David vs. Goliath: Can Small Models Win Big with Agentic AI in Hardware Design?

2025-12-04

AI Summary

Large language models (LLMs) incur prohibitive computational overhead and poor energy efficiency in hardware design, challenging their practical deployment. Method: This paper challenges the “bigger is better” paradigm by proposing a Small Language Model (SLM) + Agent AI collaborative framework tailored for Verilog design. The framework employs task decomposition, context-aware prompting, multi-turn self-correction, and iterative feedback to enable continual learning and collaborative reasoning with lightweight models. Contribution/Results: Evaluated on the NVIDIA CVDP benchmark, our approach achieves design quality comparable to state-of-the-art LLMs while reducing inference cost substantially. Experimental results demonstrate that integrating lightweight models with structured agent workflows balances performance, energy efficiency, and scalability in complex chip design tasks—paving a novel pathway toward green AI–driven adaptive hardware design.

Technology Category

• Machine Learning: Hardware-aware ML
• Natural Language Processing: (Large) Language Models
• Cognitive Modeling & Cognitive Systems: Agent Architectures

Application Category

• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Large Language Model(LLM) inference demands massive compute and energy, making domain-specific tasks expensive and unsustainable. As foundation models keep scaling, we ask: Is bigger always better for hardware design? Our work tests this by evaluating Small Language Models coupled with a curated agentic AI framework on NVIDIA's Comprehensive Verilog Design Problems(CVDP) benchmark. Results show that agentic workflows: through task decomposition, iterative feedback, and correction - not only unlock near-LLM performance at a fraction of the cost but also create learning opportunities for agents, paving the way for efficient, adaptive solutions in complex design tasks.

Problem

• Evaluating small models with agentic AI for hardware design efficiency
• Reducing computational cost and energy in domain-specific LLM tasks
• Testing if agentic workflows can match large model performance affordably

Innovation

• Small models with agentic AI framework
• Task decomposition and iterative feedback
• Near-LLM performance at lower cost