AgenticPD: A Stage-Aware Agentic Framework for Physical Design QoR Optimization

2026-07-06

AI Summary

This work addresses the high cost of quality-of-results (QoR) optimization in physical design, where evaluation typically requires a full EDA flow and decisions across stages are tightly coupled. To overcome these challenges, the paper proposes a stage-aware multi-agent framework that deploys specialized agents at each physical design stage for localized optimization, coordinated by a central referee agent to guide global search. The framework incorporates structured observation encoding, context management, and checkpoint reuse mechanisms, enabling efficient branching from intermediate states and signoff-quality evaluation without repeatedly executing the entire design flow. Experimental results demonstrate that the proposed approach significantly improves post-routing timing performance while maintaining competitive power and area metrics.

Abstract

Physical design quality-of-results~(QoR) optimization is hard and expensive. Choices made at one stage can help or hurt later stages. Each evaluation requires a costly EDA run through the full flow. While existing methods still treat optimization as flat parameter tuning or a LLM-based script generation task, we present AgenticPD, a stage-aware agentic framework for physical design QoR optimization. Instead of re-running the full flow after every trial, AgenticPD is organized around the stage boundaries of the physical design flow, where a Judge Agent navigates the search and stage-specialized agents make local decisions within their own stage using stage-local tools. Additionally, the agent harness in AgenticPD provides structured observations, execution history, and agent context management. As a result, the system can branch from prior intermediate states and reuse checkpoints to continue the optimization procedure, and every candidate is evaluated at the post-route signoff. Across these baselines, AgenticPD achieves strong post-route timing while remaining competitive in power and area.

Problem

• Physical Design
• Quality-of-Results
• QoR Optimization
• EDA Flow
• Stage-aware Optimization

Innovation

• Agentic Framework
• Stage-Aware Optimization
• Physical Design
• QoR
• Checkpoint Reuse