IBM has unveiled two prototype cognitive computing chips that are designed around the way brains work.
IBM has unveiled two prototype cognitive computing chips that are designed around the way brains work, as part of its Synapse project. Photo credit: IBM Research
The company said on Thursday that its first neurosynaptic computing chips, which have been fabricated and are being tested, use advanced algorithms and silicon circuitry to replicate the brain's neurons, synapses and axons. IBM's cognitive computing architecture integrates hardware and software, marking a "critical shift" away from the traditional von Neumann architecture of separate CPU and memory, the company said.
"This is a major initiative to move beyond the von Neumann paradigm that has been ruling computer architecture for more than half a century," project leader Dharmendra Modha said. "Future applications of computing will increasingly demand functionality that is not efficiently delivered by the traditional architecture."
Modha described the chips as "another significant step in the evolution of computers from calculators to learning systems, signalling the beginning of a new generation of computers and their applications in business, science and government".
IBM Research's Systems of Neuromorphic Adaptive Plastic Scalable Electronics (Synapse) project is being conducted in collaboration with US institutions including Columbia University, Cornell University, University of California, Merced, and University of Wisconsin, Madison.
The project has been going since 2008, when US defence agency Darpa awarded $4.9m in funding. Darpa has now also given $21m (£12.7m) to fund the next phase of the Synapse project, IBM said on Thursday.
The idea behind Synapse is to create a multi-sensory system, with size and power consumption to rival those of the human brain, that can dynamically 'rewire' itself as needed. The team is using silicon to replicate neurons, the nerve cells that transmit information; synapses, the junctions that let neurons transmit that information between each other; and axons, the part of the neuron that conducts the information.
The two chips unveiled on Thursday were fabricated using a 45nm SOI-CMOS process and each contain 256 artificial 'neurons'. One of the cores contains 262,144 programmable 'synapses', while the other contains 65,536 of what IBM calls 'learning synapses'. Ultimately, IBM wants to build a system with ten billion neurons and a hundred trillion synapses that will consume just one kilowatt of power.
According to IBM, the relatively spartan system it is showing off now has already been used to successfully demonstrate "simple applications like navigation, machine vision, pattern recognition, associative memory and classification".
"Imagine traffic lights that can integrate sights, sounds and smells and flag unsafe intersections before disaster happens or imagine cognitive co-processors that turn servers, laptops, tablets, and phones into machines that can interact better with their environments," Modha said.
Other IBM-suggested applications for future versions of the chips include the monitoring of the world's water supply using a network of sensors and actuators that would record and report temperature, pressure, wave height and other metrics, even issuing tsunami warnings.
"Similarly, a grocer stocking shelves could use an instrumented glove that monitors sights, smells, texture and temperature to flag bad or contaminated produce. Making sense of real-time input flowing at an ever-dizzying rate would be a Herculean task for today's computers, but would be natural for a brain-inspired system," IBM said.
The company has long been working on artificial intelligence research, most recently hitting the headlines with its quiz-show-winning supercomputer, Watson.
Get the latest technology news and analysis, blogs and reviews delivered directly to your inbox with ZDNet UK's newsletters.