A Lightweight Inception Boosted U-Net Neural Network for Routability Prediction

Hailiang Li,Yan Huo,Yan Wang,Xu Yang,Miaohui Hao,Xiao Wang

2024-02-07

Abstract:As the modern CPU, GPU, and NPU chip design complexity and transistor counts keep increasing, and with the relentless shrinking of semiconductor technology nodes to nearly 1 nanometer, the placement and routing have gradually become the two most pivotal processes in modern very-large-scale-integrated (VLSI) circuit back-end design. How to evaluate routability efficiently and accurately in advance (at the placement and global routing stages) has grown into a crucial research area in the field of artificial intelligence (AI) assisted electronic design automation (EDA). In this paper, we propose a novel U-Net variant model boosted by an Inception embedded module to predict Routing Congestion (RC) and Design Rule Checking (DRC) hotspots. Experimental results on the recently published CircuitNet dataset benchmark show that our proposed method achieves up to 5% (RC) and 20% (DRC) rate reduction in terms of Avg-NRMSE (Average Normalized Root Mean Square Error) compared to the classic architecture. Furthermore, our approach consistently outperforms the prior model on the SSIM (Structural Similarity Index Measure) metric.

Hardware Architecture,Artificial Intelligence,Computational Engineering, Finance, and Science,Computer Science and Game Theory,Machine Learning

What problem does this paper attempt to address?

The paper mainly addresses the issues encountered in the design of modern CPU, GPU, and NPU chips, particularly in the placement and routing stages of the Very Large Scale Integration (VLSI) backend design. As semiconductor technology nodes continue to shrink to nearly 1 nanometer, placement and routing have become the two most critical processes in the VLSI circuit backend design. To efficiently and accurately evaluate routability during the placement and global routing stages, i.e., to predict Routing Congestion (RC) and Design Rule Checking (DRC) hotspots, researchers have proposed a new deep learning model. Specifically, the main objectives of the paper are to solve the following problems: 1. **Improve prediction accuracy**: Existing prediction models still need improvement in terms of accuracy. 2. **Optimize computational efficiency**: Some recent improved models have increased accuracy but run slowly in computationally intensive scenarios. 3. **Develop new deep learning architectures**: To this end, the paper proposes a lightweight Inception-enhanced U-Net neural network model (ibUNet) to address the tasks of routing congestion prediction and DRC hotspot prediction. Through experiments on the CircuitNet dataset, the paper demonstrates that the proposed ibUNet model reduces the Average Normalized Root Mean Square Error (Avg-NRMSE) metric by 5% (RC prediction) and 20% (DRC prediction) compared to the classic model RouteNet, and also shows better performance in the Structural Similarity Index Measure (SSIM). These results indicate that ibUNet can more accurately and efficiently predict routing congestion and DRC hotspots, thereby helping to avoid routing congestion and DRC violation issues at an early stage.

A Lightweight Inception Boosted U-Net Neural Network for Routability Prediction

Inverse Design of On-Chip Interconnect via Transfer Learning-Based Deep Neural Networks

High-Correlation 3D Routability Estimation for Congestion-guided Global Routing

Predicting routability of FPGA design by learning complex network images

Global Placement with Deep Learning-Enabled Explicit Routability Optimization

Instance-level Timing Learning and Prediction at Placement Using Res-UNet Network

Automatic Routability Predictor Development Using Neural Architecture Search

Effective Heterogeneous Graph Neural Network for Routing Congestion Prediction

Optimizing Routerless Network-on-Chip Designs: An Innovative Learning-Based Framework

Techniques for Scalable and Effective Routability Evaluation

An Efficient Approach for DRC Hotspot Prediction with Convolutional Neural Network.

A deep learning based latency aware predictive routing model for network‐on‐chip architectures

HybridNet: Dual-Branch Fusion of Geometrical and Topological Views for VLSI Congestion Prediction.

Hardware architecture and routing-aware training for optimal memory usage: a case study

DRC Hotspot Prediction at Sub-10nm Process Nodes Using Customized Convolutional Network

CongestNN: an Bi-Directional Congestion Prediction Framework for Large-Scale Heterogeneous FPGAs

Layout Congestion Prediction Based on Regression-ViT

RoutePlacer: An End-to-End Routability-Aware Placer with Graph Neural Network

An Accurate Detailed Routing Routability Prediction Model in Placement

Improving Routability Prediction via NAS Using a Smooth One-shot Augmented Predictor

Fast Routability Estimation for IC Physical Design