cfaed Seminar Series

Our understanding of the human brain and its tremendously energy-efficient information processing is still quite incomplete. One reason is the immense complexity of the brain, making it difficult to understand and simulate. The latter difficulty is tackled by the

SpiNNaker2 system, aimed for simulating brain-size spiking neural networks in real-time. This talk gives some insight into the design of the system and shows how it combines biological inspiration with cutting-edge chip design. Design and abstraction methods for handling the complexity of a 144 core multi-processor chip are introduced as well. The resulting SpiNNaker2 system will be a flexible, yet energy-efficient platform for testing models of neuro-inspired information processing in real-time, as well as other applications for embedded, real-time processing.