On June 11 and 12, the seven project partners of the new DFG Research Unit “Mod4Comp” (Holistic Energy and Performance Modeling for Sustainable Computing, FOR5880) officially kicked off the project during a meeting at NHR@FAU.
In this four-year effort, research groups from FAU (Chair of Computer Architecture, NHR@FAU, Professorship for High Performance Comuting), TU Dresden (ZIH Dresden, Chair of Highly-Parallel VLSI Systems and Neuro-Microelectronics, Chair of Adaptive Dynamic Systems), and FZ Jülich (Institute for Advanced Simulation) are joining forces to develop a holistic view on the performance and energy consumption properties of a wide spectrum of computing systems, ranging from embedded and automotive devices, in-memory or near-memory computing hardware, reconfigurable hardware such as FPGAs, neuromorphic systems, and traditional GPU and CPU clusters. The goal is to develop pluggable mathematical models, simulators, and other tools that can be coupled in a modular way.
What makes this Research Unit exceptional is its unique composition of application developers, performance engineers, hardware experts, and computer architects, combined with the wide scope of performance and energy problems in modern compute platforms it aims to tackle.
Although the envisioned modeling workflow is application agnostic, several important application fields will receive special attention: (i) AI and Machine Learning (ML), (ii) autonomous driving for embedded HPC, (iii) computer vision on near-memory architectures, and (iv) spiking neural network simulation to utilize classical computing with and without neuromorphic acceleration.
Learn more about the current status and progress of Mod4Comp on the project website.