ABSTRACT
Fast and accurate pre-routing timing prediction is essential for timing-driven placement since repetitive routing and static timing analysis (STA) iterations are expensive and unacceptable. Prior work on timing prediction aims at estimating net delay and slew, lacking the ability to model global timing metrics. In this work, we present a timing engine inspired graph neural network (GNN) to predict arrival time and slack at timing endpoints. We further leverage edge delays as local auxiliary tasks to facilitate model training with increased model performance. Experimental results on real-world open-source designs demonstrate improved model accuracy and explainability when compared with vanilla deep GNN models.
- 2021. OpenCores. https://opencores.org/Google Scholar
- 2021. SkyWater. https://github.com/google/skywater-pdkGoogle Scholar
- Tutu Ajayi, Vidya A. Chhabria, Mateus Fogaça, et al. 2019. Toward an Open-Source Digital Flow: First Learnings from the OpenROAD Project. In Proc. DAC (DAC '19). Association for Computing Machinery, New York, NY, USA, Article 76, 4 pages. Google ScholarDigital Library
- C.J. Alpert, Jiang Hu, S.S. Sapatnekar, et al. 2004. Accurate estimation of global buffer delay within a floorplan. In Proc. ICCAD. 706--711. Google ScholarDigital Library
- Erick Carvajal Barboza, Nishchal Shukla, Yiran Chen, et al. 2019. Machine Learning-Based Pre-Routing Timing Prediction with Reduced Pessimism. In Proc. DAC. 1--6.Google ScholarDigital Library
- Wei-Ting J. Chan, Kun Young Chung, Andrew B. Kahng, et al. 2016. Learning-based prediction of embedded memory timing failures during initial floorplan design. In Proc. ASPDAC. 178--185. Google ScholarDigital Library
- Hongliang Chang and S.S. Sapatnekar. 2003. Statistical timing analysis considering spatial correlations using a single PERT-like traversal. In Proc. ICCAD. 621--625. Google ScholarCross Ref
- Ming Chen, Zhewei Wei, Zengfeng Huang, et al. 2020. Simple and deep graph convolutional networks. In Proc. ICML. PMLR, 1725--1735.Google Scholar
- Tianlong Chen, Kaixiong Zhou, Keyu Duan, et al. 2021. Bag of Tricks for Training Deeper Graph Neural Networks: A Comprehensive Benchmark Study. CoRR (2021). arXiv:2108.10521 https://arxiv.org/abs/2108.10521Google Scholar
- Tung-Chieh Chen, Tien-Chang Hsu, Zhe-Wei Jiang, et al. 2005. NTUplace: a ratio partitioning based placement algorithm for large-scale mixed-size designs. In Proc. ISPD. 236--238.Google ScholarDigital Library
- Chung-Kuan Cheng, Andrew B Kahng, Ilgweon Kang, et al. 2018. RePlAce: Advancing Solution Quality and Routability Validation in Global Placement. IEEE TCAD (2018).Google Scholar
- Seung-Soo Han, Andrew B. Kahng, Siddhartha Nath, et al. 2014. A deep learning methodology to proliferate golden signoff timing. In Proc. DATE. 1--6. Google ScholarCross Ref
- Thomas N. Kipf and Max Welling. 2016. Semi-Supervised Classification with Graph Convolutional Networks. CoRR abs/1609.02907 (2016). arXiv:1609.02907 http://arxiv.org/abs/1609.02907Google Scholar
- Qimai Li, Zhichao Han, and Xiao-Ming Wu. 2018. Deeper Insights Into Graph Convolutional Networks for Semi-Supervised Learning. In Proceedings of the Thirty-Second AAAI Conference on Artificial Intelligence. 3538--3545.Google ScholarCross Ref
- Yaguang Li, Yishuang Lin, Meghna Madhusudan, et al. 2020. A Customized Graph Neural Network Model for Guiding Analog IC Placement. In Proc. ICCAD. 1--9.Google ScholarDigital Library
- Rongjian Liang, Zhiyao Xie, Jinwook Jung, et al. 2020. Routing-Free Crosstalk Prediction. In Proc. ICCAD. 1--9.Google ScholarDigital Library
