Paper Collection - IR Drop & EM Research

Non-Prediction

這類論文不使用 AI 預測數值,而是透過「新硬體架構 (BSPDN)」或「演算法優化 (Optimization)」直接解決 IR Drop 問題。

Backside Power Delivery Network (BSPDN)

背面供電技術,2nm 製程後的熱門解法

  1. Reinforcement Learning-Based Optimization of Back-side Power Delivery Networks in VLSI Design for IR-drop Reduction
    • Tag: RL, Architecture Optimization
  2. A Design Methodology for Back-side Power and Clock Routing Co-Optimization
    • Tag: Co-Design, Routing

Vector Generation

主動生成測試向量以進行 Sign-off,而非被動預測

  1. Peak Power and Dynamic IR-drop Assessment via Waveform Augmenting
    • Tag: Validation, Waveform Generation, ICCAD

Differentiable Optimization

將佈局與尺寸調整數學化,進行梯度下降求解

  1. Fusion of Global Placement and Gate Sizing with Differentiable Optimization
    • Tag: Placement, Sizing, Math-heavy

Electromigration (EM) & Reliability

EM 是 IR Drop 的孿生問題。這類論文專注於金屬線壽命與可靠度,通常與電流密度優化有關。

EM-Aware Routing & Optimization

  1. Physics-Informed Neural Networks (PINNs) for Electromigration Analysis
    • Why: 這是不做純 Data-driven 預測的最佳替代方案。PINNs 結合了 PDEs 與 AI,能更準確地模擬 EM 應力演變,而不只是猜壽命。
    • Tag: Physics-ML, Reliability
  2. EM-Aware Analog/Mixed-Signal Routing with Reinforcement Learning
    • Why: 類比電路或電源線的 EM 很難解,這類論文通常使用 RL agent 來畫出不會違反 EM 規則的線。
    • Tag: Routing, RL

Multi-Physics Co-Optimization (Thermal + EM + IR)

  1. Machine Learning-Based Thermal and EM Joint Optimization for 3D-IC
    • Why: 3D-IC 的散熱會惡化 EM。這類論文探討如何同時調整電壓 (解決 IR) 和 溫度 (解決 EM)。
    • Tag: 3D-IC, Multi-physics

Generative AI for Power Grid

利用 Diffusion Model 或 GAN 直接生成完美的電源網格。

  1. Power Grid Synthesis using Diffusion Models
    • Context: (搜尋關鍵字:Diffusion models for VLSI layout, Generative PDN design)
    • Why: 這是 2024-2025 另一個大熱點。以前是人畫 Power Grid 給 AI 檢查 (Prediction);現在是讓 AI 直接畫出 (Generation) 符合 IR Drop 規範的 Grid。
    • Tag: Generative AI, Synthesis

Prediction

使用 ML (CNN/GNN/U-Net) 預測 IR Drop 熱點與數值。

  1. Real-Time Dynamic IR-drop Prediction for IR ECO
  2. Dynamic IR-Drop Prediction through a Multi-Task U-Net with Package Effect Consideration
  3. Fast IR Drop Estimation with Machine Learning

Review & Foundation

  1. Analysis of IR Drop for Robust Power Grid of Semiconductor Chip Design: A Review
  2. DynamicVoltage(IR) Drop Analysis and Design Closure: Issues and Challenges

Paper Collection - IR Drop & EM Research
https://933yee.github.io/notes/2025/12/23/paper-collection/
Author
Kevin Lee
Posted on
December 23, 2025
Licensed under