In the HiveMind project, a team of four researchers from two Fraunhofer Institutes (IIS and EMFT) develop a massively parallel scalable computer architecture for event-driven computing.
The project, which is part of the SPRIN-D “New Computing Concepts” challenge, explores a computing concept based on the asynchronous exchange of brief massages or events, which can be broadcast between a large number of simple processors. The key challenges addressed in this project are the development of a scalable routing infrastructure, algorithms to optimize the routing, and an event-driven paradigm to model and program the system.
This paradigm is inspired by neuromorphic hardware accelerators for spiking neural networks, but it may prove suitable for much broader applications, including, for example, constraint satisfaction or other graph problems.
Participating Institutes:
