Abstract
Binary Neural Networks (BNNs) are an efficient alternative to traditional neural networks as they use binary weights and activations, leading to significant reductions in memory footprint and computational energy. However, the design of efficient BNN accelerators is a challenge due to the large design space. Multiple factors have to be considered during the design, among them are the type of data flow and the organization of the accelerator architecture. To the best of our knowledge, a tool for the design space exploration of BNN accelerators with regards to these factors does not exist.
In this work, we propose DAEBI, a tool for the design space exploration of BNN accelerators, which enables designers to identify the most suitable data flow and accelerator architecture. DAEBI automatically generates VHDL-code for BNN accelerator designs based on user specifications, making it convenient to explore large design spaces. Using DAEBI, we conduct a design space exploration of BNN accelerators for traditional CMOS technology using an FPGA. Our results demonstrate the capabilities of DAEBI and provide insights into the most suitable design choices. Additionally, based on a decision model, we provide insights for the design of BNN accelerator specifications that use emerging beyond-CMOS technologies.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Ando, K., et al.: BRein memory: a single-chip binary/ternary reconfigurable in-memory deep neural network accelerator achieving 1.4 TOPS at 0.6 W. IEEE J. Solid-State Circ. (JSSC) 53(4), 983–994 (2017)
Andri, R., Cavigelli, L., Rossi, D., Benini, L.: YodaNN: an ultra-low power convolutional neural network accelerator based on binary weights. In: 2016 IEEE Computer Society Annual Symposium on VLSI (ISVLSI), pp. 236–241 (2016)
Bertuletti, M., noz Martín, I.M., Bianchi, S., Bonfanti, A.G., Ielmini, D.: A multilayer neural accelerator with binary activations based on phase-change memory. IEEE Trans. Electron Devices 70(3), 986–992 (2023)
Blott, M., et al.: FINN-R: an end-to-end deep-learning framework for fast exploration of quantized neural networks. ACM Trans. Reconfigurable Technol. Syst. (TRETS) 11(3), 1–23 (2018)
Boukhobza, J., Rubini, S., Chen, R., Shao, Z.: Emerging NVM: a survey on architectural integration and research. Challenges 23(2), 1084–4309 (2017)
Chang, L., et al.: PXNOR-BNN: in/with spin-orbit torque MRAM preset-XNOR operation-based binary neural networks. IEEE Trans. Very Large Scale Integr. (VLSI) Syst. 27(11), 2668–2679 (2019)
Chen, X., Yin, X., Niemier, M., Hu, X.S.: Design and optimization of FeFET-based crossbars for binary convolution neural networks. In: 2018 Design, Automation, Test in Europe (DATE), pp. 1205–1210 (2018)
Chen, Y.H., Emer, J., Sze, V.: Eyeriss: a spatial architecture for energy-efficient dataflow for convolutional neural networks. In: 2016 ACM/IEEE 43rd Annual International Symposium on Computer Architecture (ISCA), pp. 367–379 (2016)
Courbariaux, M., Hubara, I., Soudry, D., El-Yaniv, R., Bengio, Y.: Binarized neural networks: training deep neural networks with weights and activations constrained to +1 or \(-\)1. arXiv preprint arXiv:1602.02830 (2016)
Dave, A., Frustaci, F., Spagnolo, F., Yayla, M., Chen, J.J., Amrouch, H.: HW/SW codesign for approximation-aware binary neural networks. IEEE J. Emerg. Sel. Top. Circ. Syst. 13(1), 33–47 (2023)
Deng, J., Dong, W., Socher, R., Li, L.J., Li, K., Fei-Fei, L.: ImageNet: a large-scale hierarchical image database. In: CVPR 2009 (2009)
George, S., et al.: Nonvolatile memory design based on ferroelectric FETs. In: 2016 53nd ACM/EDAC/IEEE Design Automation Conference (DAC), pp. 1–6 (2016)
Goldberger, A.L., et al.: PhysioBank, PhysioToolkit, and PhysioNet: components of a new research resource for complex physiologic signals. Circulation 101(23), e215–e220 (2000)
Hirtzlin, T., et al.: Outstanding bit error tolerance of resistive RAM-based binarized neural networks. In: 2019 IEEE International Conference on Artificial Intelligence Circuits and Systems (AICAS), pp. 288–292 (2019)
Horowitz, M.: 1.1 computing’s energy problem (and what we can do about it). In: 2014 IEEE International Solid-State Circuits Conference Digest of Technical Papers (ISSCC), pp. 10–14. IEEE (2014)
Ko, D.H., Oh, T.W., Lim, S., Kim, S.K., Jung, S.O.: Comparative analysis and energy-efficient write scheme of ferroelectric FET-based memory cells. IEEE Access 9, 127895–127905 (2021)
Latotzke, C., Gemmeke, T.: Efficiency versus accuracy: a review of design techniques for DNN hardware accelerators. IEEE Access 9, 9785–9799 (2021)
