ChipOmni: Streamlining Electronic Design Automation with AI

Paper: ChipOmni: A General-Purpose AI Agent for Electronic Design Automation

Authors: Pengkun Wang, Rui Mao, Zhe Zhao, Dongting Xie, Haibin Wen, Song Xu, Di Wu, Xu Wang, Zhengyang Zhou, Binwu Wang, Xiaofeng Cao, Yang Wang

Executive Summary

Electronic design automation (EDA) is essential for developing complex electronic systems, yet the current workflows in this field are often disjointed. Engineers must navigate various specialized tools and representations, making the design process cumbersome. Traditional approaches using large language models have improved some aspects of EDA, primarily by assisting with task-specific code generation. However, these models typically work in isolation and do not effectively manage the entire design process, including planning, verification, and repair.

Recent research introduces ChipOmni, a new AI framework aimed at transforming EDA workflows. ChipOmni enhances the design process by offering a more integrated approach that combines several capabilities: it understands multiple types of data, plans tasks dynamically, allows collaboration among different AI agents, and interacts with verification tools. This comprehensive system adapts to feedback from various engineering tools, ensuring that the task context evolves as the design progresses.

The research highlights how ChipOmni performs across several benchmarks related to EDA tasks, including analog circuit design, register transfer level (RTL) design, testbench creation, and formal assertion generation. It demonstrates improved performance on these benchmarks compared to existing methods. Notably, ChipOmni successfully completed all 30 tasks in the AnalogCoder-Pro benchmark, a significant achievement in the context of EDA automation.

One of the standout features of ChipOmni is its capability to handle long-term design projects without human intervention after the initial setup. In a detailed case study, it took a natural-language specification for a RISC-V processor and transformed it through three levels of design, ultimately producing a micro-system on chip (micro-SoC) layout that met all necessary physical sign-off criteria. This ability to autonomously manage complex design tasks marks a substantial advancement in EDA practices.

The implications of this research are significant. ChipOmni could facilitate a shift from the current fragmented, tool-centric processes in EDA to more cohesive, agent-driven systems. By streamlining the design workflow, it may reduce the time and effort required for engineers to complete complex projects, potentially leading to faster innovation in electronic design. The successful integration of multimodal data understanding and dynamic planning highlights a promising direction for future EDA tools, aiming for a more automated and efficient design environment.

This research underscores the potential for AI to transform EDA, encouraging enterprises to consider how such advancements could enhance their design capabilities and overall productivity.

Academic Abstract

Electronic design automation (EDA) workflows remain highly fragmented, requiring engineers to coordinate heterogeneous representations, specialised software tools, and iterative verification across multiple design stages. Although large language models have brought more general reasoning and code-generation capabilities to EDA, many existing approaches remain focused on isolated tasks and struggle to plan, verify, and repair designs autonomously within executable engineering environments. Here we introduce ChipOmni, a knowledge-enhanced, context-evolving AI agent framework for EDA. ChipOmni integrates multimodal understanding, dynamic planning, multi-agent collaboration, and verification-driven tool interaction into a unified autonomous system. By continually incorporating feedback from external engineering tools, the framework maintains an evolving task context, integrates domain knowledge, and iteratively refines intermediate designs. Across benchmarks for analog circuit, RTL, testbench, and formal-assertion generation, ChipOmni improves multiple task-specific measures over the evaluated baselines and solves all 30 AnalogCoder-Pro tasks. In a long-horizon case study, ChipOmni advances a natural-language RISC-V specification through three design levels to a micro-SoC GDSII layout that passes physical sign-off, without human intervention after initialization. Together, these results show that ChipOmni can coordinate verification-driven design across heterogeneous tasks and sustain long-horizon engineering workflows, advancing a shift from fragmented, tool-centred processes towards agent-centred, closed-loop systems.