CRADLE: Conversational RTL Design Space Exploration with LLM-based Multi-Agent Systems

TL;DR

CRADLE is a conversational framework utilizing LLM-based multi-agent systems for flexible, user-guided RTL design space exploration, achieving significant FPGA resource optimizations through internal verification and correction.

Contribution

It introduces a novel LLM-based multi-agent system for interactive RTL design exploration with self-verification and optimization capabilities.

Findings

Achieves 48% LUT reduction on average

Achieves 40% FF reduction on average

Demonstrates effectiveness on RTLLM benchmark

Abstract

This paper presents CRADLE, a conversational framework for design space exploration of RTL designs using LLM-based multi-agent systems. Unlike existing rigid approaches, CRADLE enables user-guided flows with internal self-verification, correction, and optimization. We demonstrate the framework with a generator-critic agent system targeting FPGA resource minimization using state-of-the-art LLMs. Experimental results on the RTLLM benchmark show that CRADLE achieves significant reductions in resource usage with averages of 48% and 40% in LUTs and FFs across all benchmark designs.