New NSF Grant: Exploring The Risks and Rewards of Large Language Models in Enabling Energy-Efficient Data Center Software Infrastructure

The CS Department Chair, Dr. George K. Thiruvathukal, has received an EAGER award from the National Science Foundation’s Clean Energy Technology initiative.

New NSF Grant: Exploring The Risks and Rewards of Large Language Models in Enabling Energy-Efficient Data Center Software Infrastructure
New NSF Grant: Exploring The Risks and Rewards of Large Language Models in Enabling Energy-Efficient Data Center Software Infrastructure

The CS Department Chair, Dr. George K. Thiruvathukal, has received an EAGER award from the National Science Foundation’s Clean Energy Technology initiative. The project is a collaboration with Dr. Jamie Davis and Dr. Yung-Hsiang Lu at Purdue University and Dr. Konstantin Läufer at Loyola University Chicago aims to tackle a pressing issue: the growing energy consumption of data centers.

Data centers currently consume about 2% of the U.S.'s energy, and this number is set to rise. The goal is simple but ambitious: make data centers–and the software running on them--more energy-efficient. The project leverages Large Language Models (LLMs) to optimize software energy use. By optimizing software to use less energy, there could be a cut down on the overall energy footprint of data centers.

The project is also focused on education by developing new educational materials on energy efficiency, which will be available online and presented at a workshop in the second year of the project.

The approach is twofold:

  1. Energy-Efficient Software: LLMs are used to create smarter, energy-saving software solutions for data centers. These models learn and adapt, providing real-time feedback to improve efficiency.
  2. Educational Materials: a new framework for teaching energy efficiency will be created, emphasizing systems thinking. This holistic approach helps students understand the broader impact of energy use and find innovative solutions.

The project is structured around three main thrusts:

  • Measuring the impact of LLMs on energy-efficient programming.
  • Creating and sharing new educational materials.
  • Hosting a workshop to disseminate our findings.

If successful, this project could revolutionize energy-aware software engineering and education. We’re excited to see the positive changes it can bring to both the tech industry and the environment.

For more information about our NSF grant, see https://www.nsf.gov/awardsearch/showAward?AWD_ID=2343595.

Congratulations to our great faculty Dr. George K. Thiruvathukal and Dr. Konstantin Läufer.