SpiNNaker - A Universal Spiking Neural Network Architecture
SpiNNaker Overview
The SpiNNaker project's architecture mimics the human brain's biological structure and functionality. This offers the possibility of utilising massive parallelism and redundancy to provide resilience in an environment of unreliability and failure of individual components.
In the human brain, communication between its computing elements, or Neurons, is achieved by the transmission of electrical "Spikes" along connecting Axons. The biological processing of the Neuron can be modelled by a digital processor and the Axon connectivity can be represented by messages, or information packets, transmitted between a large number of processors which emulate the parallel operation of the billions of Neurons comprising the brain.
The engineering of the SpiNNaker concept is illustrated in the System Diagram where the hierachy of components can be identified. Each element of the toroidal interconnection mesh is a multi-core processor known as the "SpiNNaker Chip" comprising 20 processing cores, or "Fascicle Processors" in an initial implementation. Each Fascicle is a complete processing sub-system with local memory and a DMA capability. It is connected to its local peers via a Network-on-Chip (NoC) which provides local high bandwidth communication and to other SpiNNaker chips via links between SpiNNaker chips. In this way the massive parallelism extending to thousands or millions of processors is possible.
The knowledge content and learning ability of the brain is embodied in its evolvable interconnection pattern; this routes a Spike generated by one Neuron to others which are interconnected with it by Axons and these interconnections are modified and extended as a result of the learning and processes.
In SpiNNaker the neural interconnection is controlled by a packet Router within each multi-core processor. Each transmitted packet representing a Spike contains information which identifies its source Neuron; this is used by a multi-core processor's Router to identify whether this packet should be routed to one of its contained Fascicle processors to respond, or should be routed on to one of the six adjacent multi-core processors connected to it as part of the overall SpiNNaker network.
Sponsors and Collaborators
