SpiNNaker: A 1-W 18-Core System-on-Chip for Massively-Parallel Neural Network Simulation

The modelling of large systems of spiking neurons is computationally very demanding in terms of processing power and communication. SpiNNaker - Spiking Neural Network architecture - is a massively parallel computer system designed to provide a cost-effective and flexible simulator for neuroscience e...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:IEEE journal of solid-state circuits Jg. 48; H. 8; S. 1943 - 1953
Hauptverfasser: Painkras, Eustace, Plana, Luis A., Garside, Jim, Temple, Steve, Galluppi, Francesco, Patterson, Cameron, Lester, David R., Brown, Andrew D., Furber, Steve B.
Format: Journal Article Tagungsbericht
Sprache:Englisch
Veröffentlicht: New York, NY IEEE 01.08.2013
Institute of Electrical and Electronics Engineers
Schlagworte:
Applied sciences
Asynchronous interconnect
Biological system modeling
Brain modeling
chip multiprocessor
Computational modeling
Computer systems
Electric, optical and optoelectronic circuits
Electronics
energy efficiency
Exact sciences and technology
globally asynchronous locally synchronous (GALS)
Hardware
Integrated circuits
Integrated circuits by function (including memories and processors)
network-on-chip
Neural networks
neuromorphic hardware
Neurons
real-time simulation
Semiconductor electronics. Microelectronics. Optoelectronics. Solid state devices
spiking neural networks (SNNs)
System-on-chip
Transistors
Spiking neuron
Processor
Asynchronous interconnect
neuromorphic hardware
Modeling
Implementation
Multiprocessor
Microelectronic fabrication
Parallel system
globally asynchronous locally synchronous (GALS)
energy efficiency
Parallel processing
chip multiprocessor
Chemical mechanical polishing
Computer design
spiking neural networks (SNNs)
Large scale system
Globally asynchronous locally synchronous
Packet switching
network-on-chip
System design
System on a chip
Parallel architectures
Parallel computer
Neural network
Synapse
real-time simulation
Integrated circuit
Computer system
Simulator
Asynchronous transmission
ISSN:0018-9200, 1558-173X
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:The modelling of large systems of spiking neurons is computationally very demanding in terms of processing power and communication. SpiNNaker - Spiking Neural Network architecture - is a massively parallel computer system designed to provide a cost-effective and flexible simulator for neuroscience experiments. It can model up to a billion neurons and a trillion synapses in biological real time. The basic building block is the SpiNNaker Chip Multiprocessor (CMP), which is a custom-designed globally asynchronous locally synchronous (GALS) system with 18 ARM968 processor nodes residing in synchronous islands, surrounded by a lightweight, packet-switched asynchronous communications infrastructure. In this paper, we review the design requirements for its very demanding target application, the SpiNNaker micro-architecture and its implementation issues. We also evaluate the SpiNNaker CMP, which contains 100 million transistors in a 102-mm 2 die, provides a peak performance of 3.96 GIPS, and has a peak power consumption of 1 W when all processor cores operate at the nominal frequency of 180 MHz. SpiNNaker chips are fully operational and meet their power and performance requirements.
ISSN:0018-9200
1558-173X
DOI:10.1109/JSSC.2013.2259038