Open-AI Driven Open-source Open-access Sustainable ICs Design Flow
Tee Hui Teo, Maoyang Xiang, Emil Goh, Huan-Ke Hsu
Abstract
The work first demonstrates a design flow that uses open-source electronic design automation (EDA) software and open-access process design kits (PDKs) to design artificial neural network (ANN) circuits. Based on this design flow, a large-language model (LLM) is trained and used to design a three-phase pulse-width modulator (PWM). The three-phase PWM was implemented in SkyWater 130 nm CMOS technology and tapeout through a low-cost multi-project wafer (MPW). This work demonstrates the possibility of using open-AI, open-source EDA design flow to design practical circuits. Sanctions, expensive software license subscriptions, and complicated NDA do not restrict it. All academic institutions and companies should have open-AI, open-source, open-access EDA and PDKs for a sustainable IC design flow.