An Energy-Aware Resource Scheduling Strategy Of Power System Tasks In Supercomputer

With the rise and development of high-performance computing platforms,such as super-computing platforms,data centers,cloud computing service platforms,etc.,the problem of high energy consumption has become prominent increasingly.The demand for green computing has increased significantly.Reducing the energy generated by servers not only improves service efficiency,but also reduces the platforms' operating costs.In view of the high energy consumption and insufficient resource utilization of high-performance computing platform,summarize the main energy consumption components by monitoring resource usage data from platform's integrated resource software.An energy model is created by regression analysis,which deals the relationship between resource usage data and energy consumption.Relying on the China Electric Power Research Institute's super-computing platform,design a resource scheduling scheme.Combining with the proposed energy model,a balanced load strategy based on energy perception is proposed,which can improve the overall resource utilization rate from the perspective of improving the platform's load balancing;A node adjustment algorithm to adjust the number of nodes dynamically is proposed.The resource scheduling scheme aims to improve the efficiency of resource utilization and green energy saving of the super-computing platform.Use linux OS and CloudSim,which is a simulation tool for simulating the environment of high-performance computing platforms,measure the energy model's accuracy and the resource scheduling's status of energy saving.Test results show that the energy model(based on the usage of CPU and memory)can achieve almost 90% accuracy,and the scheduling strategy has about 14% energy saving compared with the basic scheduling strategy.