Spinnaker

SpiNNaker ( Spiking Neural Network Architecture ) is a multi-core computer architecture designed by the Advanced Processor Technologies Research Group (APT) at the School of Computer Science, University of Manchester , [1] led by Steve Furber , to simulate the human brain (see Human Brain Project ). It is planned to use 1 million ARM processors (currently 0.5 million) [2] in a massively parallel computing platform based on spiking neural networks . [3] [3] [4] [5] [6] [7][8] [9] [10] [11]

The completed design is to be housed in 10 19-inch racks each rack holds 100,000 cores [12] The cards themselves holding the chips are held in 5 Blade Enclosures and Each Core Emulates 1000 Neurons . [12]

SpiNNaker is used as a component of the neuromorphic computing platform for the Human Brain Project . [13] [14]

See also

  • TrueNorth – a processor architecture designed for spiking neural networks.

References

  1. Jump up^ Advanced Processor Technologies Research Group
  2. Jump up^ Steve Furber interviewed on BBC Click
  3. ^ Jump up to:b Spinnaker Home Page , University of Manchester , retrieved 11 June 2012
  4. Jump up^ Furber, SB ; Galluppi, F .; Temple, S .; Plana, LA (2014). “The SpiNNaker Project”. Proceedings of the IEEE : 1. doi :10.1109 / JPROC.2014.2304638 .
  5. Jump up^ Xin Jin; Furber, SB ; Woods, JV (2008). “Efficient modeling of spiking neural networks on a scalable multiprocessor chip”. 2008 IEEE International Joint Conference on Neural Networks (IEEE World Congress on Computational Intelligence) . pp. 2812-2819. doi : 10.1109 / IJCNN.2008.4634194 . ISBN  978-1-4244-1820-6 .
  6. Jump up^ A million ARM cores to host the brain simulatorNews article on the project in theEE Times
  7. Jump up^ Temple, S .; Furber, S. (2007). “Neural systems engineering”. Journal of the Royal Society Interface . 4 (13): 193. doi :10.1098 / rsif.2006.0177 . A manifesto for the SpiNNaker project, surveying and reviewing the general level of understanding of brain function and approaches to building the model of the brain.
  8. Jump up^ Plana, LA; Furber, SB ; Temple, S .; Khan, M .; Shi, Y .; Wu, J .; Yang, S. (2007). “A GALS Infrastructure for a Massively Parallel Multiprocessor”. IEEE Design & Test of Computers . 24 (5): 454. doi : 10.1109 / MDT.2007.149 . A description of the Globally Asynchronous, Locally Synchronous (GALS) nature of SpiNNaker, with an overview of the asynchronous communications hardware designed to transmit neural ‘spikes’ between processors.
  9. Jump up^ Navaridas, J .; Luján, M .; Miguel-Alonso, J .; Plana, LA; Furber, S. (2009). “Understanding the interconnection network of SpiNNaker”. Proceedings of the International Conference on Conference on Supercomputing – ICS ’09 . p. 286. doi : 10.1145 / 1542275.1542317 . ISBN  9781605584980 . SpiNNaker is a multicomputer-based machine, showing the suitability of the packet-switched network for large-scale spiking neural network simulation.
  10. Jump up^ Rast, A .; Galluppi, F .; Davies, S .; Plana, L .; Patterson, C .; Sharp, T .; Lester, D .; Furber, S. (2011). “Concurrent heterogeneous neural model simulation on real-time neuromimetic hardware”. Neural Networks . 24 (9): 961-978. doi :10.1016 / j.neunet.2011.06.014 . PMID  21778034 . A demonstration of SpiNNaker’s ability to simulate different neural models (simultaneously, if necessary) in contrast to other neuromorphic hardware.
  11. Jump up^ Sharp, T .; Galluppi, F .; Rast, A .; Furber, S. (2012). “Power-efficient simulation of detailed cortical microcircuits on Spinnaker”. Journal of Neuroscience Methods . 210 (1): 110-118. doi : 10.1016 / j.jneumeth.2012.03.001 . PMID  22465805 . Four-chip, real-time simulation of a four-million-synapse cortical circuit, showing the extreme energy efficiency of the SpiNNaker architecture
  12. ^ Jump up to:b Video interview with Steve Furber by computerphile
  13. Jump up^ Calimera, A; Macii, E; Poncino, M (2013). “The Human Brain Project and Neuromorphic Computing” . Functional neurology . 28 (3): 191-6. PMC  3812737  . PMID  24139655 .
  14. Jump up^ Monroe, D. (2014). “Neuromorphic computing gets ready for the (really) big time”. Communications of the ACM . 57(6): 13-15. doi : 10.1145 / 2601069 .

Leave a Reply

Your email address will not be published. Required fields are marked *

Copyright computerforum.eu 2018
Shale theme by Siteturner