Accelerating Detailed Routing Convergence through Offline Reinforcement Learning

2025-12-03

AI Summary

Modern detailed routing suffers from slow convergence due to shrinking feature sizes and increasingly stringent design rules. Conventional iterative pathfinding relies on static edge-cost weights, making it difficult to simultaneously eliminate design rule violations (DRVs) and maintain runtime efficiency. This paper proposes a dynamic cost-weight scheduling method based on offline reinforcement learning, employing Conservative Q-Learning (CQL) to learn weight-adjustment policies from historical routing data—enabling cross-process-node knowledge transfer and generalization. The approach requires no online training and integrates seamlessly into existing iterative pathfinding frameworks. Evaluated on the ISPD’19 benchmark suite, it achieves an average 1.56× speedup (up to 3.01×), while maintaining or reducing DRV counts. This significantly improves both routing efficiency and robustness.

Technology Category

• Planning, Routing, and Scheduling: Planning/Scheduling and Learning
• Search and Optimization: Learning to Search
• Intelligent Robots: Learning & Optimization for ROB

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

Detailed routing remains one of the most complex and time-consuming steps in modern physical design due to the challenges posed by shrinking feature sizes and stricter design rules. Prior detailed routers achieve state-of-the-art results by leveraging iterative pathfinding algorithms to route each net. However, runtimes are a major issue in detailed routers, as converging to a solution with zero design rule violations (DRVs) can be prohibitively expensive. In this paper, we propose leveraging reinforcement learning (RL) to enable rapid convergence in detailed routing by learning from previous designs. We make the key observation that prior detailed routers statically schedule the cost weights used in their routing algorithms, meaning they do not change in response to the design or technology. By training a conservative Q-learning (CQL) model to dynamically select the routing cost weights which minimize the number of algorithm iterations, we find that our work completes the ISPD19 benchmarks with 1.56x average and up to 3.01x faster runtime than the baseline router while maintaining or improving the DRV count in all cases. We also find that this learning shows signs of generalization across technologies, meaning that learning designs in one technology can translate to improved outcomes in other technologies.

Problem

• Accelerating detailed routing convergence using offline reinforcement learning
• Reducing runtime by dynamically selecting routing cost weights
• Improving generalization across technologies with learned designs

Innovation

• Offline reinforcement learning for dynamic cost weight selection
• Conservative Q-learning model reduces algorithm iterations
• Generalizable across technologies for faster routing convergence