Deep Reinforcement Learning-Based Power Management for Chiplet-Based Multicore Systems
Xiao Li,Lin Chen,Shixi Chen,Fan Jiang,Chengeng Li,Wei Zhang,Jiang Xu
DOI: https://doi.org/10.1109/tvlsi.2024.3415487
2024-08-28
IEEE Transactions on Very Large Scale Integration (VLSI) Systems
Abstract:Chiplet technology has emerged as a promising solution to address the increasing demand for high-performance computing in light of the slowdown of Moore's law. While chiplet-based multicore systems offer higher performance through heterogeneous integration, they also pose challenges for power delivery system (PDS) design. The integration of additional vertical and inter-chiplet connections, along with higher power density, impose stringent requirements on power delivery. Moreover, PDS efficiency is affected by workload variations at runtime, necessitating the need to design and manage PDSs and processors as a whole to improve system energy efficiency while balancing performance. In this article, we propose an offline-online co-design optimization methodology that combines offline PDS design optimization with online power management. To address the power consumption and delivery mismatch, we introduce a centralized deep Q-network (DQN)-based online control scheme for power co-management in chiplet-based multicore systems. By carefully designing the state space and reward functions, our approach achieves workload-aware adaptive control to reduce the energy-delay-product (EDP) while maintaining PDS efficiency under a given performance target (PT). We conduct evaluations on realistic applications to validate the effectiveness of our approach. For 64-core systems, our method achieves an average EDP reduction of 67% while meeting a 90% PT, surpassing state-of-the-art modular Q-learning (MQL)-based and heuristic-based approaches by up to 4% and 16%, respectively. Additionally, our approach demonstrates wiser action selection policies, higher control stability, and lower implementation overhead compared to the MQL-based approach.
engineering, electrical & electronic,computer science, hardware & architecture