Paper Info
Title
Power profiling and evaluating the effect of frequency scaling on NWChem.
Abstract
Modern high-performance computing system designers are becoming increasingly aware of the energy requirements for computing while needing to lower the operational costs and raise reliability. At the same time, high-performance application developers are taking pro-active steps towards less energy consumption without a significant performance loss. One way to accomplish this is to change the processor frequency dynamically during application execution. In this paper, a representative HPC application, NWChem, is considered with the aim to investigate its compute and memory intensiveness along with the energy-saving potential of various stages of execution. This work presents energy consumption characteristics of three different algorithms within NWChem that compute the Møller-Plesset (or many-body) perturbation theory second-order correction (MP2) to the Hartree-Fock energy (HF), which radically differ in their computer resource usages. Power consumption measurements are obtained with the application of dynamic voltage and frequency scaling (DVFS) to both processor and memory to determine the runtime behavior of the three MP2 algorithms. Finally, a frequency scaling strategy is proposed and tested. For a 55-atom system, this strategy yields energy savings of up to 10.1% with a modest performance degradation of 2.4%.
Year
DOI
Venue
2016
10.22360/SpringSim.2016.HPC.044
SpringSim (HPS)
Field
DocType
Citations 
Profiling (computer programming),Computer science,Parallel computing,Voltage,Frequency scaling,Operational costs,Computer engineering,Energy consumption,Computing systems,Power consumption
Conference
0
PageRank 
References 
Authors
0.34
0
4
Name
Order
Citations
PageRank
Vaibhav Sundriyal1171.58
Ellie Fought201.01
Masha Sosonkina327245.62
Theresa L. Windus422930.66