“Democratizing Integrated Circuit Designs with Deep Learning”
Thursday, Sept. 14 at 1:00pm
LAR 234

Abstract

As we are near the end of Moore’s Law scaling, there is a great need to design various kinds of customized accelerators for much higher performance and energy efficiency. Given the success of deep learning in many domains in recent years, we apply it to chip designs, especially with high-level synthesis for automated RTL generation for high-quality accelerator designs. Using a set of deep learning techniques, such as graph-based neural networks, meta learning, and cross-modality learning, as well as microarchitecture guided optimization for regular structures, such as systolic arrays and stencil computation, we show that it is possible to automate IC designs so that most software programmers can design their own customized ICs for a wide range of applications.

Biography

Dr. Jason Cong is Volgenau Chair for Engineering Excellence Professor at the UCLA Computer Science Department (and a former department chair), with joint appointments from the Electrical and Computer Engineering Department. He is the director of Center for Domain-Specific Computing (CDSC) and the director of VLSI Architecture, Synthesis, and Technology (VAST) Laboratory. Dr. Cong’s research interests include novel architectures and compilation for customizable computing, synthesis of VLSI circuits and systems, and quantum computing. He has over 500 publications in these areas, including 17 best paper awards, and four papers in the FPGA and Reconfigurable Computing Hall of Fame. He and his former students co-founded AutoESL, which developed the most widely used high-level synthesis tool for FPGAs (renamed to Vivado HLS and Vitis HLS after Xilinx’s acquisition). He is a member of the National Academy of Engineering, and a Fellow of ACM, IEEE, and the National Academy of Inventors. He is the recipient of the 2022 IEEE Robert Noyce Medal for fundamental contributions to electronic design automation and FPGA design methods.

About the Series

The Walter G. Fredrickson Distinguished Interdisciplinary Lecture Series is named in honor of Walter Fredrickson (BSEE ’57), whose long and illustrious career in the world of electronics and engineering serves as the inspiration for the interdisciplinary lecture series. After he received his BSEE (with a specialization in electronics), Fredrickson entered active duty with the Air Force Security Service. Upon finishing his service, he completed some graduate work in computer circuits and logic at the University of Maryland, ultimately joining Radiation, Inc., a small electronics company in Melbourne, Florida. There he designed airborne and ground based telemetry data systems, which led to engineering management responsibility of complex computer-based data processing, display and communications systems for military and space applications.

In 1968, after acquisition of Radiation by Harris-Intertype, he was selected to lead the development of a new line of text editing and ad layout products for the newspaper industry, which became the precursor of modern word processing and publication products and software. Mr. Fredrickson served as Chairman of the Florida High Technology & Industry Council’s Applied Research Program, and received the Governor’s Award for leadership of this industry/academia program.

During the last decade of Mr. Fredrickson’s nearly 40 year career with Harris, he became the senior corporate executive responsible for assisting all Harris divisions to improve the success rate of their new product/program activities, through emphasis on the product-to-market process with associated metrics, and a climate for innovation among interdisciplinary hardware, software and marketing teams.