Crafting better digital systems with ECE PhD student Jie-Fang Zhang

Growing up on the other side of the world in the bustling city of Taipei, Taiwan, Jie-Fang Zhang discovered he had a passion and a talent for digital system design. This year, the first-year ECE PhD student won the Chia-Lun Lo Fellowship – an award that assists outstanding students in Rackham programs who have earned a previous degree from a university in Taiwan.

Zhang’s research is focused on building efficient hardware solutions and designing domain-specific architectures for emerging applications. These applications, such as computer vision and machine learning, currently rely on general purpose processors like central processing units (CPUs) and graphics processing units (GPUs). However, such devices are not well-suited to these applications, which limits their performance and leads to higher power consumption. Zhang hopes to design new architectures that will better fit the application.

“Think of it like having a lot of different cars,” Zhang said. “You have sports cars, you have SUV’s, you have trucks. Each one is good at something, but no car is good at everything. We’re trying to find the best car for the job.”

Zhang works with Prof. Zhengya Zhang (no relation) as part of the VLSI Signal Processing Group. He’s designing a chip prototype for sparse deep learning acceleration, which could be applied to object detection and image recognition.

“I think that the most exciting thing is that I get to experiment with the entire process from designing the architecture to building the prototype to testing the chip,” Zhang said. “I have a lot of freedom to decide what I want to do.”

Zhang chose U-M for its strong reputation in the fields of integrated circuit design and computer architecture. While he’s only in his first year of the PhD program, Zhang has been a wolverine since 2016. He began as a CSE master’s student, earning his degree last year.

“U-M offers plenty of interesting courses, and I get fundamental knowledge and a solid background about the field,” Zhang said. “This was definitely a very good choice.”

To learn more about Zhang and the work of the VLSI Signal Processing group, visit their website.