Yang 'Cindy' Yi and students lay foundation for future of chip design

Whether or not you have a technical background, you’ve probably become accustomed to Moore’s Law, the observation that computing power doubles every two years. That expansion of computing ability has led to exponential technological growth. But creating better efficiency in technology isn’t just something that happens automatically.

Those technological foundations are laid years in advance in places like the new BRain Inspired Computing, Communication, and Security Lab (BRICCS), which is the centerpiece of the new computer systems research area at Virginia Tech’s Institute for Advanced Computing in Alexandria. It starts with researchers like institute faculty member Yang “Cindy” Yi and her students attacking the potential areas of improvement from every angle, at the smallest, most refined level. Yi is a professor and Bradley Senior Faculty Fellow of the Bradley Department of Electrical and Computer Engineering.

That can mean trying to find ways to reduce how much power a chip uses. Or it can mean finding ways of using the same power to deliver better performance. Everything converges in the name of developing smaller, faster, more efficient chips that can lead to smaller, faster, more wearable devices.

“Within the institute, we span the full computing stack, from AI [artificial intelligence] applications and the data centers that support them, down through systems, interfaces, and microarchitectures to the design of chips and emerging substrates,” said Kirk Cameron, the institute's managing director. “Cindy’s work focuses on the evolution of computing architectures to support emerging applications envisioned by domain scientists — applications that will ultimately shape next-generation systems, devices, and industries.”

When it came to a recent project for a leading technology company, just how much smaller and faster were they hoping to get the hardware from its existing standards?

“Our industry collaborator is targeting a 100 times reduction in power consumption along with a much smaller silicon footprint plus higher speed and wider bandwidth,” said Yi. “What they ask the university to do is create what the next future product could be, determine how much power we can really save, and how much speed we can improve.”

Yi’s team members delivered, designing and manufacturing a new chip within a year. They did so by approaching the problem in a few different ways. By replacing a camera-based eye-tracking system with a micro-electromechanical system (MEMS)-based approach, they were able to move everything into a single module. This calculates indirect time of flight, a method that estimates distance by measuring the phase difference between emitted and reflected modulated light signals. Yi and her team were able to reduce the power load required to run this indirect time of flight system by 100 to 1,000 times, taking a device that formerly required constant power to run and transforming it into something that could be portable and battery-powered.

“The current bottleneck today is power consumption,” said Yi. “Most current devices still need to recharge after two to three hours of usage. What we are working on is to embed the computing and sensing modules into glasses. That’s why these systems rely on highly integrated, ultra‑low‑power components, so they can operate efficiently without adding bulk or requiring external modules.”

The hope is that the technology from these labs might find their way into commercial products in 10 or 20 years. It’s a vision of the future of what augmented or virtual reality devices might be capable of producing, for everything from search and rescue operations to daily use.

The BRICCS Lab, which brings together eight faculty members from the College of Engineering and the Pamplin College of Business, is particularly well suited to embrace these kinds of challenges. It houses a host of high-end equipment, from oscilloscopes to spectrum and network analyzers. But the centerpiece is a top-of-the line, armoire-sized engineering probe system usually found in research and dvelopment labs and semiconductor foundries for research‑grade chip evaluation. It is so large, it required pulling the doors off of the lab to get it inside.

“My students are very excited about this probe station because it has the capability to measure nanometer and micrometer chips,” said Yi. “It’s great for them to have hands-on experience in chip measurements and testing. When they go to industrial companies, they can show that they already have the experience through the training and the research projects at the university.”

The team works at the forefront of neuromorphic computing, wireless communication, and security, advancing next‑generation intelligent systems through interdisciplinary research. Ph.D. student Ramashish Gaurav was getting his master’s degree at the University of Waterloo, Canada in January 2022 and could only find a handful of schools that offered neuromorphic computing then, including Virginia Tech.

“I’m getting to do things that I could have only imagined doing had I not been here,” he said. “Working on the next generation of AI systems, neural networks, spiking, neuromorphic chips, hardware.”

Ph.D. student Alberta Dadeboe said the lab and its capabilities are things you sort of have to see in person to really understand the scope of what she and her classmates are working on, something she loves to do for visitors.

“It’s always exciting to show them the new equipment we have and run our demos,” she said. “It’s great to see the excitement on their faces when we show them what we’re doing.”

While Yi hopes to use the BRICCS Lab to continue her work into areas such as heterogeneous integration, trying different circular chip components, stacking them in different ways, and working to optimize drivers to achieve even better performance, the most important components of her lab aren’t pieces of equipment. 

“Our students are really the most precious part of our group,” she said.