Event Details

Title: Sustainability in Supercomputing: A Holistic View on Energy Consumption in HPC and AI

Abstract: The demand for computing is constantly increasing, with artificial intelligence intensifying the necessity for increased computing capabilities even further. This trend is particularly evident in supercomputing centers globally, where exascale computing demands correlate with higher electrical power consumption. Consequently, prioritizing energy efficiency is not merely a desirable feature but an imperative to enhance performance in simulation, modeling, artificial intelligence, and machine learning. The Leibniz Supercomputing Centre (LRZ) stands at the forefront of energy-efficient practices, implementing direct hot-water cooling for its leadership-class systems. Our comprehensive approach encompasses all facets of energy efficiency, spanning from the point of electrical energy entry into the building to the utilization of waste heat, including the incorporation of emerging technologies to facilitate the seamless provision of energy for scientific endeavors. In fact, the learnings on energy efficient HPC also propels better solution for large scale AI infrastructures.

Speaker Bio

Dieter Kranzlmueller is full professor of Computer Science at the Ludwig-Maximilians-Universitaet (LMU) in Munich and the chairman of the board at the Leibniz Supercomputing Centre (LRZ) of the Bavarian Academy of Sciences and Humanities. He serves on a number of boards, including the national Gauss Center for Supercomputing (GCS), the National Research Data Infrastructure (NFDI), and the Center for Digital Technology & Management (CDTM). His research interests focus on research infrastructures, in particular integrating supercomputers, AI technologies and quantum computing.