AI-Powered EDA & Circuit Exploration

OpenFASoC: An Open Source Framework for Fully Automated Mixed-Signal SoC Generation (IC Design Automation)

• Human Language to Analog Layout Using GLayout Layout Automation Framework. MLCAD’24.
• Disrupting Conventional Chip Design through the Open Source EDA Ecosystem. EDTM’24.
• Openfasoc: An open platform towards analog and mixed-signal automation and acceleration of chip design. ISDCS’23.
• An open source compatible framework to fully autonomous digital LDO generation. ISCAS’23.
• An open-source and autonomous temperature sensor in SkyWater 130 nm for comprehensive design space exploration. SSC-L’22.
• The missing pieces of open design enablement: A recent history of Google efforts. ICCAD’20. Bridging academic open-source EDA to real-world usability. ICCAD’20.

Genetic Alogorithm for Unconventional Circuit Synthesis

We use genetic algorithms to explore unconventional circuit designs across multiple topics, including approximate computing and transistor-level synthesis. By evolving candidate circuits with fitness functions that capture power, accuracy, and area, we can discover non-obvious topologies that traditional flows often miss. This approach lets us search large design spaces quickly while keeping targets and constraints explicit.

Here shows framework for using genetic algorithm to search for approximate compute circuits.