Recent breakthroughs in large language models (LLMs) and autonomous agents open new opportunities for hardware design optimization. This paper investigates three interrelated directions toward autonomous RTL and synthesis optimization:
- Realistic benchmarking: It discusses reference‑based and open‑ended RTL optimization benchmarks that evaluate opportunities under authentic EDA workflows, balancing reproducibility and exploratory freedom.
- Tool‑grounded RTL optimization: It introduces Dr. RTL, a multi‑agent framework that automates critical‑path analysis, RTL rewriting, and implementation‑level evaluation, continuously improving via reusable optimization skills. Evaluated on 20 real‑world RTL designs, Dr. RTL achieves average WNS/TNS improvements of 21%/17% and a 6% area reduction compared with commercial synthesis tools.
- Design‑specific synthesis optimization: It explores agent‑driven synthesis script tuning and the discovery of executable synthesis algorithms across intermediate representations, technology‑independent optimization, and technology mapping.
Finally, the authors envision AI‑native design automation that jointly explores architectural, RTL, and EDA algorithm design spaces, guided by verification constraints and implementation feedback.
Review: The work convincingly demonstrates how LLMs and multi‑agent systems can be integrated into the hardware design flow, providing measurable benchmarks and substantial experimental gains, thereby laying a solid foundation for future AI‑driven end‑to‑end hardware optimization.