Brain-inspired circuit board 9000 times faster than an average PC
Thu 1 May 2014
Another ‘the future is here’ article, this time about bioengineers at Stanford University who have developed Neurocore processors designed to act like a human brain – well, 1 million neurons at least – which makes them ‘9000 times’ faster than a PC and consume a fraction of the power.
It has packed 16 of them on to an iPad-sized board called the Neurogrid and uses analogue computation alongside digital – a key factor in keeping power consumption low. Not sure about the speed claims. It is suggested that the Neurogrid can do in real time what took the fourth most powerful computer 40 mins to do i.e. simulate 1 minute of brain activity. But it did this running at 5W rather than the supercomputer’s 1MW.
Anyway, cost is also an issue. Currently a Neurogrid costs about $40,000 but the team believes that could come down $400.
“Analog computation constrains the number of distinct ion-channel populations that can be simulated—unlike digital computation, which simply takes longer to run bigger simulations,” the Neurogrid website explains. “Digital communication constrains the number of synaptic connections that can be activated per second, unlike analog communication, which simply sums additional inputs onto the same wire. Working within these constraints, Neurogrid achieves its goal of simulating multiple cortical areas in real-time by making judicious choices.”