The Impact of Dynamic Power Management in Computational Clusters with Multi-Core Processors

Abstract

<span style="font-size:9.0pt;font-family: " times="" new="" roman";mso-fareast-font-family:"times="" roman";mso-bidi-font-family:="" mangal;mso-ansi-language:en-gb;mso-fareast-language:en-us;mso-bidi-language:="" hi"="" lang="EN-GB">In this paper, we study a question related to the execution of jobs in computing clusters built from servers with multi-core processors. Scenarios where a single job is executed or multiple jobs are simultaneously processed by multi-core processors are investigated. Workloads based on captured traces are used in our study. Numerical results demonstrate that the computing resources are efficiently utilized in a multiple job executing scenario and the setup time has a slight impact on the average response time of a computing cluster. Furthermore, a scenario where multiple jobs are simultaneously executed by cores and Dynamic Power Management (DPM) is applied for each processor core yields the most efficient energy consumption. As a consequence, schedulers should take account the feature of multicore processors to save the energy consumption of computing clusters.</span>