Neurogrid is a piece ofcomputer hardware that is designed specifically forsimulation of biological brains. It usesanalog computation to emulateion channel activity, and digital communication to softwire structured connectivity patterns. Neurogrid simulates one millionneurons[1] and six billionsynapses in real time. The neurons spike at a rate of ten times a second. In terms of the total number of simulated neurons, it rivals simulations done by theBlue Brain Project. However, by running the simulation of whole neurons, instead of simulation on molecular level, it needs only one millionth of Blue Brain's power. The entire board consumes less than two watts of electrical energy.
Neurogrid was designed and built by theBrains in Silicon group atStanford University, led byKwabena Boahen. The Neurogrid hardware was first up and running in late 2009. Since then, it has been used to start performing simulation experiments.[2]
The Neurogrid board contains sixteenNeurocores, each of which has 256 x 256 silicon neurons in an 11.9 mm x 13.9 mm chip. An off-chipRAM and an on-chip RAM (in each Neurocore) softwire horizontal and vertical cortical connections, respectively. With 61 graded and 18 binary programmable parameters, common to all of its silicon neurons, a Neurocore can model a variety of spiking and interaction patterns.[3]