Two Georgia Tech School of Electrical and Computer Engineering (ECE) students took home top awards at the Semiconductor Research Corporation (SRC) TECHCON 2024 in Austin, Texas this fall.

Ph.D. candidate Zishen Wan won the Best Presentation Award, while fourth-year electrical engineering student Avanish Narumanch placed top three in the Undergraduate Research Awards.

Wan presented his research, “Heterogeneous 3D Integrated CIM for Factorization with Holographic Perceptual Representations,” which aims to improve AI systems' ability to perceive and reason by disentangling complex sensory signals.

"The research enhances AI systems' ability to disentangle complex sensory signals, which is critical for advancing perception and reasoning in future cognitive AI systems."

Along with the award, Wan was one of just four students selected to present their research to the SRC director and keynote speaker.

The research presents a heterogeneous 3D integrated in-memory compute engine to breakdown complex object representations. The technology is able to store and recognize complex patterns through holographic perceptual representations, which holds detailed information similar to that of a hologram.

This process can store a lot of information in a way that mimics human perception.

The system uses a technology called "memristors," which are tiny devices that can remember and adjust their electrical resistance depending on the task. These memristors have built-in stochasticity or non-volatility, combined with volatile memory as a heterogeneous system, enabling more effective simulation of natural systems and better handling of uncertainty in computing.

The results have been very promising. Wan’s design has shown a five-order magnitude increase in factorization capacity. It’s also more efficient, packing in five and a half times more computing power into almost six times less space than 2D designs, while using 20 percent less energy.

Narumanch was recognized for his research on ways to achieve faster and more energy-efficient memory devices. It focused on spintronic memory devices, which are non-volatile, low-power, and scalable electronic devices that use electron spin to store information.

The goal of the project was to reduce switching time and energy consumption without sacrificing reliability.

Both students completed the research through the JUMP 2.0 COCOSYS Center.

Wan’s work was in collaboration with senior research engineer Mohamed Ibrahim, Associate Professor Tushar Krishna, and Professor Arijit Raychowdury, along with fellow ECE Ph.D. students Che-Kai Liu (co-first author), Hanchen Yang, and Samuel Spetalnick.

This is the latest of many awards his work has received, which include being named a 2023 Machine Learning and Systems Rising Star and 2024 Cyber-Physical Systems Rising Star.

Narumanch worked on his research under the guidance of ECE Ph.D. candidate Md Nahid Haque Shazon, who works in ECE Professor Azad Naeemi’s Nanoelectronics Research Lab.

SRC TECHCON is the annual flagship technical conference hosted by the SRC. The conference draws top researchers, scientists, and a large field of students from across the United States and world to present cutting-edge research in the field of semiconductor electronics.