Open-source solutions for monitoring and system settings for energy-optimized data centers

Acronym: EE-HPC

Coordination: Dr. -Ing. Jan Eitzinger, NHR@FAU 

Centers involved: NHR4CES@RWTH, NHR@FAU

Further institutions involved: HLRS, DKRZ, HPE Deutschland

Motivation: High-performance computing (HPC) is now one of the fundamental research methods in many scientific disciplines, for example in climate modeling, astrophysics and biology. All data centers in Germany consume around 3% of the country's electricity. Even small energy savings in data centers lead to a relevant reduction in CO2. The aim of the “GreenHPC” funding guideline is to strengthen innovation in Germany by improving energy efficiency in high-performance computing in research and also in commercial data centers.

Goals and methods: The aim of the project is to automatically optimize the energy efficiency of HPC systems. An innovative monitoring system should help to reduce energy consumption while simultaneously increasing computing power. This goal is to be achieved through new software-based control mechanisms for system parameters. The system parameters, such as the utilization of the computing nodes, are to be adjusted automatically. Monitoring software, which is coupled with a new type of user interface, is intended to offer users a transparent platform so that they can also decide on the energy efficiency part of the computing load themselves. This holistic approach ensures flexible and broad use for a wide range of applications.

Innovation and perspectives: The automated optimization solution represents a significant and innovative approach for HPC systems to save energy. The project results can be adapted by existing HPC systems. The underlying open-source approach also promises a high degree of widespread effectiveness and ease of use.

Publications: 

  • Jan Eitzinger, Thomas Gruber, Christian Terboven, Radita Liem, Jose Gracia, Kingshuk Haldar, Pay Giesselmann, Jan Frederik Engels, Torsten Wilde, Marcel Marquardt, Jan Maeder, Christian Simmendinger, David Brayford: EE-HPC – A Framework for Energy Efficient HPC System Operation . Project Poster at ISC24, Hamburg, Germany (2024).
  • Jan Eitzinger, Thomas Gruber, Christian Terboven, Radita Liem, Jose Gracia, Kingshuk Haldar, Pay Giesselmann, Jan Frederik Engels, Torsten Wilde, Marcel Marquardt, Jan Maeder, Christian Simmendinger, David Brayford: EE-HPC – A Framework for Energy Efficient HPC System Operation . Research Poster at SC23, Denver, CO, November 12-17 (2023).

Funding: BMBF, Förderlinie: Green HPC

Project duration: September 2022-August 2025