Salvato in:
Dettagli Bibliografici
Autori principali: Wei, Yangbo, Huang, Zhen, Li, Huang, Xing, Wei W., Lin, Ting-Jung, He, Lei
Natura: Preprint
Pubblicazione: 2025
Soggetti:
Accesso online:https://arxiv.org/abs/2504.03723
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866909685648457728
author Wei, Yangbo
Huang, Zhen
Li, Huang
Xing, Wei W.
Lin, Ting-Jung
He, Lei
author_facet Wei, Yangbo
Huang, Zhen
Li, Huang
Xing, Wei W.
Lin, Ting-Jung
He, Lei
contents Hardware design automation faces challenges in generating high-quality Verilog code efficiently. This paper introduces VFlow, an automated framework that optimizes agentic workflows for Verilog code generation. Unlike traditional approaches relying on fixed prompts or manually designed flows, VFlow treats workflow discovery as a search over graph-structured LLM invocation sequences. It introduces a multi-population cooperative evolution (CEPE-MCTS) algorithm that balances multiple hardware objectives -- functional correctness, area, power, timing and token cost -- while sharing successful patterns and avoiding repeated failures. Integrated multi-level verification ensures syntactic correctness, functional behavior, and synthesizability. Experiments on VerilogEval and RTLLM2.0 show VFlow improves pass@1 by 20--30\% over prompting baselines and closely matches designer-level area/power. Remarkably, VFlow enables small LLMs to outperform larger models with up to 10.9$\times$ ROI, offering a cost-effective solution for RTL design. This work paves the way for intelligent, automated hardware development, advancing LLM applications in EDA.
format Preprint
id arxiv_https___arxiv_org_abs_2504_03723
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle VFlow: Discovering Optimal Agentic Workflows for Verilog Generation
Wei, Yangbo
Huang, Zhen
Li, Huang
Xing, Wei W.
Lin, Ting-Jung
He, Lei
Hardware Architecture
Multiagent Systems
Hardware design automation faces challenges in generating high-quality Verilog code efficiently. This paper introduces VFlow, an automated framework that optimizes agentic workflows for Verilog code generation. Unlike traditional approaches relying on fixed prompts or manually designed flows, VFlow treats workflow discovery as a search over graph-structured LLM invocation sequences. It introduces a multi-population cooperative evolution (CEPE-MCTS) algorithm that balances multiple hardware objectives -- functional correctness, area, power, timing and token cost -- while sharing successful patterns and avoiding repeated failures. Integrated multi-level verification ensures syntactic correctness, functional behavior, and synthesizability. Experiments on VerilogEval and RTLLM2.0 show VFlow improves pass@1 by 20--30\% over prompting baselines and closely matches designer-level area/power. Remarkably, VFlow enables small LLMs to outperform larger models with up to 10.9$\times$ ROI, offering a cost-effective solution for RTL design. This work paves the way for intelligent, automated hardware development, advancing LLM applications in EDA.
title VFlow: Discovering Optimal Agentic Workflows for Verilog Generation
topic Hardware Architecture
Multiagent Systems
url https://arxiv.org/abs/2504.03723