AI-Powered EDA & Circuit Exploration

glayout: Analog Layout Automation Framework [1]

gLayout is a Python-based, programmatic analog layout framework that enables highly parameterized physical design generation with integrated design rule validation. By abstracting process design kits (PDKs) into portable, rule-aware primitives, gLayout supports scalable and technology-agnostic layout synthesis. The framework establishes a foundation for agentic analog layout automation. In our prior work, we demonstrated reinforcement-learning–driven operational amplifier synthesis, where thousands of DRC-clean, performance-optimized layout instances were automatically generated in a single training run. These results illustrate a viable path toward closed-loop, learning-based analog physical design.

OpenFASoC: An Open-Source Framework for Fully Automated Mixed-Signal SoC Generation

• 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 Algorithms 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.

The framework below uses a genetic algorithm to search for approximate-computing circuits.

[1] A. Hammoud et al., Human Language to Analog Layout Using GLayout Layout Automation Framework. MLCAD’24.