ASIC Draws First Industry Partners
Duke’s Alternative Sustainable & Intelligent Computing consortium gives industry partners early access to pioneering research
Electrical and Computer Engineering associate professor Yiran Chen will drive a new Industry University Co-operative Research Center (IUCRC) headquartered at Duke, with partner sites at Notre Dame and Syracuse Universities. The new center, named the Alternative Sustainable & Intelligent Computing (ASIC) consortium, will focus on emerging computing platforms for cognitive applications.
ASIC’s operational costs will be covered by $150,000 in yearly funding from the National Science Foundation, but the center is capable of drawing additional private industry investments; ASIC investors have the opportunity to leverage the consortium’s groundbreaking research in novel circuits, architecture, and computational models at a pre-competitive stage, with the potential for earning significant returns on their investments.
The combined expertise of ASIC’s three investigators—Duke ECE associate professor Hai “Helen” Li and associate professor Benjamin Lee are co-leads—has already proven attractive enough for three industry partners to make initial investments of $50,000 each.
Chen and Li lead the Duke Center for Evolutionary Intelligence (CEI), a research group that designs computer architecture and circuits that mimic the human brain in the way they convey information. Neuromorphic, or “cognitive,” computing employs compact architecture that allows tasks to be accomplished much more quickly, using less energy. While other cognitive computing researchers around the world focus on software alone, CEI occupies a tiny, influential niche in which its software development experience is matched by expertise in hardware and deployment.
Lee’s lab focuses on understanding the nature of cognitive computing workloads and how different business entities compete for resources within shared systems. For example, within a large company that aggregates and processes data for both search and mail applications, the two business units might be tempted to each purchase its own computing cluster to guarantee performance when they need it. Lee’s work has shown that they would spend less money and use less power if they pooled their processors, memory, and network resources.
As a team, the three Duke researchers, together with fellow ECE faculty members Robert Calderbank, Krishnendu Chakrabarty, Xin Li and Miroslav Pajic, can create computing systems that are markedly stronger in architecture and performance than those found in the mainstream.
The other two sites, Notre Dame and Syracuse, will focus on the applications of cognitive computing platforms in medical and military sectors, respectively.
“The knowledge and expertise developed via ASIC will effectively enable a boom in cognitive applications,” said Chen.