Chapter

Integrated Circuit and System Design. Power and Timing Modeling, Optimization and Simulation

Volume 3254 of the series Lecture Notes in Computer Science pp 553-563

Exploiting Dynamic Workload Variation in Offline Low Energy Voltage Scheduling

  • Lap-Fai LeungAffiliated withDepartment of Electrical and Electronic Engineering, Hong Kong University of Science and Technology
  • , Chi-Ying TsuiAffiliated withDepartment of Electrical and Electronic Engineering, Hong Kong University of Science and Technology
  • , Xiaobo Sharon HuAffiliated withDepartment of Computer Science and Engineering, University of Notre Dame

* Final gross prices may vary according to local VAT.

Get Access

Abstract

In this paper, a novel off-line voltage scheduling algorithm, which exploit the dynamic workload variation is proposed. During the construction of the voltage schedule, instead of optimizing the energy consumption assuming all the tasks are running in the worst case workload, we derive a schedule that results in low energy consumption when the tasks are running at a given workload distribution while at the same time can guarantee no deadline violation when the worst-case scenario really happens. By doing so, more slacks are generated and lower voltages can be used when the tasks are really running at workloads that are less than the worst case values. This work can be viewed as an interaction between the off-line voltage scheduling and on-line dynamic voltage scaling. The problem is formulated as a constrained optimization problem and optimal solution is obtained. Simulation and trace-based results show that, by using the proposed scheme, significant energy reduction is obtained for both randomly generated task sets and real-life applications when comparing with the existing best off-line voltage scheduling approach.