UNIVERSITY PARK, Pa. — A paper that explores possible computer performance breakthroughs by using the new computation paradigms enabled by new technologies has won the 2017 Institute for Electrical and Electronics Engineers (IEEE) Transactions on Multi-Scale Computing Systems (TMSCS) Best Paper award.
“Enabling New Computation Paradigms with HyperFET - An Emerging Device,” which was funded by the National Science Foundation’s Expeditions-in-Computing program, was a collaboration with Wei-Yu Tsai, a recent computer science and engineering graduate. Also collaborating was Xueqing Li, Matthew Jerry, Baihua Xie, Nikhil Shukla, Huichu Liu and Nandhini Chandramoorthy, graduate students in the School of Electrical Engineering and Computer Science; and John Sampson, professor, and Vijay Narayanan, distinguished professor, both of computer science and engineering; as well as several others. It was the last work co-authored by the University of Pittsburgh’s Steve Levitan, who recently passed away.
Tsai said there are three key elements that can be learned from the research: It is the first work that provides thorough modelings from device-, circuit- and system-level operations of the oscillations and couplings of the proposed nano-scale voltage controlled oscillators; it is the first work that shows the structure and circuit design of the network constructed by the nano-scale spiking neuron cells; and it is supported by the real measurement data — it provides the evaluations of the systems built with the proposed oscillators and the projections of the performance improvements.
“This work helps me move from pure circuit-level design to system-level design,” said Tsai, about the work for the project. “Starting from the circuit simulation which I already have expertise in, I've learned to build a system and stand from the application perspective to evaluate the performance and efficiency. It also pushes me to look deeper into neuromorphic hardware, and further to the machine learning/deep neural networks areas.”
"IEEE Transactions on Multi-Scale Computing Systems" (TMSCS) is a peer-reviewed publication devoted to computing systems that exploit multi-scale and multi-functionality. Contributions to TMSCS must address computation of information and data at higher system-levels for processing by digital and emerging domains.
A copy of the award winning work can be found here.