Abstract
With the increasing demand for edge AI application scenarios, as the most popular deep learning models, Convolutional Neural Networks (CNNs) need advanced solutions for the deployment of highly energy-efficient implementations. This paper presents a novel approach to improve the efficiency of CNN inference on Field-Programmable Gate Arrays (FPGAs) using Partial Reconfiguration (PR). Our method deconstructs CNN topology into different layers for runtime reconfiguration with fewer resources, aiming to significantly reduce static power and overall energy consumption. To identify the conditions for practical PR efficiency, we present a thorough design space exploration study with three CNN benchmarks, each evaluated across three different implementations. The comparison results demonstrate that our PR approach can achieve up to 3.88 and 1.67 times energy savings compared to software and static hardware implementations, respectively. These results also show that the benefits of PR improve with the depth of the network, suggesting very promising levels of gains as the network gets larger and under the key conditions of using fast optimized reconfiguration controllers and methodical system-level exploration of the increased hardware implementation complexity.
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
Similar content being viewed by others
References
Farhadi, M., Ghasemi, M., Yang, Y.: A novel design of adaptive and hierarchical convolutional neural networks using partial reconfiguration on FPGA. In: IEEE High Performance Extreme Computing Conference (HPEC), Waltham, MA, USA, pp. 1-7 (2019). https://doi.org/10.1109/HPEC.2019.8916237.
Youssef, E., Elsimary, H.A., El-Moursy, M.A., Mostafa, H., Khattab, A.: Energy-efficient precision-scaled cnn implementation with dynamic partial reconfiguration. IEEE Access 10, 95571–95584 (2022). https://doi.org/10.1109/ACCESS.2022.3204704
Youssef, E., Elsemary, H.A., El-Moursy, M.A., Khattab, A., Mostafa, H.: Energy Adaptive convolution neural network using dynamic partial reconfiguration. in: IEEE 63rd International Midwest Symposium on Circuits and Systems (MWSCAS), vol. 2020, pp. 325–328. Springfield, MA, USA (2020). https://doi.org/10.1109/MWSCAS48704.2020.9184640
Irmak, H., Ziener, D., Alachiotis, N.: Increasing Flexibility of FPGA-based CNN Accelerators with Dynamic Partial Reconfiguration. In: 2021 31st International Conference on Field-Programmable Logic and Applications (FPL), Dresden, Germany, 2021, pp. 306-311. https://doi.org/10.1109/FPL53798.2021.00061.
Alberto de Albuquerque Silva, C., Andrey Ramalho Diniz, A., Duarte Dória Neto, A., Alberto Nicolau de Oliveira, J.: Use of partial reconfiguration for the implementation and embedding of the artificial neural network (ANN) in FPGA. In: 4th International Conference on Pervasive and Embedded Computing and Communication System (2014). https://doi.org/10.5220/0004716301420150.
Venieris, S.I., Bouganis, C.S.: fpgaConvNet: a framework for mapping convolutional neural networks on FPGAs. In: IEEE International Symposium on Field-Programmable Custom Computing Machines (FCCM), Washington, DC, USA, vol. 2016, pp. 40–47 (2016). https://doi.org/10.1109/FCCM.2016.22
Kastner, F., Janben, B., Kautz, F., Hubner, M., Corradi, G.: Hardware, software codesign for convolutional neural networks exploiting dynamic partial reconfiguration on PYNQ. In: IEEE International Parallel and Distributed Processing Symposium Workshops (IPDPSW), Vancouver, BC, Canada, vol. 2018, pp. 154–161 (2018). https://doi.org/10.1109/IPDPSW.2018.00031
Bonamy, R., et al.: Energy efficient mapping on manycore with dynamic and partial reconfiguration: application to a smart camera. Inter. J. Circuit Theory Appli. (2018). https://doi.org/10.1002/cta.2508
Bonamy, R., Bilavarn, S., Chillet, D., Sentieys, O.: Power modeling and exploration of dynamic and partially reconfigurable systems. J. Low Power Electr. (2016). https://doi.org/10.1166/jolpe.2016.1448
Bonamy, R., Pham, H.-M., Pillement, S., Chillet, D.: UPaRC-Ultra-fast power-aware reconfiguration controller. In: Design, Automation & Test in Europe Conference & Exhibition (DATE), Dresden, Germany, vol. 2012, pp. 1373–1378 (2012). https://doi.org/10.1109/DATE.2012.6176705
Duhem, F., Muller, F., Bonamy, R., Bilavarn, S.: Fortress: a flow for design space exploration of partially reconfigurable systems. Design Autom. Embedded Syst., 301-326 (2015). https://doi.org/10.1007/s10617-015-9160-2.
Sadek, A., et al.: Supporting utilities for heterogeneous embedded image processing platforms (STHEM): an overview. In: International Symposium on Applied Reconfigurable Computing (ARC) (2018). https://doi.org/10.1007/978-3-319-78890-6_59
Rongshi, D., Yongming, T.: Accelerator implementation of Lenet-5 convolution neural network based on FPGA with HLS. In: 2019 3rd International Conference on Circuits, System and Simulation (ICCSS), Nanjing, China, pp. 64-67 (2019). https://doi.org/10.1109/CIRSYSSIM.2019.8935599.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2024 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Li, Z., Bilavarn, S. (2024). Improving the Energy Efficiency of CNN Inference on FPGA Using Partial Reconfiguration. In: Dias, T., Busia, P. (eds) Design and Architectures for Signal and Image Processing. DASIP 2024. Lecture Notes in Computer Science, vol 14622. Springer, Cham. https://doi.org/10.1007/978-3-031-62874-0_8
Download citation
DOI: https://doi.org/10.1007/978-3-031-62874-0_8
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-62873-3
Online ISBN: 978-3-031-62874-0
eBook Packages: Computer ScienceComputer Science (R0)