Research and visualization of algorithms for various energy efficient scheduling models

Abstract

Li et al. and Yao et al. propose their own scheduling algorithms for computing the minimum energy schedule for n jobs [3, 8]. Therefore, we provide a visualizer to see their output schedules and thereby comparing their differences in term of the performance, energy consumption, sequencing of the job execution and so on. In addition, we observe that Li et al.’s schedule may require a processor to adjust it processing speed frequently. And we believe that the number of speed adjustment of their schedule will affect the total energy consumption when the acceleration problem is considered. Moreover, we also find that Li et al.’s and Yao et al.’s schedule require a processor to switch the tasks frequently and this may increase the total energy consumption. Therefore, we introduce two heuristic algorithms, Canyon Reduction (CR) and Switching Reduction (SR). CR is used to minimize the number of speed transition of Li et al.’s schedule and SR is applied to managing the number of context switching of their schedules. Both algorithms take O(n2) to complete. Besides, we use statistical method to evaluate their effectiveness. Finally, we give an algorithm which constructs an optimal single speed schedule with capability of acceleration for n jobs with patterns. And the total running time is O(n).