R2G: A Multi-View Circuit Graph Benchmark Suite from RTL to GDSII

2026-04-09

AI Summary

This work addresses the limitations imposed by inconsistent circuit graph representations and the absence of standardized evaluation protocols in physical design, which have hindered the advancement of graph neural networks. To bridge this gap, the authors introduce the first multi-view circuit graph benchmark suite spanning the full RTL-to-GDSII flow, offering aligned and information-equivalent graph representations across five key design stages. Accompanied by an end-to-end data processing pipeline and a unified evaluation protocol, this benchmark enables stage-aware, information-equivalent standardization for the first time, effectively decoupling graph representation from model choice. Experimental results demonstrate that the node-centric view exhibits the strongest generalization capability, achieving prediction performance with R² > 0.99 using only 3–4 decoder layers, and reveal that the choice of graph perspective exerts a far greater impact on performance than differences in model architecture.

Technology Category

• Machine Learning: Graph-based Machine Learning
• Reasoning under Uncertainty: Graphical Models
• Knowledge Representation and Reasoning: Computational Complexity of Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Graph neural networks (GNNs) are increasingly applied to physical design tasks such as congestion prediction and wirelength estimation, yet progress is hindered by inconsistent circuit representations and the absence of controlled evaluation protocols. We present R2G (RTL-to-GDSII), a multi-view circuit-graph benchmark suite that standardizes five stage-aware views with information parity (every view encodes the same attribute set, differing only in where features attach) over 30 open-source IP cores (up to $10^6$ nodes/edges). R2G provides an end-to-end DEF-to-graph pipeline spanning synthesis, placement, and routing stages, together with loaders, unified splits, domain metrics, and reproducible baselines. By decoupling representation choice from model choice, R2G isolates a confound that prior EDA and graph-ML benchmarks leave uncontrolled. In systematic studies with GINE, GAT, and ResGatedGCN, we find: (i) view choice dominates model choice, with Test R$^2$ varying by more than 0.3 across representations for a fixed GNN; (ii) node-centric views generalize best across both placement and routing; and (iii) decoder-head depth (3--4 layers) is the primary accuracy driver, turning divergent training into near-perfect predictions (R$^2$$>$0.99). Code and datasets are available at https://github.com/ShenShan123/R2G.

Problem

• circuit representation
• evaluation protocol
• physical design
• graph neural networks
• benchmark

Innovation

• multi-view circuit graph
• graph neural networks
• physical design automation
• benchmark suite
• representation-model decoupling