Breaking the Memory Wall with Optical Interconnects and In-Memory Computing

Abstract: The explosive growth of emerging applications in data analytics and high-performance computing is placing unprecedented demands on today’s computing systems. Consequently, hyperscale computing systems now face critical bottlenecks in data transfer rather than computational power. I will talk about two projects we are working on to address this challenge. In the first project, we are developing glass panel-enabled 3D optical interconnects to significantly increase bandwidth density and energy efficiency for panel-scale computing. We employ volumetric waveguides and 3D routing in glass to enhance the density of waveguides and optical components and eliminate the shoreline density limitations in planar photonics. In the second project, we are developing non-volatile memory devices for in-memory computing to greatly reduce the data shuffling between processors and memory. With numerous memory devices developed in recent years, there is a need for benchmarking their performance in the deep neural network computing. We developed such a comprehensive methodology for such benchmarking and found that electrochemical memory has great potential for high-performance large-scale analog in-memory computing.

Additional Information:

BIO: Ning Li is an associate professor in the Department of Electrical Engineering and Materials Research Institute at The Pennsylvania State University. He was a research staff member at IBM T.J. Watson Research Center from 2010 to 2022. His research experience includes photonic components and links for communications and interconnects, heterogeneous integration of materials and devices for new applications, nonvolatile memories for in-memory computing. He was awarded more than 250 U.S. patents, many High Value Patent Awards, and multiple Master Inventor Awards. He published in scientific journals and conferences including Nature Photonics, Nature Communications, Advanced Materials, Optical Fiber Communication (OFC), etc. His work has been featured on Nature Research Highlight, Semiconductor Today, etc. He received his BS degree from Tsinghua University and PhD degree from The University of Texas at Austin.