Reproduce “Agentic AI for Heterogeneous System Design Across Chip, Scale-Up, and Scale-Out Domains” 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

Murali Emani, Kashyap R Manjusha

ArchAgent searches facility designs for AI and HPC workloads. An LLM planner proposes hardware portfolios; specialists advise, and a deterministic analytical evaluator scores proposals. Its best modeled score across six workloads is 1.93, versus hill climbing’s 1.65 under 100 evaluator calls. The comparison uses a B200-based design against an H100 reference, so it does not isolate heterogeneity’s effect.

Summary by alphaXiv.

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