Reproduce “TicTacBench: Benchmarking Timing Closure Capabilities of Coding Agents” 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

Bowei Wang, Zhigang Fang, Zhijie Yang, Renzhi Chen, Shanshan Li, Lei Wang

TicTacBench evaluates whether agents can repair functionally correct RTL for setup-timing closure after placement and routing, rather than relying on pre-PnR estimates. It supplies constraints and timing reports, then validates submissions with a full post-PnR flow across 30 tasks. GPT-5.4 closes 53.3% of tasks, but all eight agents worsen ADP.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.