AutoPPA: Automated Circuit PPA Optimization via Contrastive Code-based Rule Library Learning

2026-04-20

AI Summary

PPA (performance, power, area) optimization in RTL design has long relied on manual expertise, while existing automated approaches suffer from limited efficiency and generalizability. This work proposes AutoPPA, a novel framework that introduces the first automatic rule induction mechanism based on contrastive code pairs. AutoPPA learns generalizable optimization rules from diverse designs through an Explore-Evaluate-Induce pipeline, eliminating the need for human prior knowledge. By integrating contrastive learning, adaptive multi-step search, and large language model–driven circuit code generation, AutoPPA consistently achieves superior PPA results across multiple benchmark circuits, significantly outperforming both hand-optimized designs and state-of-the-art methods such as SymRTLO and RTLRewriter.

Technology Category

• Search and Optimization: Learning to Search
• Machine Learning: Optimization
• Natural Language Processing: Learning & Optimization for NLP

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Cost models of using LLMs in production systems
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

Performance, power, and area (PPA) optimization is a fundamental task in RTL design, requiring a precise understanding of circuit functionality and the relationship between circuit structures and PPA metrics. Recent studies attempt to automate this process using LLMs, but neither feedback-based nor knowledge-based methods are efficient enough, as they either design without any prior knowledge or rely heavily on human-summarized optimization rules. In this paper, we propose AutoPPA, a fully automated PPA optimization framework. The key idea is to automatically generate optimization rules that enhance the search for optimal solutions. To do this, AutoPPA employs an Explore-Evaluate-Induce ($E^2I$) workflow that contrasts and abstracts rules from diverse generated code pairs rather than manually defined prior knowledge, yielding better optimization patterns. To make the abstracted rules more generalizable, AutoPPA employs an adaptive multi-step search framework that adopts the most effective rules for a given circuit. Experiments show that AutoPPA outperforms both the manual optimization and the state-of-the-art methods SymRTLO and RTLRewriter.

Problem

• PPA optimization
• RTL design
• automated optimization
• circuit optimization
• rule learning

Innovation

• AutoPPA
• PPA optimization
• contrastive rule learning
• automated RTL design
• adaptive multi-step search