Energy efficiency, at the present time, becomes an important consideration in computer software engineering. Dynamic voltage and frequency scaling (DVFS) is an efficient approach for reducing CPU power consumption. This paper exploits DVFS technique to schedule the frame tasks to cores with higher or lower VF settings depending on a task’s time-varying compute intensity. The DVFS based scheduling scheme minimizes the number of voltage and frequency transition as possible to solve the limitation of traditional DVFS for multicore systems. The proposed DVFS based scheduling scheme can improve the energy consumption by 9% to 16% with great speedup for parallel execution. This proposed scheme can effectively improve the energy saving efficiency of video processing applications for mobile computing and healthcare systems.
Relation:
2016 IEEE International Conference on Systems, Man, and Cybernetics