- Yibo Lin, Zixuan Jiang, Jiaqi Gu, et al. 2020. DREAMPlace: Deep Learning Toolkit-Enabled GPU Acceleration for Modern VLSI Placement. IEEE TCAD (2020).Google Scholar
- Mingjie Liu, Walker J. Turner, George F. Kokai, et al. 2021. Parasitic-Aware Analog Circuit Sizing with Graph Neural Networks and Bayesian Optimization. In Proc. DATE. 1372--1377. Google ScholarCross Ref
- Jingwei Lu, Pengwen Chen, Chin-Chih Chang, et al. 2015. ePlace: Electrostatics-based placement using fast fourier transform and Nesterov's method. ACM TODAES 20, 2 (2015), 17.Google ScholarDigital Library
- Yi-Chen Lu, Sai Surya Kiran Pentapati, Lingjun Zhu, et al. 2020. TP-GNN: A Graph Neural Network Framework for Tier Partitioning in Monolithic 3D ICs. In Proc. DAC. 1--6. Google ScholarCross Ref
- Azalia Mirhoseini, Anna Goldie, Mustafa Yazgan, et al. 2021. A graph placement methodology for fast chip design. Nature 594, 7862 (Jun 2021), 207--212. Google ScholarCross Ref
- Joelle Pineau, Philippe Vincent-Lamarre, Koustuv Sinha, et al. 2021. Improving Re-producibility in Machine Learning Research(A Report from the NeurIPS 2019 Re-producibility Program). Journal of Machine Learning Research 22, 164 (2021), 1--20. http://jmlr.org/papers/v22/20-303.htmlGoogle Scholar
- Haoxing Ren, George F. Kokai, Walker J. Turner, et al. 2020. ParaGraph: Layout Parasitics and Device Parameter Prediction using Graph Neural Networks. In Proc. DAC. 1--6. Google ScholarCross Ref
- Miodrag Vujkovic, David Wadkins, Bill Swartz, et al. 2004. Efficient Timing Closure without Timing Driven Placement and Routing. In Proc. DAC. 268--273. Google ScholarDigital Library
- Hanrui Wang, Kuan Wang, Jiacheng Yang, et al. 2020. GCN-RL Circuit Designer: Transferable Transistor Sizing with Graph Neural Networks and Reinforcement Learning. In Proc. DAC. Article 201, 6 pages.Google ScholarCross Ref
- Minjie Wang, Da Zheng, Zihao Ye, et al. 2019. Deep Graph Library: A Graph-Centric, Highly-Performant Package for Graph Neural Networks. arXiv preprint arXiv:1909.01315 (2019).Google Scholar
- Zhiyao Xie, Rongjian Liang, Xiaoqing Xu, et al. 2021. Net2: A Graph Attention Network Method Customized for Pre-Placement Net Length Estimation. In Proc. ASPDAC. 671--677.Google ScholarDigital Library
- Keren Zhu, Mingjie Liu, Hao Chen, et al. 2020. Exploring Logic Optimizations with Reinforcement Learning and Graph Convolutional Network. In Proceedings of the 2020 ACM/IEEE Workshop on Machine Learning for CAD. 145--150. Google ScholarDigital Library
Recommendations
Machine Learning-Based Pre-Routing Timing Prediction with Reduced Pessimism
DAC '19: Proceedings of the 56th Annual Design Automation Conference 2019Optimizations at placement stage need to be guided by timing estimation prior to routing. To handle timing uncertainty due to the lack of routing information, people tend to make very pessimistic predictions such that performance specification can be ...
How accurately can we model timing in a placement engine?
DAC '05: Proceedings of the 42nd annual Design Automation ConferenceThis paper presents a novel placement algorithm for timing optimization based on a new and powerful concept, which we term differential timing analysis. Recognizing that accurate optimization requires timing information from a signoff static timing ...
Slack in static timing analysis
The notion of slack is central in static timing analysis and very large scale integration (VLSI) design in general. Negative slack means that a timing constraint is violated, while a positive slack of x ps is intended to mean that an extra delay of x ps ...
Comments