ORFS-agent: LLM-Based Tool for Chip Design Optimization
A new AI agent called ORFS-agent uses large language models to automate parameter tuning in open-source chip design. It iteratively explores configurations, outperforming standard Bayesian optimization in resource efficiency and final design metrics across six benchmarks on ASAP7 and SKY130HD nodes. The system leverages thinking-model backends such as Sonnet 4.6 and Kimi K2.5.
Key facts
- ORFS-agent is an LLM-based iterative optimization agent for chip design.
- It automates parameter tuning in an open-source hardware design flow.
- It outperforms standard Bayesian optimization in resource efficiency and design metrics.
- Tested on six benchmarks using ASAP7 and SKY130HD nodes.
- Uses thinking-model backends: Sonnet 4.6 and Kimi K2.5.
- Published on arXiv with ID 2506.08332.
- Targets register-transfer level to physical layout flows.
- Addresses high-dimensional optimization with thousands of parameters.
Entities
Institutions
- arXiv