Stanford bioengineers have developed a new type of circuit board based on the human brain. The chip is faster and consumes less power than the average personal computer.
The new circuit board is the product of the Stanford research team's efforts to mimic at least part of the human brain's efficiency as a computer. While modern computers are smaller and much more powerful compared to the early room sized computers that were used decades ago, even the most expensive personal computer available today is still a far cry from the human brain in terms of performance. Even when compared to the brains of small animals such as mice, personal computers still lag behind considerably. Experts estimate that a mouse's brain is still around 9,000 times faster than a personal computer trying to emulate it.
Aside from being much slower than its biological counterparts, personal computers are also relatively power-hungry devices. Compared to a biological brain, a personal computer uses 40,000 times more energy to operate.
"From a pure energy perspective, the brain is hard to match," said Stanford associate professor of bioengineering Boahen.
The chip developed by the researchers however, may be the key to unlocking the powerful computational abilities of biological brains and translating them for use with modern technology. It is said to be 9,000 times faster than a normal commercially available computer and uses only a fraction of the power required to run a personal computer. The researchers hope that the new development will also spur new advances in robotics, prosthetic technology for amputees and computing in general.
The new chip has been christened as the Neurogrid. This circuit board is comprised of 16 "Neurocore" microchips that were designed by the team specifically for this purpose. Since the chip is based on the human brain, the 16 cores can mimic the functions of 1 million neurons as well as the large number of synaptic connections that can be made between the neurons.
Currently, the Neurogrid is roughly the size of a tablet computer. As the technology is refined however, Neurogrids of the future may be a lot smaller and quite possible, faster and even more power efficient as well.
"Right now, you have to know how the brain works to program one of these," said Boahen, gesturing at the $40,000 prototype board on the desk of his Stanford office. "We want to create a neurocompiler so that you would not need to know anything about synapses and neurons to able to use one of these."
The research was funded by the National Institutes of Health with the hope that the technology could soon be used to create robotic prosthetic limbs for amputees and people who suffer from paralysis. In the future, the researchers also hope to use the technology to control an advanced humanoid robot in a manner similar to the way the human brain makes complex movements possible.
