Agentic data science with the Data Agent Kit
cloud.google.com·8h ago
AI Research
arXiv CS.AI·Sep 3·5 min read
Dictionary-Guided Mutation Operators for Automated HDL Repair
TL;DR
A dictionary-guided mutation system for automated Hardware Description Language (HDL) repair was developed, achieving significant improvements in repair accuracy and speed. It successfully repaired 14 bug variants, including complex multi-bug instances, outperforming existing methods like CirFix by 18 times on specific benchmarks.
✦ Why It Matters
Consider implementing dictionary-guided mutation strategies in your HDL repair processes to enhance efficiency and accuracy.
Key Takeaways
- Explore dictionary-guided mutation for HDL repairs to improve outcomes.
- Utilize regex-based modifications to streamline code repair processes.
- Implement fault localization to focus repair efforts on critical code areas.
- Evaluate your current HDL repair methods against the CirFix benchmark for performance insights.
How It Works
The proposed system employs a dictionary of mutation vocabularies tailored to specific designs, allowing for targeted modifications of HDL code. By using regex-based matching, it applies token substitutions, insertions, and deletions directly to the Verilog source code.
The simulation-divergence fault localization module identifies which output signals diverge from expected results during a single simulation run, scoring the source lines based on their structural proximity to these signals. This targeted approach ensures that the most suspicious areas of the code are prioritized for mutation, enhancing the likelihood of successful repairs.