NHR Research Projects
Research collaborations within the NHR network connecting expertise across Germany
The NHR network enables NHR centers to collaborate on research topics aimed at promoting and advancing scientific high-performance computing. To this end, the NHR Verein e.V. funds so-called future projects for two years at a time, each involving at least three NHR centers. The Scientific Advisory Board reviews the applications for these research projects.
Current Future Projects
Acronym: AIC
Coordination: Dr. Anita Ragyanszki
Contact: Dr. Anita Ragyanszki (NHR@ZIB), Dr. Robert Schade (PC2)
Centers involved: NHR@ZIB, PC2, NHR@FAU
Other researchers involved: Prof. Dr Martin Brehm (UPB), Dr. Nicholas Charron (ZIB), Dr. Alireza Ghasemi (FAU), Dr. Georg Hager (NHR@FAU), Prof. Dr. Petra Imhof (FAU), Dr. Matthias Läuter (NHR@ZIB), Prof. Dr. Christian Plessl (PC2), Prof. Dr. Gerhard Wellein (NHR@FAU), Dr. Xin Wu (PC2).
Motivation: The project integrates AI/ML into HPC to enhance molecular simulations and structure prediction, developing open-source pipelines for efficiency and accessibility.
Goals and methods: Scaling limits in molecular science make simulations and structure predictions costly. AI/ML reduces these constraints, improving speed and resource efficiency.
Innovations und perspectives: This project introduces AI-driven methodologies for accelerating molecular simulations and structure predictions, enabling more efficient and precise modeling. Its open-source approach fosters collaboration, expands AI’s role in HPC, and lays the groundwork for future advancements in data-driven molecular science.
Projectduration: 24 months, start: Q4-2024
Acronym: EEC
Coordination: Prof. Dr. Julian Kunkel, NHR@Göttingen
Other researchers involved: Dr. Christian Boehme (NHR@Göttingen), Dr. René Caspart (NHR@KIT), Dr. Georg Hager (NHR@FAU), Dr. Thomas Steinke (NHR@ZIB), Dr. Christian Terboven (NHR4CES@RWTH), Dr. Sandra Wienke (NHR4CES@RWTH), Dr. A.W. (NHR4CES@TUDa)
Centers involved: NHR4CES@RWTH, NHR4CES@TUDa, NHR@FAU, NHR@Göttingen, NHR@KIT, NHR@TUD, NHR@ZIB
Motivation: The project aims to standardize the assessment of HPC energy consumption and provide guidelines for optimizing infrastructure, system software, and user software. By improving energy efficiency, it seeks to reduce total cost of ownership (TCO) and minimize the ecological footprint of compute centers.
Goals and Methods: With rising energy costs and increasing computing demands, energy-efficient HPC operations are crucial. This project addresses the issue by optimizing HPC systems to lower costs and energy consumption, ensuring sustainable computing while meeting growing performance needs.
Innovations and perspectives: Focusing on energy efficiency in HPC, this project researches methods and KPIs under the NHR initiative. It aims to develop a standardized benchmark suite to assess energy use, establish an optimization baseline, and enhance HPC performance.
Projectduration: 24 months, start: Q4-2024
Acronym: HAI
Coordination: Dr. Matthias Lieber, NHR@TUD
Other researchers involved: Prof. C.B. (TUDa),Dr. Charlotte Debus (NHR@KIT), Dr. Siavash Ghiasvand (ScaDS.AI), Prof. Harald Koestler NHR@FAU, Jaison Lewis (NHR@Göttingen), Prof. Sarah Neuwirth (NHR@SW), Lincoln Sherpa (NHR@TUD), Dr. Christian Terboven (NHR4CES@RWTH).
Centers involved: NHR4CES@RWTH, NHR4CES@TUDa, NHR@FAU, NHR@Göttingen, NHR@KIT, NHR@SW, NHR@TUD
Motivation: Data analytics and AI pipelines place high demands on software development and computational resources
Goals and methods: This project aims to contribute strategies and tools for efficient and collaborative development as well as high computational efficiency with three focus points: (1) data processing pipelines with LLM-automated data engineering for rapid development; (2) efficient use of HPC resources for scalable model training; and (3) collaborative code development and execution, paired with FAIR data management practices.
Innovations und perspektives: HAI will align to use cases of NHR users and increase their productivity and competences. It will contribute strategies for increased efficiency and usability of HPC resources by providing tools, tutorials, and documentation. Software will be provided open source and can also be used by other computing centers.
Projectduration: 24 months, start: Q4-2024