Hybrid dynamic thermal management method with model predictive control
Jian Ma, Hai Wang, Sheldon X.-D. Tan, Chi Zhang, He Tang · 2014
Dynamic thermal management methods are important to control the temperature of microprocessor at runtime and improve the reliability performance of the chip. In this paper, a model predictive control based dynamic thermal management method is proposed with hybrid thermal regulation techniques combining task migration and dynamic voltage frequency scaling (DVFS) methods. The new method is able to track the specified safe temperature ceiling, reduce processor performance degradation, and lower temperature variation among different cores.