Sree Krishna Raja K,
P. Jaya Krishna,
- Student, Nalla Malla Reddy Engineering College, Hyderabad, India
- Assistant Professor, Nalla Malla Reddy Engineering College, Hyderabad, India
Abstract
The performance of modern VLSI systems is heavily influenced by power constraints, necessitating precise power estimation and effective optimization techniques. Traditional methods, such as gate-level simulations, are often slow and computationally intensive. This paper introduces DRPENN (Differential Algorithm for Routing and Placement Optimization using Neural Networks), an innovative solution that combines a Switching Activity Estimator (SAE) with a neural network-assisted differential algorithm. By leveraging toggle rates from simulations to train a Graph Neural Network (GNN), DRPENN circumvents the need for extensive gate-level simulations. This trained model accurately predicts switching activity, optimizing the placement and routing of circuit elements while minimizing computational demands. The methodology involves converting gate-level netlists into graphical representations and using a differential algorithm for back-propagation. Our experimental results show that DRPENN achieves lower error rates and faster inference times compared to conventional probabilistic SAE methods, offering a promising approach for efficient and accurate VLSI design optimization.
Keywords: VLSI design, routing optimization, placement optimization, differential algorithm, neural networks
[This article belongs to Journal of VLSI Design Tools and Technology(jovdtt)]
References
- Aisha Fayomi; Amal S. Hassan; Hanan Baaqeel; Ehab M. Almetwally., Bayesian Inference and Data Analysis of the Unit–Power Burr X Distribution. in Axioms, 2023, pp. 12(3), 297.
- Seyed Morteza Nabavinejad; Sherief Reda and Masoumeh Ebrahimi., BatchSizer: Power-Performance Trade-off for DNN Inference. in ACM, 2023, pp 7-13.
1.Yuan Zhou; Haxong Ren et al., “PRIMAL: Power Inference Using Machine Learning,” in DAC, 2019, pp. 39:1–39:6. 2.Yanqing Zhang; Haoxing Ren et al., “GRANNITE: Graph Neural Network Inference for Transferable Power Estimation” in IEEE, 2020, 978-1- 7281-1085-1/20
- Murata, H. Ishibuchi. “Performance evaluation of genetic algorithms for flowshop scheduling problems,” Proceedings of 1st IEEE Conference on Evolutionary Computation, 1994, pp. 812-817.
- Liu, T. Qingfu, K.L. Ma. “A novel differential evolution algorithm for solving integer programming problems,” IEEE Transactions on Systems, Man, and Cybernetics, Part C (Applications and Reviews), 2010, Vol. 40, Issue 5, pp. 714-724.
- Kennedy, R. Eberhart. “Particle Swarm Optimization,” Proceedings of IEEE International Conference on Neural Networks, 1995, pp. 1942-1948.
- Dorigo, T. Stützle. “Ant Colony Optimization,” MIT Press, 2004.
- Mitchell. “An Introduction to Genetic Algorithms,” MIT Press, 1996.
- Kirkpatrick, C. D. Gelatt, M. P. Vecchi. “Optimization by Simulated Annealing,” Science, 1983, Vol. 220, No. 4598, pp. 671-680.
- Shi, R. Eberhart. “A Modified Particle Swarm Optimizer,” Proceedings of IEEE International Conference on Evolutionary Computation, 1998, pp. 69-73.
- Loshchilov, F. Hutter. “CMA-ES for Hyperparameter Optimization of Deep Neural Networks,” Proceedings of the International Conference on Learning Representations (ICLR), 2016.
Journal of VLSI Design Tools and Technology
Volume | 14 |
Issue | 02 |
Received | July 19, 2024 |
Accepted | July 31, 2024 |
Published | August 7, 2024 |