An Energy-aware Dynamic Clustering-based Scheduling Algorithm for Parallel tasks on Clusters

The paper proposes an energy-aware dynamic clustering-based scheduling algorithm that aims at reducing communication energy consumption through clustering dependent tasks. A job can be described by a direct acyclic graph (DAG) of parallel tasks. Because the execution time is hard to estimate accurately, the current static scheduling strategies may cause energy increase due to task waiting. The dynamic scheduling method adjusts the clustering group based on the energy consumption threshold. The results of a comparison of this algorithm with static clustering shows that the proposed algorithm has less energy consumption and obtain a shorter makespan to some extent.