Laboratory for Emerging Devices and Circuits Team Wins Best Paper Award for AI Computing Memory Research

Georgia Tech School of Electrical and Computer Engineering (ECE) professor Shimeng Yu and his team at the Laboratory for Emerging Devices and Circuits won the Association for Computer Memory (ACM) Transactions on Design Automation of Electronic Systems (TODAES) 2024 Best Paper Award.

The prestigious award recognizes the best paper published in the TODAES, the ACM's flagship publications in the area of electronic design automation (EDA).

Yu accepted the award at the 61st Design Automation Conference in San Francisco, Calif. in June.

This is the second consecutive year Yu’s team won an award for research printed in a flagship publication in the area of EDA, and the third year in a row research from ECE has received such an honor.

The paper titled, “Hardware-aware quantization/mapping strategies for compute-in-memory accelerators,” analyzed different combinations of settings for emerging non-volatile memory (eNVM) technologies.

This new technology is important for mixed-signal Compute-in-Memory (CIM) accelerators, which are very energy efficient, thus making them crucial for artificial intelligence hardware design, which are notoriously resource intensive platforms.

Ultimately, the research found the right configuration and settings can significantly improve output and efficiency. Yu’s team was able to achieve an increase in processing speed by up to 60 percent, while doubling the energy efficiency and reducing overall hardware size by up to 25 percent.

The findings provide design guidelines to engineers who continue to research eNVM and CIM technology.

Yu co-authored the paper with ECE Ph.D. graduates Shanshi Huang and Hongwu Jiang, who are now both assistant professors in Hong Kong University of Science and Technology.

Yu’s lab won the IEEE’s Donald O. Pederson Best Paper Award in 2023 for their research on an end-to-end benchmark framework to evaluate state-of-the-art CIM accelerators. The award honors the best paper in the IEEE’s Transactions on Computer-Aided Design of Integrated Circuits and Systems, the flagship journal of the IEEE Council on Electronic Design Automation.

In 2022, ECE Professor Sung Kyu Lim and his research team won the Donald O. Pederson Best Paper Award for their paper on a physical design tool named Compact-2D that automatically builds high-density and commercial-quality monolithic three-dimensional integrated circuits.

Yu also recently received a 2023 Intel Outstanding Research Award for his work on a chip that will help quantify uncertainty that is beyond the capabilities of existing binary computing systems, and improve computing robustness.