Li, G., Zhang, M., Zhang, Q., Lin, Z.: Efficient binary 3D convolutional neural network and hardware accelerator. J. Real-Time Image Process. 19(1), 61–71 (2022)
Li, Y., Chen, Y., Jones, A.K.: A software approach for combating asymmetries of non-volatile memories. In: Proceedings of the 2012 ACM/IEEE International Symposium on Low Power Electronics and Design, ISLPED 2012, pp. 191–196 (2012)
Ni, K., Li, X., Smith, J.A., Jerry, M., Datta, S.: Write disturb in ferroelectric FETs and its implication for 1T-FeFET AND memory arrays. IEEE Electron Device Lett. 39(11), 1656–1659 (2018)
Nurvitadhi, E., Sheffield, D., Sim, J., Mishra, A., Venkatesh, G., Marr, D.: Accelerating binarized neural networks: comparison of FPGA, CPU, GPU, and ASIC. In: 2016 International Conference on Field-Programmable Technology (FPT), pp. 77–84 (2016)
Resch, S., et al.: PIMBALL: binary neural networks in spintronic memory. ACM Trans. Archit. Code Optim. (TACO) 16(4), 1–26 (2019)
Sari, E., Belbahri, M., Nia, V.P.: How does batch normalization help binary training? arXiv:1909.09139 (2019)
Soliman, T., et al.: Efficient FeFET crossbar accelerator for binary neural networks. In: 2020 IEEE 31st International Conference on Application-specific Systems, Architectures and Processors (ASAP), pp. 109–112 (2020)
Stadtmann, T., Latotzke, C., Gemmeke, T.: From quantitative analysis to synthesis of efficient binary neural networks. In: 2020 19th IEEE International Conference on Machine Learning and Applications (ICMLA), pp. 93–100. IEEE (2020)
Sun, X., et al.: Fully parallel RRAM synaptic array for implementing binary neural network with (+1, \(-\)1) weights and (+1, 0) neurons. In: 2018 23rd Asia and South Pacific Design Automation Conference (ASP-DAC), pp. 574–579 (2018)
Sunny, F.P., Mirza, A., Nikdast, M., Pasricha, S.: Robin: a robust optical binary neural network accelerator. ACM Trans. Embed. Comput. Syst. (TECS) 20(5), 1–24 (2021)
Suresh, A., Cicotti, P., Carrington, L.: Evaluation of emerging memory technologies for HPC, data intensive applications. In: 2014 IEEE International Conference on Cluster Computing (CLUSTER), pp. 239–247 (2014)
Tu, Z., Chen, X., Ren, P., Wang, Y.: AdaBin: improving binary neural networks with adaptive binary sets. In: Avidan, S., Brostow, G., Cissé, M., Farinella, G.M., Hassner, T. (eds.) ECCV 2022. LNCS, vol. 13671, pp. 379–395. Springer, Cham (2022). https://doi.org/10.1007/978-3-031-20083-0_23
Wu, Q., et al.: A Non-volatile computing-in-memory ReRAM macro using two-bit current-mode sensing amplifier. In: 2021 IEEE 10th Non-volatile Memory Systems and Applications Symposium (NVMSA), pp. 1–6 (2021)
Yayla, M., Chen, J.J.: Memory-efficient training of binarized neural networks on the edge. In: Proceedings of the 59th ACM/IEEE Design Automation Conference (DAC) (2022)
Yayla, M., et al.: Reliable binarized neural networks on unreliable beyond Von-Neumann architecture. IEEE Trans. Circuits Syst. I Regul. Pap. 69(6), 2516–2528 (2022)
Zangeneh, M., Joshi, A.: Performance and energy models for memristor-based 1T1R RRAM cell. In: Proceedings of the Great Lakes Symposium on VLSI (GLSVLSI 2012), pp. 9–14 (2012)
Zhang, Y., Chen, G., He, T., Huang, Q., Huang, K.: ViraEye: an energy-efficient stereo vision accelerator with binary neural network in 55 nm CMOS. In: Proceedings of the 28th Asia and South Pacific Design Automation Conference, pp. 178–179 (2023)
Acknowledgements
This paper has been supported by Deutsche Forschungsgemeinschaft (DFG) project OneMemory (405422836), by the Collaborative Research Center SFB 876 “Providing Information by Resource-Constrained Analysis” (project number 124020371), subproject A1 (http://sfb876.tu-dortmund.de) and by the Federal Ministry of Education and Research of Germany and the state of NRW as part of the Lamarr-Institute for ML and AI, LAMARR22B.
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
Yayla, M., Latotzke, C., Huber, R., Iskif, S., Gemmeke, T., Chen, JJ. (2023). DAEBI: A Tool for Data Flow and Architecture Explorations of Binary Neural Network Accelerators. In: Silvano, C., Pilato, C., Reichenbach, M. (eds) Embedded Computer Systems: Architectures, Modeling, and Simulation. SAMOS 2023. Lecture Notes in Computer Science, vol 14385. Springer, Cham. https://doi.org/10.1007/978-3-031-46077-7_8
Download citation
DOI: https://doi.org/10.1007/978-3-031-46077-7_8
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-46076-0
Online ISBN: 978-3-031-46077-7
eBook Packages: Computer ScienceComputer Science (R0)