SCALE: Self-Supervised Constraint-Aware Layout GEneration for Local P&R DRV Fixing at Advanced Nodes
Quick Answer
SCALE introduces a self-supervised layout generation framework for fixing local design-rule violations (DRV) in sub-2nm semiconductor nodes, achieving a 12-25% improvement in solve rates for complex violations.
Quick Take
By leveraging a fine-tuned language model, it generates DRC-annotated layout-violation pairs, enhancing rule-aware geometric guidance for DRV repair, validated by an industrial DRC checker.
Key Points
- SCALE addresses complex rule interactions and dense multi-layer routing geometries.
- The framework uses a fine-tuned language model to reconstruct masked polygons.
- It produces DRC-annotated layout-violation pairs for further model training.
- Achieves up to 97% solve rates on 100 real sub-2nm DRV cases.
- Improves state-of-the-art agents' performance by 12-25%.
DeepSignal Analysis
What happened
SCALE is a new framework designed to address local design-rule violations (DRV) in semiconductor manufacturing at sub-2nm nodes. It utilizes a self-supervised layout generation process that improves the solve rates for complex violations by 12-25%, validated by an industrial DRC checker.
Key evidence
- SCALE employs a fine-tuned language model to reconstruct polygons from surrounding context, enhancing the understanding of layout violations.
- The framework generates DRC-annotated layout-violation pairs, which are used to fine-tune a domain-adapted DRC-VLM for improved guidance.
- SCALE achieved a solve rate improvement of up to 97% on 100 real sub-2nm cases, addressing various types of violations.
Why it matters
As semiconductor technology progresses towards smaller nodes, the complexity of design-rule violations increases, making efficient fixing methods crucial. SCALE's approach leverages advanced AI techniques to enhance the layout generation process, potentially leading to more efficient manufacturing workflows and reduced time-to-market for advanced semiconductor products.
What to watch
Paper Resources
📖 Reader Mode
~2 min readAbstract:As semiconductor manufacturing advances toward sub-2nm nodes, local place-and-route (P&R) design-rule violation (DRV) fixing is increasingly limited by complex rule interactions, dense multi-layer routing geometries, and foundry-specific constraints. While Large Language Models (LLMs) have recently demonstrated strong capabilities in EDA scripting and documentation, their application to visual layout understanding remains largely unexplored: diagnosing DRC violations from layout imagery demands precise geometric reasoning and foundry-specific rule knowledge absent from general-purpose VLM training. We propose SCALE, a framework with a self-supervised layout-generation stage for local DRV fixing at advanced nodes. Multi-layer layout geometry is serialized into structured text, and a fine-tuned language model learns to reconstruct randomly masked polygons from surrounding BEOL context alone without violation labels. At inference, natural-language rule constraints and high-temperature sampling steer generation toward diverse, violation-prone layout variants validated by an industrial signoff DRC checker, producing DRC-annotated layout--violation pairs used to fine-tune a domain-adapted DRC-VLM. This VLM provides rule-aware geometric guidance for local DRV repair, boosting state-of-the-art agents' solve rates by +12--25% (up to 97%) on 100 real sub-2nm cases spanning enclosure, spacing, width, and color-spacing violations.
| Comments: | 8 pages, 5 Figures, 6 Tables |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI) |
| Cite as: | arXiv:2607.21850 [cs.CV] |
| (or arXiv:2607.21850v1 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2607.21850 arXiv-issued DOI via DataCite (pending registration) |
Submission history
From: Chiatung Ho [view email]
[v1]
Thu, 23 Jul 2026 22:38:04 UTC (10,296 KB)
— Originally published at arxiv.org
Want this in your inbox every morning?
Daily brief at your local 8am — bilingual EN/中文, free.
More from arXiv cs.CV
See more →ProMoE-FL: Prototype-conditioned Mixture of Experts for Multimodal Federated Learning with Missing Modalities
ProMoE-FL introduces a Prototype-conditioned Mixture-of-Experts framework for multimodal federated learning, effectively addressing missing modalities. It outperforms existing methods on four chest X-ray datasets, demonstrating superior feature synthesis capabilities in both homogeneous and heterogeneous settings.