Reinforcement Learning-Based Optimization of Workload-Aware Power Delivery Networks

2026-10-05

AI Summary

This work addresses the inefficiency and resource over-provisioning inherent in traditional power delivery network (PDN) design, which relies on worst-case assumptions. We propose a workload-aware adaptive PDN optimization framework that generates architecture-level power traces via system-level simulation and maps them into spatial density distributions. By integrating SPICE analysis with electromigration (EM) lifetime assessment, a Deep Q-Network (DQN) dynamically optimizes wire widths to minimize routing area. Evaluated on PARSEC and SPLASH-2 benchmarks, the proposed method reduces PDN area by 47% on average while satisfying EM and IR-drop constraints. Furthermore, it achieves an approximately 26× speedup in optimization time compared to simulated annealing. This approach enables efficient, workload-driven PDN resource allocation, significantly improving upon conventional conservative design methodologies.

Abstract

Power Delivery Networks (PDNs) are critical components of modern VLSI chips, providing stable voltage levels while satisfying electromigration (EM) and IR-drop constraints. Conventional PDN design methodologies typically rely on worst-case assumptions, often resulting in over-provisioned networks and inefficient use of resources. This paper presents a reinforcement learning-based framework for the optimization of workload-aware PDNs. The proposed methodology first generates workload-aware PDNs using architectural power traces obtained from system-level simulations. These power traces are mapped to spatial power density distributions, enabling adaptive allocation of PDN resources according to local current demand. A reinforcement learning agent then performs wire-width optimization to minimize PDN area while maintaining EM and voltage integrity constraints. Electrical and reliability metrics are obtained using SPICE-based circuit analysis and EM lifetime estimation. Experimental evaluation is performed on a dataset of workload-aware PDNs generated from 4-, 8-, and 16-core multiprocessor floorplans using PARSEC and SPLASH-2 benchmark workloads. Furthermore, the proposed Deep Q-Network (DQN)-based optimizer reduces the average normalized PDN area by 47\% while satisfying all EM and IR-drop constraints. Compared to simulated annealing, the proposed approach achieves comparable optimization quality while providing approximately 26$\times$ faster optimization.

Problem

• Power Delivery Networks
• Workload-Aware Optimization
• Electromigration
• IR-drop
• VLSI

Innovation

• Reinforcement Learning
• Power Delivery Networks
• Workload-Aware Optimization
• Deep Q-Network
• VLSI