Reproduce “CHIA: An open-source framework for principled, agentic AI-driven hardware/software co-design research” 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

Angela Cui, Ferran Hermida-Rivera, Jack Toubes, Raghav Gupta, Jim Fang, Chengyi Lux Zhang, Ella Schwarz, Junha Kim and 4 more

CHIA provides an open-source framework for composing and running AI-assisted hardware and software design workflows as reusable loops and clusters. Python-orchestrated tasks connect agents to builds, simulators, tests, and physical-design tools across resources. In one demonstration, RTL optimization of a 4-wide MegaBOOM core raised Sky130 frequency from 46.8 to 95.1 MHz, but the result did not generalize wholesale to commercial 16nm.

Key insights

  • Treating the productive construction and scalable deployment of AI-driven co-design flows as a primary objective is essential for advancing principled and reproducible scientific inquiry into agentic AI for hardware design.
  • A modular, containerized, and fault-tolerant framework effectively addresses the engineering and scientific challenges of integrating AI into complex HW/SW co-design workflows, moving beyond bespoke implementations.
  • Balancing agentic autonomy with programmatic control and robust, isolated verification mechanisms ensures that AI-generated designs and contributions meet high standards of quality and verifiability.

Summary by alphaXiv.

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