Reproduce “AssertMiner: Module-Level Spec Generation and Assertion Mining using Static Analysis Guided LLMs” 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
Hongqin Lyu, Yonghao Wang, Jiaxin Zhou, Zhiteng Chao, Tiancheng Wang, Huawei Li
AssertMiner combines static analysis of Register-Transfer Level (RTL) code with Large Language Models (LLMs) to automatically generate reliable module-level specifications and deep assertions for hierarchical hardware designs. The framework significantly boosts fault coverage by detecting errors often missed by top-level assertions and enhances structural coverage compared to existing LLM-based methods.
Key insights
- Reliable module-level specifications can be accurately inferred by LLMs when guided by static structural analysis, without directly interpreting potentially buggy RTL implementation details.
- Deep assertions, targeting specific modules, significantly enhance error detection and localization capabilities compared to only using top-level assertions.
- A hybrid approach combining static code analysis for structural context with LLMs for semantic reasoning enables comprehensive and robust assertion generation for complex hierarchical designs.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.