IR Drop

IR drop, the voltage drop across a power delivery network (PDN) in integrated circuits, is a critical factor impacting circuit performance and reliability. Research focuses on developing faster and more accurate prediction methods, primarily using machine learning approaches such as convolutional neural networks (CNNs) and, increasingly, graph neural networks (GNNs) integrated with CNNs to better capture the complex relationships between PDN structure and current consumption. These advancements aim to replace computationally expensive simulations, enabling faster design cycles and improved optimization of PDNs. The improved accuracy and speed of these predictive models are crucial for designing increasingly complex and power-hungry integrated circuits.

Papers

August 6, 2024

Static IR Drop Prediction with Attention U-Net and Saliency-Based Explainability
Lizi Zhang, Azadeh Davoodi
Tetromino Pixel Saliency Explanation IR Drop

March 27, 2024

PDNNet: PDN-Aware GNN-CNN Heterogeneous Network for Dynamic IR Drop Prediction
Yuxiang Zhao, Zhuomin Chai, Xun Jiang, Yibo Lin, Runsheng Wang, Ru Huang
Heterogeneous GNN IR Drop

IR Drop

Papers

Static IR Drop Prediction with Attention U-Net and Saliency-Based Explainability

PDNNet: PDN-Aware GNN-CNN Heterogeneous Network for Dynamic IR Drop Prediction