Spinnaker3
Table of contents
Spinnaker3
The European Union offers a variety of funding programs to support research and education as a valuable public good. Our "Spinnaker3" project (100759174) is co-funded by the European Regional Development Fund (ERDF) and co-financed by tax revenue on the basis of the budget approved by the Saxon state parliament.
Project name
Spinnaker3 - event-based hardware platform for Artificial General Intelligence
Application number : 100759174
Brief description
The widespread use of AI is generating ever higher economic and social costs due to the increasing demand for electrical energy. Significantly more efficient AI hardware modeled on our brains can counteract this trend. However, such neuromorphic systems are still nowhere near the performance
capabilities of current AI solutions. One of the few large-scale systems is SpiNNaker. Like our brain, it focuses on efficient communication and event-based
event-based processing and is generally programmable, which enables a wide range of applications. The aim of this project is to develop a new hardware architecture based on the established SpiNNaker idea and thus create the basis for the world's most efficient hardware platform for artificial general intelligence (AGI). Spinnaker3 will combine efficient communication and event-based processing with novel types of neural networks, as well as integrating algorithms of symbolic AI and other AI approaches. One focus is on algorithms for continuous learning. Spinnaker3 combines the potential of three areas in a unique solution:
- It creates hardware for more efficient AI models,
- it takes developments in neuromorphic systems further and
- it forms a new, generic computer architecture with broad applicability.
Project duration
Period: 25.07.2025 to 31.05.2028
Contact
© Prof. Dr.-Ing. habil. Christian Mayr
Mr Prof. Dr.-Ing. habil. Dr. sc. nat. Christian Georg Mayr
Send encrypted email via the SecureMail portal (for TUD external users only).