Optimizing parallel HPC applications for green energy sources
Cong Wang, Michael Zink, David Irwin · 2015
While there has been significant prior research on optimizing the energy-efficiency of parallel applications, there has been much less on optimizing them for green energy sources, which expose rapid changes in power's availability (or cost) due to the use of local renewable energy (or utility demand response programs). In this paper, we present energy management policies that utilize active and inactive power capping to maximize the performance of rigid and elastic parallel tasks when subject to variable power constraints from green energy sources. We implement our policies on CloudLab, and evaluate their performance on multiple applications. Our results demonstrate the importance of designing for green energy with variable power. For example, we show that Graph500 requires 17% more time and 9% more energy to complete when power varies based on real-time electricity prices versus when power is unlimited at a fixed price. However, since real-time prices are lower than fixed prices, the total electricity cost of our best energy management policy when using real-time prices is 67% less than when using fixed prices.