CCF:SHF:Medium: Automated End-to-End Synthesis for Programmable Analog & Mixed-Signal Systems

Georgia Tech Research Corporation, Atlanta GA

Investigators

Abstract

Automated digital-design tools have enabled a wide range of digital innovations, and automating mixed-signal and analog design tools have the potential for an equally wide range of electronic innovations. Today’s analog system design processes are impeding the development of new integrated circuits (ICs) processing due to the required manual effort. The emerging demand for low-power high-efficiency analog and mixed-signal-computing (e.g. machine learning) further is only increasing the bottlenecks imposed by the state of the art in analog system design. The success of this project would drastically decrease the design time, design cost, and design uncertainty of large-scale analog and mixed-signal ICs. These innovations will require educational innovation at multiple levels and could enable future commercial opportunities built around these tools. This project proposes a design-automation framework that enables end-to-end synthesis from the high-level description down to targeted configurable devices or custom silicon IC layouts. It lays a solid technology-independent foundation to abstract and automate analog design. This tool capability enables a new vision for analog and mixed-signal computing, similar to that has already seen in digital design spaces. This effort creases the first analog/mixed-signal high-level synthesis, i.e., from behavioral-level description such as Python, as well as text-based analog/mixed-signal mid-level synthesis. The co-design of mixed analog/digital architectures enables real-time to near-zero latency computing for ultra-low-energy applications, while expanding the architectural design space for analog circuits. This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.