Open source EDA for Machine Learning
summary
Open Road is an open source RTL‑to‑GDS physical design and synthesis flow used for tape‑outs down to 12 nm and being pushed for estimation at ~2 nm, with commercial support via Precision Innovations in a Red Hat‑style model. The flow’s fast estimation mode reports 4–20× faster runs than proprietary tools, and an LLM‑based OFS agent cuts tuning experiments by about 40% fewer runs. A nonprofit initiative and NSF funding back governance and ecosystem work.
AI-written summary · about
chapters (11)
- 0:00Open Road and Precision Innovations
- 1:45ML barriers with proprietary EDA
- 3:31Why open source for ML in EDA
- 4:12Open Road for high-level synthesis
- 5:39Open source EDA’s innovation role
- 6:02OFS agent for LLM hyperparameter tuning
- 7:09Open Road Initiative governance
- 8:58Future of Open Road and support
- 9:18Q&A: AI in routing and CI
- 10:44Q&A: Red Hat vs 2nm efforts
- 11:26Q&A: Analog expansion and closing
related talks
22mArchitecture 2.0: Toward Open Source Foundation Models and Datasets for Hardware Design
29mAn introduction to Open Accelerated Discovery
14mAutomating Design Space Exploration Using Advanced Simulation Technologies
15mComputational Mobility: Learning, Prediction, and Algorithms
21mOpen Sourcing Project Sparrow
20mRoute Optimization Engine for the Real World
04mAI-Powered Observability in Semiconductor Design
01h 06mAutomated design space exploration and generation of AI accelerators