Optimizing Power and Energy in HPC Data Centers with Energy Aware Runtime

The increasing power consumption of High Performance Computing (HPC) clusters is a concern because of the high cost of electrical power, sustainability considerations and actual limitations in data center power infrastructures. The Energy Aware Runtime (EAR) is an Open Source software developed in a collaboration with Lenovo®, Barcelona Supercomputing Center and Energy Aware Solutions (EAS). EAR provides system and application power monitoring as well as Energy Optimization and Power Capping capabilities and helps HPC data centers to address those challenges.

This paper is intended for HPC system administrators, application developers and data center managers that want to better understand the power usage of their HPC system, optimize the energy consumption of the applications and limit the power consumption of the HPC cluster in operation.

Introduction
Overview of the Energy Aware Runtime
Energy Aware Runtime deep dive
Optimizing the HPC system Energy consumption
System level power capping – managing power limits with EAR
Service and Support for EAR
Conclusions and outlook
References

To view the document, click the Download PDF button.

Changes in the December 13, 2022 update:

• Added additional Energy Aware Runtime part numbers to Table 2 on page 19

Related product families

Product families related to this document are the following:

• High Performance Computing