Reproduce “Agentic AI-based Coverage Closure for Formal Verification” on OpenResearch
Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.
Universal app for Apple silicon and Intel. Requires macOS 11 or later.
About this paper
Sivaram Pothireddypalli, Ashish Raman, Deepak Narayan Gadde, Aman Kumar
An agentic AI framework automates coverage closure in formal hardware verification by iteratively analyzing uncovered RTL regions and generating targeted SystemVerilog properties. This method consistently achieves a 10% to 20% increase in formal coverage across diverse designs, accelerating the verification sign-off process.
Key insights
- Large Language Models (LLMs) can effectively automate complex formal verification tasks, such as identifying coverage gaps and generating targeted properties, by leveraging their contextual reasoning abilities.
- An iterative feedback loop, where agents analyze coverage reports, generate properties, and re-run verification, systematically drives towards comprehensive coverage closure.
- The performance of different LLMs varies in this application, with advanced models like OpenAI's GPT-5 demonstrating superior reasoning and higher coverage metrics compared to others like Meta's Llama3.3.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.