Neurogrid, a new circuit board modeled after the human brain, is far faster and uses less energy than traditional designs. The new microchip is able to process information 9,000 times faster than the circuit boards in today's computers.
Stanford bioengineers developed the new board by mimicking the plethora of connections found within the human brain. Despite their power, computers pale in comparison to even simple brains, in many key aspects. Computer simulations of the functioning of the brain cortex of a mouse take 9,000 times longer to process than the actual biological event.
"From a pure energy perspective, the brain is hard to match," Kwabena Boahen of Stanford University wrote in a press release announcing the development.
The Neurogrid board consists of a network of 16 Neurocore chips. The components are capable of simulating the functions of neurons, connected together through synapses. The boards are small enough to easily hold in your hands.
This structure "Real time - for the first time - using 16 Neurocores integrated on a board that consumes three watts," researchers wrote in the article announcing development of the new technology.
Challenges facing the new board design will include the making the device easier to operate. Previous detail knowledge of neurophysiology is required to run the current design. The advanced circuit board will also need to become less expensive. Prototypes of the new design cost $40,000 each, making them prohibitively expensive - so far - for the consumer market.
Neurocores each have 65,536 electronic neurons, which are manufactured using 15-year-old technology. By switching to more efficient modern methods of production, Boahen believes he may be able to cut the cost of the boards by 99 percent, to just around $400.
IBM is working on similar technology, called SyNapse, which will connect 256 digital neurons, each with 1,024 connections. They hope to develop larger chips, capable of even greater processing speeds and efficiencies.
Future applications of the technology could include weather and climate forecasting, making models more accurate and reliable. Computer simulations of how drugs act within the human body could also be made more efficient by use of the new circuit board design. Robotics and prosthetic limbs could each see significant advances from the development of this new chip.
"The human brain, with 80,000 times more neurons than Neurogrid, consumes only three times as much power. Achieving this level of energy efficiency... is the ultimate challenge neuromorphic engineers face," Boahen wrote in the press release.
