Performance and energy efficient dynamic voltage and frequency scaling scheme for multicore embedded system
Yen‐Lin Chen, Ming-Feng Chang, Wen-Yew Liang, Chao-Hsien Lee · 2016
This paper proposes a multicore DVFS method that predicts the best voltage/frequency setting for the embedded system. A smart adjustment mechanism for performance is also provided to users under various performance requirements. The algorithm has been implemented on the platform NVIDIA JETSON Tegra K1. The results show that the proposed DVFS scheme could accurately predict the suitable frequency, given runtime statistics information of a running program. Through this method, users can simply achieve effective energy consumption and performance by specifying the allowable performance loss factor.