Abstract
The energy efficiency of hardware implementations of convolutional neural networks (CNNs) is critical to their widespread deployment in low-power mobile devices. Recently, a plethora of methods have been proposed providing energy-optimal mappings of CNNs onto diverse hardware accelerators. Their estimated power consumption is related to specific implementation details and hardware parameters, which does not allow for machine-independent exploration of CNN energy measures. In this paper, we introduce a simplified theoretical energy complexity model for CNNs, based on only two-level memory hierarchy that captures asymptotically all important sources of power consumption of different CNN hardware implementations. We calculate energy complexity in this model for two common dataflows which, according to statistical tests, fits asymptotically very well the power consumption estimated by the Time/Accelergy program for convolutional layers on the Simba and Eyeriss hardware platforms. The model opens the possibility of proving principal limits on the energy efficiency of CNN hardware accelerators.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Alwani, M., et al.: Fused-layer CNN accelerators. In: Proceedings of the IEEE/ACM MICRO 2016, pp. 22:1–22:12 (2016). https://doi.org/10.1109/MICRO.2016.7783725
Chen, Y., Emer, J.S., Sze, V.: Eyeriss: A spatial architecture for energy-efficient dataflow for convolutional neural networks. In: Proceedings of the ACM/IEEE ISCA 2016, pp. 367–379 (2016). https://doi.org/10.1109/ISCA.2016.40
Kao, S.C., Krishna, T.: GAMMA: Automating the HW mapping of DNN models on accelerators via genetic algorithm. In: Proceedings of the ACM/IEEE ICCAD 2020, pp. 44:1–44:9 (2020). https://doi.org/10.1145/3400302.3415639
Mittal, S.: A survey of techniques for approximate computing. ACM Comput. Surv. 48(4), 62:1–62:33 (2016). https://doi.org/10.1145/2893356
Parashar, A., et al.: Timeloop: A systematic approach to DNN accelerator evaluation. In: Proceedings of the IEEE ISPASS 2019, pp. 304–315 (2019). https://doi.org/10.1109/ISPASS.2019.00042
Shao, Y.S., et al.: Simba: Scaling deep-learning inference with multi-chip-module-based architecture. In: Proceedings of the IEEE/ACM MICRO 2019, pp. 14–27 (2019). https://doi.org/10.1145/3352460.3358302
Šíma, J., Cabessa, J.: Energy complexity of fully-connected layers. In: Proceedings of the IWANN 2023. LNCS 14134, Part I, Springer, Berlin (2023)
Sze, V., Chen, Y., Yang, T., Emer, J.S.: Efficient Processing of Deep Neural Networks. Synthesis Lectures on Computer Architecture. Morgan & Claypool Publishers (2020). https://doi.org/10.2200/S01004ED1V01Y202004CAC050
Wu, Y.N., Emer, J.S., Sze, V.: Accelergy: An architecture-level energy estimation methodology for accelerator designs. In: Proceedings of the IEEE/ACM ICCAD 2019 (2019). https://doi.org/10.1109/ICCAD45719.2019.8942149
Yang, T., Chen, Y., Emer, J.S., Sze, V.: A method to estimate the energy consumption of deep neural networks. In: Proceedings of the IEEE ACSSC 2017, pp. 1916–1920 (2017). https://doi.org/10.1109/ACSSC.2017.8335698
Acknowledgements
The research was supported by the Czech Science Foundation grant GA22-02067S and the institutional support RVO: 67985807 (J. Šíma, P. Vidnerová). We thank Jan Kalina for his advice on statistical tests.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Šíma, J., Vidnerová, P., Mrázek, V. (2023). Energy Complexity Model for Convolutional Neural Networks. In: Iliadis, L., Papaleonidas, A., Angelov, P., Jayne, C. (eds) Artificial Neural Networks and Machine Learning – ICANN 2023. ICANN 2023. Lecture Notes in Computer Science, vol 14263. Springer, Cham. https://doi.org/10.1007/978-3-031-44204-9_16
Download citation
DOI: https://doi.org/10.1007/978-3-031-44204-9_16
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-44203-2
Online ISBN: 978-3-031-44204-9
eBook Packages: Computer ScienceComputer Science (R0)