Application of Neural Networks for Image Segmentation in the Prob-lem of Fast Global Routing
https://doi.org/10.15514/ISPRAS-2022-35(5)-10
Abstract
The paper explores the possibilities of using neural network methods to solve the problem of global routing for VLSI ASIC design. An algorithm has been developed for generating a training dataset based on the Lee algorithm, which allows one to synthesize three-dimensional matrices with obstacles and points that need to be connected. The U-Net fully convolutional neural network, effective for semantic segmentation of images, was selected for training. The quality of the results was assessed using a validation data. A significant reduction in routing time compared to the Lee algorithm was shown, but the share of unbroken routes was only 37%. Ways to improve the training dataset and adapt the approach to real conditions using DEF and GUIDE files are proposed. In general, the work demonstrated the potential of neural network methods to speed up the global routing task, but continued research is required to improve the quality and reliability of the results. The work is useful for specialists in the field of integrated circuit design and machine learning.
About the Authors
Timur Maratovich KADIRLIEV
Institute of Design Problems in Microelectronics, RAS (IPPM RAS)
Russian Federation
Russian Federation
University graduate, engineer-researcher of the Institute of Design Problems in Microelectronics RAS. Research interests: machine learning and neural networks, data science, web development, automation of digital VLSI design.
Dmitry Vladimirovich TELPUKHOV
Institute of Design Problems in Microelectronics, RAS (IPPM RAS)
Russian Federation
Russian Federation
Dr. Sci. (Tech.), Deputy Director for Scientific Work of the Institute of Design Problems in Microelectronics RAS. Research interests: automation of digital VLSI design, logic synthesis, radiation-resistant design, machine learning and neural networks, residual class system.
Roman Aleksandrovich SOLOVYEV
Institute of Design Problems in Microelectronics, RAS (IPPM RAS)
Russian Federation
Russian Federation
Dr. Sci. (Tech.), member-corr. RAS, chief researcher at the Institute of Design Problems in Microelectronics RAS. Area of scientific interests: automation of digital VLSI design, neural networks and machine learning, residue number system, physical synthesis of integrated circuits.
References
1. F. Rubin, "The Lee Path Connection Algorithm," in IEEE Transactions on Computers, vol. C-23, no. 9, pp. 907-914, Sept. 1974, doi: 10.1109/T-C.1974.224054.
2. James, Gareth & Witten, Daniela & Hastie, Trevor & Tibshirani, Robert. (2013). An Introduction to Statistical Learning: With Applications in R., Springer, New York, NY https://doi.org/10.1007/978-1-4614-7138-7.
3. O'Shea, K., & Nash, R. (2015). An introduction to convolutional neural networks. arXiv preprint arXiv:1511.08458.
4. Ronneberger, O., Fischer, P., Brox, T. (2015). U-Net: Convolutional Networks for Biomedical Image Segmentation. In: Navab, N., Hornegger, J., Wells, W., Frangi, A. (eds) Medical Image Computing and Computer-Assisted Intervention – MICCAI 2015. MICCAI 2015. Lecture Notes in Computer Science, vol 9351. Springer, Cham. https://doi.org/10.1007/978-3-319-24574-4_28.
5. Sudre, C.H., Li, W., Vercauteren, T., Ourselin, S., Jorge Cardoso, M. (2017). Generalised Dice Over-lap as a Deep Learning Loss Function for Highly Unbalanced Segmentations. In: Cardoso, M., et al. Deep Learning in Medical Image Analysis and Multimodal Learning for Clinical Decision Support. DLMIA ML-CDS 2017 2017. Lecture Notes in Computer Science, vol 10553. Springer, Cham. https://doi.org/10.1007/978-3-319-67558-9_28
6. Ghazy, A.A., & Shalan, M. (2020). OpenLANE: The Open-Source Digital ASIC Implementation Flow. in Proc. Workshop on Open-Source EDA Technol. (WOSET), 2020, Art. no. 21.
7. LEF/DEF 5.8 Language Reference. Доступно по ссылке: https://coriolis.lip6.fr/doc/lefdef/lefdefref/DEFSyntax.html, дата обращения — апрель 2023.
Review
For citations:
KADIRLIEV T.M., TELPUKHOV D.V., SOLOVYEV R.A. Application of Neural Networks for Image Segmentation in the Prob-lem of Fast Global Routing. Proceedings of the Institute for System Programming of the RAS (Proceedings of ISP RAS). 2023;35(5):145-156. (In Russ.) https://doi.org/10.15514/ISPRAS-2022-35(5)-10
Email this article 