Genetic Scheduling on Minimal Processing Elements in the Grid

  • Authors:
  • Wensheng Yao;Baiyan Li;Jinyuan You

  • Affiliations:
  • -;-;-

  • Venue:
  • AI '02 Proceedings of the 15th Australian Joint Conference on Artificial Intelligence: Advances in Artificial Intelligence
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper addresses the problem of scheduling parallel program tasks onto computational grid to minimize the execution time of the parallel program and the number of required processing elements. This task scheduling problem is known to be NP-complete. Existing scheduling algorithms either assume a fixed number of processing elements, or generate schedules that need more processing elements than necessary, which is especially obvious when using task duplication technique. To overcome the weaknesses, we propose a genetic scheduling algorithm using task duplication. The proposed algorithm can yield schedules with shorter execution time and fewer required processing elements, and without useless task duplications. The conditions under which the algorithm performs best were highlighted.