LAN-message Passing Architectures using PVM: Application to the Routing Problem , 1994, PPSN.

[9]  Edward D. Lazowska,et al.  A Comparison of Receiver-Initiated and Sender-Initiated Adaptive Load Sharing , 1986, Perform. Evaluation.

[10]  Stephen Taylor,et al.  A Practical Approach to Dynamic Load Balancing , 1998, IEEE Trans. Parallel Distributed Syst..

[11]  David E. Goldberg,et al.  Genetic Algorithms in Search Optimization and Machine Learning , 1988 .

[12]  Peter M. Dew,et al.  A periodic symmetrically-initiated load balancing algorithm for distributed systems , 1994, OPSR.

[13]  Emile H. L. Aarts,et al.  Simulated Annealing: Theory and Applications , 1987, Mathematics and Its Applications.

[14]  Chien-Chung Shen,et al.  A Graph Matching Approach to Optimal Task Assignment in Distributed Computing Systems Using a Minimax Criterion , 1985, IEEE Transactions on Computers.

[15]  David S. Johnson,et al.  Computers and Intractability: A Guide to the Theory of NP-Completeness , 1978 .

[16]  Thomas L. Casavant,et al.  A Taxonomy of Scheduling in General-Purpose Distributed Computing Systems , 1988, IEEE Trans. Software Eng..

[17]  Paul Bryant Grosso,et al.  Computer Simulations of Genetic Adaptation: Parallel Subcomponent Interaction in a Multilocus Model , 1985 .