Reproduce “DeepAssert: An LLM-Aided Verification Framework with Fine-Grained Assertion Generation for Modules with Extracted Module Specifications” 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
Yonghao Wang, Jiaxin Zhou, Hongqin Lyu, Zhiteng Chao, Tiancheng Wang, Huawei Li
DeepAssert introduces an LLM-aided framework that generates fine-grained deep assertions for internal hardware modules by extracting module-level specifications from high-level design specifications and inter-module relationships, significantly improving bug detection and localization efficiency in IC verification.
Key insights
- Large Language Models (LLMs) can effectively infer implicit module functionalities and signal behaviors by contextualizing a module's role within the overall design hierarchy and its interactions, even without explicit module-level documentation.
- Generating module-level specifications independently of the module's RTL implementation is crucial to prevent assertions from inadvertently replicating existing design errors.
- Fine-grained deep assertions, when combined with top-level assertions, provide superior fault coverage and significantly faster, more precise bug localization compared to top-level assertions alone.
Summary by alphaXiv.
Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.