Published Paper Details:

Energy, non-renewable resource has at-most global importance. Conserving this energy in computational grids especially in large data centers has been a hot topic of discussion since many years. A small amount of reduction in energy consumption in a single node would add up resulting in a significant amount energy savings in the data centers. In this paper the energy model is derived based on various parameters that are obtained using PAPI and Perf, a Linux profiler. The work also concentrate on scaling Central Processing Unit frequency and obtain results for various workload. Molecular dynamics a systolic approach is the program that is used. The program is run at different workloads and at different CPU frequencies. Based on that, energy equation is derived and graphical representation of the results are shown for different frequency ranges and workloads. The result can be utilized to identify the CPU frequency for running the applications which can provide better performance in acceptable amount of time with reduced energy consumption.