IBM Research Lab Unveils Chip Design Modeled After Human Brain
The chip attempts to recreate the behavior of neurons and synapses using advanced algorithms and silicon circuitry. These cognitive computing prototype chips could reduce the power consumption and space required for a microprocessor by orders of magnitude over today's technology, IBM said in a statement.
The chip might strike some as nothing more than science fiction, but Tim Ulmen, principal at Midwest IT solutions group, a Wichita, Kan.-based system builder, suggests a second glance. "We might think [the prototype] is far-fetched, but we've seen incredible things happen with technology before. When I was a kid we wouldn't have expected to be able to have phones in our pockets, but here we are," he said.
The IBM Research group has two working prototypes of the new chip, both fabricated using 45-nm, Silicon-on-Insulator/Complementary Metal Oxide Semiconductor (SOI-CMOS) technology. The chips boast integrated memory, computation and communication modeled after biological synapses, neurons and axons, respectively. According to IBM, the prototypes successfully run simple applications including navigation, pattern recognition and classification.
Computing in the future will require a massive evolution in microprocessors, said Dharmendra Modha, project leader for IBM Research. "Future applications of computing will increasingly demand functionality that is not efficiently delivered by the traditional architecture," he said in a statement. "These chips are another significant step in the evolution of computers from calculators to learning systems.”
While IBM strives to change the basic building blocks of processor technology, Ulmen thinks software confines computing capability more than hardware. "There are very few software processes that utilize any real computing power," he said.
IBM worked software into the mix of the new processor by creating an on-chip network with a single integrated system of hardware and software. The end goal is microprocessor architecture with no set programming, integrated memory and parallel processing to mimic the human brain. One prototype contains 262,144 programmable synapses, while the other has 65,536 learning synapses, IBM said.
IBM received approximately $21 million in new funding from DARPA [the U.S. Defense Advanced Research Projects Agency] for the second phase of its chip development project, known as the Sytems of Neuromorphic Adaptive Plastic Scalable Electronics (SyNAPSE) project. In conjunction with university research collaborators, IBM hopes to build upon the prototypes created in the first phase of the project to invent a small, low-power system that rewires itself based on experience.
“Imagine traffic lights that can integrate sights, sounds and smells and flag unsafe intersections before disaster happens," said Modha. "Or imagine cognitive co-processors that turn servers, laptops, tablets, and phones into machines that can interact better with their environments."
Researchers from IBM, Columbia University, Cornell University, University of California Merced and University of Wisonsin Madison are testing the cognitive computing prototype chips at IBM research labs in Yorktown Heights, N.Y. and San Jose, Calif.