ABSTRACT
In order to realize the low cost, high efficiency and safety of the cloud computing system, high energy consumption has become a problem that cannot be ignored. In an environment of computing resource voltage and dynamic adjustment, two energy-efficient scheduling algorithms are proposed, being the energy priority scheduling algorithm and energy genetic scheduling algorithm. Parallel task scheduling in the cloud platform is based on the full consideration of the dependencies among tasks, and the process of parallel and cooperative computation of each sub-task in the graph are distributed to the resource nodes on the Directed Acyclic Graph. Different voltage calculation times and power consumption levels of each sub-task can be different, because of the difference of each computing node under different supply voltage calculation speeds and powers; the communication bandwidth between different nodes is different. Parallel tasks are completed in the final period to reduce the energy consumption of parallel task execution.
