Boosting Power Efficiency of HPC Applications with GEOPM
TimeSunday, 17 November 20198:30am - 12pm
DescriptionPower and energy are critical constraints for exascale supercomputing. Optimizing for application performance under such constraints is becoming challenging due to the dynamic phase behavior of scientific codes, increasing variation in processor power efficiency resulting from manufacturing, and due to complexities arising from upcoming heterogeneous architectures.
In order to address some of these challenges, this half-day hands-on tutorial will discuss GEOPM (Global Extensible Open Power Manager) as an emerging, cross-community, cross-platform, open source (BSD 3-clause), job-level power management framework that is capable of providing feedback to job schedulers and resource managers in order to drive system power efficiency improvements. This framework is part of OpenHPC and a key pillar of the software stack in the upcoming first exascale system at Argonne National Laboratory in the US. It optimizes for time-to-solution by leveraging techniques from learning and control systems. GEOPM leverages hardware monitoring and control knobs provided by modern processors.
In this hands-on tutorial, we will discuss the state-of-the-art power management techniques and show the areas where GEOPM can provide benefit over and above those techniques. A high-level overview of the GEOPM architecture, a walkthrough of the GEOPM plugin infrastructure, and example exercises for hands-on practice will be demoed.