RAG-Enhanced Kernel-Based Heuristic Synthesis (RKHS): A Structured Methodology Using Large Language Models for Hardware Design

TL;DR

This paper introduces RKHS, a novel methodology combining retrieval-augmented generation and kernel heuristics with LLMs to improve hardware design automation tasks like scheduling.

Contribution

It presents a structured LLM-based approach that synthesizes reusable heuristics, demonstrating improved scheduling performance with manageable overhead.

Findings

Prototype reduces schedule length by up to 11%

Achieves only 1.3x runtime overhead

Generalizes to other EDA optimization problems

Abstract

Heuristic design upholds modern electronic design automation (EDA) tools, yet crafting effective placement, routing, and scheduling strategies entails substantial expertise. We study how large language models (LLMs) can systematically synthesize reusable optimization heuristics beyond one-shot code generation. We propose RAG-Enhanced Kernel-Based Heuristic Synthesis (RKHS), which integrates retrieval-augmented generation (RAG), compact kernel heuristic templates, and an LLM-driven refinement loop inspired by iterative self-feedback. Applied to latency-minimizing list scheduling in high-level synthesis (HLS), a prototype reduces average schedule length by up to 11 percent over a baseline scheduler with only 1.3x runtime overhead, and the structured retrieval-synthesis loop generalizes to other EDA optimization problems.