HighTide: An Agent-Curated Open-Source VLSI Benchmark Suite

2026-06-02

AI Summary

Existing open-source VLSI benchmark suites lack diversity, scalability, and a complete compilation-to-verification pipeline, hindering their utility in AI-driven EDA research. This work proposes and constructs the first publicly available, reproducible, and continuously evolving open-source VLSI benchmark suite, encompassing multiple hardware description languages and process nodes, and integrating Bazel-based incremental builds, remote caching, and a full RTL-to-GDS flow. Innovatively, it introduces an AI agent endowed with twelve distinct capabilities, augmented with decision logs serving as long-term memory, to autonomously curate designs, manage their lifecycle, and perform meta-maintenance of the benchmark itself. This effort establishes a standardized evaluation platform for AI-for-EDA research and catalyzes the advancement of the open-source hardware ecosystem.

Abstract

We introduce HighTide, an evolving AI-assisted benchmark suite. Specifically, the contributions are: (i) a diverse open-source suite spanning multiple design languages and technology nodes, (ii) Bazel-based incremental RTL-to-GDS compilation with remote caching, (iii) AI-assisted design curation through twelve agent skills covering the design lifecycle, flow optimization, tool reference, and meta-maintenance, backed by per-design decision logs that serve as long-term memory of tuning rationale across the suite, and (iv) an infrastructure with RTL compilation verification for stable releases. The suite is publicly available and designed to grow with the open-source hardware ecosystem.

Problem

• VLSI benchmark
• open-source hardware
• AI-assisted design
• RTL-to-GDS flow
• design curation

Innovation

• AI-assisted design
• VLSI benchmark suite
• incremental RTL-to-GDS compilation
• agent-based curation
• open-source hardware