Researchers at IBM have created a one-of-a-kind chip/processor named TrueNorth that may in the long run exceed expectations of calculations done by today’s supercomputers. Details of the chip were first published in an article on Thursday in the journal Science.
The chip's electronic "neurons" can indicate when a sort of information, light, for instance, passes a certain limit. The neurons work in parallel and initiate sorting data into patterns suggesting when the light is growing brighter, changing shape or color.
The processor might in this manner have the capacity to perceive similar to how humans do, for example, perceive acts without cognizant thought which today's machines and robots battle to translate.
The chip holds 5.4 billion transistors, however draws only 70 milliwatts of force. Intel processors in today's Pcs and data centers may have 1.4 billion transistors and utilize significantly more power, say between 35 to 140 watts.
Conventional microprocessors and graphics processors in present age are equipped for performing billions of numerical operations a second; however the new chip system performs calculations scarcely a thousand times per second. As per IBM analysts, due to the vast number od circuits working in parallel, the new chip is equipped for performing 46 billion operations per second for every watt of energy utilized.
TrueNorth’s vast network of one million “neurons” is about as mind boggling as the brain of a honeybee. “It is a remarkable achievement in terms of scalability and low power consumption,” said Horst Simon, deputy director of the Lawrence Berkeley National Laboratory. Simon compared the new outline of the chip to the dawn of parallel supercomputers in the 1980s, saying it was similar to moving from a two-path street to a highway. Let’s hope TrueNorth star promises much better hi-tech future than that.
Yann Lecun, director of artificial intelligence at Facebook and a pioneer in neural networks, remarked that he was dubious whether IBM's methodology would ever outpace today's speediest commercial processors.
"The chip gives off an impression of being extremely restricted from various perspectives, and the execution is not what it appears," Mr. Lecun composed in an email sent to columnists from various publications. Specifically, he condemned as insufficient the testing of the chip's capability to detect moving subjects.
Several neuroscience scientists and computer scientists however, chose to believe otherwise. “The TrueNorth chip is like the first transistor,” said Terrence J. Sejnowski, director of the Salk Institute’s Computational Neurobiology Laboratory. “It will take many generations before it can compete, but when it does, it will be a scalable architecture that can be delivered to cellphones, something that Yann’s G.P.U.s (graphic processing unit) will never be able to do.”
IBM's research was monetarily funded by the Defense Advanced Research Projects Agency, a research arm of the Pentagon, under a program called Systems of Neuromorphic Adaptive Plastic Scalable, or SyNAPSE.
Dr. Gill Pratt, the program manager at the agency says, they are pursuing twin goals in its effort to create biological processors.