Neurogrid: Simulating a million neurons and a billion synapses with sixteen neuromorphic chips
-
1
Stanford University, United States
Large-scale brain simulations link high-level cognitive phenomena to low-level biophysical mechanisms, helping neuroscientists understand how cognition emerges from the brain’s wetware. These simulations use a digital approach to model ion-channels that was pioneered by Hodgkin and Huxley in the 1940s. Computer performance has increased over a billionfold since then (Moore’s Law), enabling supercomputers to simulate networks with millions of neurons connected by billions of synapses in real-time. This scale is only about 0.01% of the human cortex, however. And clock speed has plateaued in recent years, putting real-time full-brain simulations out of reach for the foreseeable future—even for the fastest supercomputers. Fortuitously, with the recently developed ability to emulate (i.e., simulate in real-time) various types of ion-channels, the analog technique pursued by neuromorphic engineers over the past two decades has matured. The brain can now be modeled using subthreshold analog computation to emulate ion-channel activity and asynchronous digital communication to route synaptic connections. Neurogrid, an entirely clockless system with sixteen mixed-analog-digital chips created at Stanford, emulates a million cortical neurons connected by six billion synapses. It rivals the performance of 20 IBM Blue Gene L racks on this particular task while consuming five orders of magnitude less energy. By providing an affordable platform to perform large-scale simulations, Neurogrid is helping neuroscientists vet various hypotheses about how the brain works.
Keywords:
neuromorphic chips,
neurogrid,
Neurons,
Large-scale brain simulations,
how cognition emerges from the brain’s wetware
Conference:
Bernstein Conference 2012, Munich, Germany, 12 Sep - 14 Sep, 2012.
Presentation Type:
Invited Talk
Topic:
Other
Citation:
Boahen
K
(2012). Neurogrid: Simulating a million neurons and a billion synapses with sixteen neuromorphic chips.
Front. Comput. Neurosci.
Conference Abstract:
Bernstein Conference 2012.
doi: 10.3389/conf.fncom.2012.55.00040
Copyright:
The abstracts in this collection have not been subject to any Frontiers peer review or checks, and are not endorsed by Frontiers.
They are made available through the Frontiers publishing platform as a service to conference organizers and presenters.
The copyright in the individual abstracts is owned by the author of each abstract or his/her employer unless otherwise stated.
Each abstract, as well as the collection of abstracts, are published under a Creative Commons CC-BY 4.0 (attribution) licence (https://creativecommons.org/licenses/by/4.0/) and may thus be reproduced, translated, adapted and be the subject of derivative works provided the authors and Frontiers are attributed.
For Frontiers’ terms and conditions please see https://www.frontiersin.org/legal/terms-and-conditions.
Received:
18 Sep 2012;
Published Online:
12 Sep 2012.
*
Correspondence:
Prof. Kwabena Boahen, Stanford University, Stanford, United States, boahen@stanford.edu