📄 中文摘要
随着电子产品的快速发展，2.5D集成电路的应用日益普及，自动化芯片放置在更大且更异构的芯片组件中变得至关重要。现有的放置方法通常侧重于最小化布线长度，或通过加权和将多目标优化转化为单一目标，这限制了其处理竞争设计需求的能力。布线长度的减少与热管理是固有的冲突目标，使得先前的方法在实际应用中显得不足。为了解决这一挑战，提出了TDPNavigator-Placer，这是一种新颖的多智能体强化学习框架，能够动态优化芯片放置，兼顾热管理和布线长度的需求。
TDPNavigator-Placer：基于多智能体强化学习的热管理与布线长度优化的2.5D芯片放置
📄 English Summary
TDPNavigator-Placer: Thermal- and Wirelength-Aware Chiplet Placement in 2.5D Systems Through Multi-Agent Reinforcement Learning
The rapid advancement of electronics has led to the widespread adoption of 2.5D integrated circuits, where effective automated chiplet placement is crucial as systems scale to larger and more heterogeneous chiplet assemblies. Existing placement methods typically focus on minimizing wirelength or transforming multi-objective optimization into a single objective through weighted sums, which limits their ability to address competing design requirements. Wirelength reduction and thermal management are inherently conflicting objectives, rendering prior approaches inadequate for practical deployment. To tackle this challenge, TDPNavigator-Placer is proposed, a novel multi-agent reinforcement learning framework that dynamically optimizes chiplet placement while considering both thermal management and wirelength reduction.