الملخص
Reinforcement learning (RL) has emerged as a promising approach for placement optimization, particularly when combined with graph neural networks (GNNs) that capture circuit connectivity. However, most learning-based placement approaches focus on floorplanning, macro placement, or global placement, while detailed placement refinement remains relatively unexplored. In this paper, we present GPlaceRL, an open-source graph reinforcement learning framework for detailed placement refinement. GPlaceRL represents legalized placements as graphs and provides a modular environment for studying graph encoders, policy architectures, reward formulations, and local placement actions. To demonstrate the capabilities of GPlaceRL, we conduct a systematic evaluation of proximal policy optimization (PPO) policies with graph attention network (GAT) encoders in a per-design optimization setting. Across five placement benchmarks, the best greedy evaluation results achieve HPWL improvements ranging from $3.27\%$ to $32.87\%$. The results highlight the importance of compact GAT architectures and flexible local action spaces for placement optimization. Overall, GPlaceRL provides a reproducible and extensible framework for systematic research on RL-based detailed placement refinement.
الكلمات المفتاحية
الموضوع
بيانات النشر
- المجلة
- غير متاح
- وصول مفتوح
- وصول مفتوح أخضر
اقتبس هذه المقالة
APA 7
Stoikos, P., Oikonomou, F., Poulos, C., Pantazi-Kypriou, M., Tziouvaras, A., Anagnostopoulos, C., Karakonstantis, G., & Floros, G. (2026). GPlaceRL: An Open-Source Graph Reinforcement Learning Framework for Detailed Placement. https://omanscience.com/ar/articles/gplacerl-an-open-source-graph-reinforcement-learning-framework-for-detailed-placement
MLA 9
Stoikos, Pavlos, et al. "GPlaceRL: An Open-Source Graph Reinforcement Learning Framework for Detailed Placement." https://omanscience.com/ar/articles/gplacerl-an-open-source-graph-reinforcement-learning-framework-for-detailed-placement.
شيكاغو (المؤلف–التاريخ)
Stoikos, Pavlos, Foteini Oikonomou, Christos Poulos, Maria Pantazi-Kypriou, Athanasios Tziouvaras, Christos Anagnostopoulos, Georgios Karakonstantis, and George Floros. 2026. "GPlaceRL: An Open-Source Graph Reinforcement Learning Framework for Detailed Placement." https://omanscience.com/ar/articles/gplacerl-an-open-source-graph-reinforcement-learning-framework-for-detailed-placement.
هارفارد
Stoikos, P., Oikonomou, F., Poulos, C., Pantazi-Kypriou, M., Tziouvaras, A., Anagnostopoulos, C., Karakonstantis, G. and Floros, G. (2026) 'GPlaceRL: An Open-Source Graph Reinforcement Learning Framework for Detailed Placement', Available at: https://omanscience.com/ar/articles/gplacerl-an-open-source-graph-reinforcement-learning-framework-for-detailed-placement.
فانكوفر
Stoikos P, Oikonomou F, Poulos C, Pantazi-Kypriou M, Tziouvaras A, Anagnostopoulos C, et al. GPlaceRL: An Open-Source Graph Reinforcement Learning Framework for Detailed Placement. https://omanscience.com/ar/articles/gplacerl-an-open-source-graph-reinforcement-learning-framework-for-detailed-placement
IEEE
P. Stoikos, F. Oikonomou, C. Poulos, M. Pantazi-Kypriou, A. Tziouvaras, C. Anagnostopoulos, G. Karakonstantis, and G. Floros, "GPlaceRL: An Open-Source Graph Reinforcement Learning Framework for Detailed Placement," https://omanscience.com/ar/articles/gplacerl-an-open-source-graph-reinforcement-learning-framework-for-detailed-placement.