On-Line Predictive Thermal Management Under Peak Temperature Constraints for Practical Multi-Core Platforms

The power and thermal issues have become one of the major design challenges for the development of modern computing systems. Developing an effective Dynamic thermal management algorithm for the practical multi-core computing system to deliver maximal throughput without encountering temperature emergencies becomes highly demanded. In this paper, we first identify the limitation of the existing theoretical work, and we introduce an enhanced reactive thermal management algorithm based on the dynamic voltage and frequency scaling (DVFS) technique. Then, we develop a new temperature prediction technique and migration scheme that take the local temperature of a core as well as the impacts from neighboring cores into consideration and we validate our algorithm on an Intel desktop. The experimental results show that our approach can significantly improve the throughput while satisfying the temperature constraint compared to the conventional approach.