IEEE Transactions on Parallel and Distributed Systems
Power-aware resource allocation in high-end systems via online simulation
Proceedings of the 19th annual international conference on Supercomputing
Hi-index | 0.00 |
This paper analyzes workload characteristics and power usage of parallel high-performance computing system. According to the conclusions of the analysis, we propose two kinds of power-aware scheduling policies to reduce the power consumption of system by power down idle nodes reasoningly. Detailed experiments using data from the Parallel Workloads Archive indicates that both policies can achieve considerable overall system power savings while meeting the user-defined switch frequency limit for each node.