From New Professor Rajit Manohar, Brain-Inspired Computing
Extremely powerful and efficient at processing information, the human brain has much to offer anyone looking to build a better computer chip. That’s the thinking behind TrueNorth, a 4-square-centimeter chip with 5.4 billion transistors, and 1 million "neurons" that communicate via 256 million "synapses."
One of its designers, Rajit Manohar, is the new John C. Malone Professor of Electrical Engineering and Computer Science. Manohar, who started Jan. 1, was previously on faculty at Cornell from 1998 to 2016. He is the recipient of an NSF CAREER award, nine best paper awards, nine teaching awards, and was included in MIT Technology Review's list of “35 Innovators Under 35” for his work on low-power microprocessor design. He is also the founder of Achronix Semiconductor, a company that specializes in high-performance asynchronous field programmable gate arrays (FPGA) chips.
Asynchronous circuits are his primary area of research. Typically, digital devices rely on synchronous circuits – those with built-in clocks that allow the same amount of time for the completion of each computational function. It’s reliable, but it also means that the system can run only as fast as the slowest function in the chain. In asynchronous systems, each function is allowed as little or as much times as needed to complete its task. “It’s like a relay race - you hand the baton to the person when you’re there,” he said.
All of these functions working in parallel – similar to how the brain operates – allow for greater complexity and use much less energy. Through his work with asynchronous circuits, Manohar came to work with a team of IBM researchers in a years-long collaboration that resulted in TrueNorth. We still don’t know exactly how the brain works, Manohar said, so it’s a stretch to call TrueNorth a copy of the brain’s functions. But neuroscience has given us a much better understanding of what’s happening in the brain, and that information inspired the architecture of the TrueNorth chip.
The field of neuromorphic computing – systems with electronic circuits mimicking neurological architectures – has been around since the 1980s. The computing power of the TrueNorth chip, however, is unprecedented. To see what kind of real-world applications TrueNorth might have, the research team developed a multi-object detection and classification application and tested it with two challenges: one was to detect people, bicyclists, cars, trucks, and buses that appear periodically on a video; the other was to correctly identify each object. TrueNorth proved adept at both tasks.
“The chip is really good at pattern recognition – it does it much more efficiently than a conventional computer chip,” Manohar said.
Now, he and the other researchers are working on a new chip that improves upon TrueNorth by performing certain computations even more efficiently. He predicts it won’t be long before this kind of technology ends up in everyday devices.
“These neuro-computing algorithms currently provide state-of-the-art performance for tasks like object detection, recognizing faces – tasks that a lot of companies care about today,” he said. “Imagine having photos or videos that you search for in the same way that you search for text today; these types of chips are way more efficient at that kind of computation.”