Enhanced Min-Energy Schedules for Optimizing Throughput, Preemptions and On-line Execution
Project: Research
Researcher(s)
- Minming LI (Principal Investigator / Project Coordinator)Department of Computer Science
- Frances Foong YAO (Co-Investigator)Department of Computer Science
Description
Energy consumption has become one of the major concerns for the computer industry in recent years. Current dynamic voltage scaling (DVS) techniques allow the speed of microprocessors to be set dynamically and therefore make it possible to save energy by scheduling jobs wisely. The associated scheduling problems for DVS techniques are classified into two categories: continuous model and discrete model. In the continuous model, the processor can run at arbitrary speeds, while in the discrete model, only a finite number of speed levels are available. Furthermore, the scheduling environment is either offline where all jobs are specified in advance, or online where each job is not known until it arrives. The principal goal of this project is to design online scheduling algorithms for the discrete model under reasonable assumptions and to provide performance guarantee for these online heuristics by comparing them with the min-energy offline schedule. The researchers will also study the characteristics of the min-energy schedule and online heuristics when the original model is enhanced with additional practical considerations, such as job-switch costs and throughput. Completing this project will lead to deeper insights into online heuristics for DVS models and therefore provide useful information for chip designers.Detail(s)
Project number | 9041244 |
---|---|
Grant type | GRF |
Status | Finished |
Effective start/end date | 1/01/08 → 8/09/10 |