Power-Aware Resource Allocation for Independent Tasks in Heterogeneous Real-Time Systems

  • Authors:
  • Affiliations:
  • Venue:
  • ICPADS '02 Proceedings of the 9th International Conference on Parallel and Distributed Systems
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

In recent years, power management and power reductionhas become a critical issue in portable systemsthat are designed for real-time use. In this paper, westudy the problem of static allocation of a set of independenttasks onto a real-time system consisting of heterogeneousprocessing elements, each enabled with discreteDynamic Voltage Scaling. The allocation problemis first formulated as an extended Generalized AssignmentProblem. A linearization heuristic (LR-heuristic)is then extended for solving the problem. An analysisof the upper bound on the number of tasks that theheuristic may fail to allocate is also presented. Ourexperiments show that when the real-time constraintsare tight, the LR-heuristic achieves 15% off the optimalenergy consumption for small size problems, whilethe performance of a classic greedy heuristic is around90% off the optimal. A relative performance improvementof up-to 40% over the classic greedy heuristic isalso observed for large size problems.