Advertisement

DVAFS—Dynamic-Voltage-Accuracy-Frequency-Scaling Applied to Scalable Convolutional Neural Network Acceleration

  • Bert Moons
  • Marian VerhelstEmail author
Chapter
  • 296 Downloads

Abstract

Dynamic-voltage-accuracy-frequency-scaling (DVAFS) is a generalization of the more common concept of dynamic-voltage-frequency-scaling (DVFS) where, instead of modulating voltage and frequency with changing throughput requirements, voltage and frequency are changed in scenarios with different accuracy requirements. The chapter introduces DVAFS, discusses its performance on the block- and system-level, and compares it to existing similar techniques. Finally, the chapter discusses Envision, a DVAFS-compatible silicon for convolutional neural network accelerator, and its use case in a hierarchical cascade for face recognition.

Keywords

Low power Embedded Deep learning Variable precision Computer architecture 

References

  1. 1.
    B. Barrois, Methods to evaluate accuracy-energy trade-off in operator-level approximate computing, Ph.D. thesis, Universite de Rennes, 2017Google Scholar
  2. 2.
    B. Barrois, K. Parashar, O. Sentieys, Leveraging power spectral density for scalable system-level accuracy evaluation, in Design, Automation & Test in Europe Conference & Exhibition (DATE), 2016 (IEEE, Piscataway, 2016), pp. 750–755Google Scholar
  3. 3.
    Y.-H. Chen, T. Krishna, J. Emer, V. Sze, Eyeriss: an energy-efficient reconfigurable accelerator for deep convolutional neural networks. ISSCC Digest of Technical Papers, pp. 262–263, 2016Google Scholar
  4. 4.
    S. Chippa, V. Venkataramani, S.T. Chakradhar, K. Roy, A. Raghunathan, Approximate computing: an integrated hardware approach, in Asilomar Conference on Signals, systems and computers, 2013Google Scholar
  5. 5.
    V.K. Chippa, S.T. Chakradhar, K. Roy, A. Raghunathan, Analysis and characterization of inherent application resilience for approximate computing, in 2013 50th ACM/EDAC/IEEE Design Automation Conference (DAC), 2013Google Scholar
  6. 6.
    M. de la Guia Solaz, W. Han, R. Conway, A flexible low power DSP with a programmable truncated multiplier, in TCAS-I, 2012Google Scholar
  7. 7.
    K. Goetschalckx, B. Moons, S. Lauwereins, M. Andraud, M. Verhelst, Optimized hierarchical cascaded processing. IEEE J. Emerging Sel. Top. Circuits Syst. 8(4), 884–894 (2018)CrossRefGoogle Scholar
  8. 8.
    G.B. Huang, M. Ramesh, T. Berg, E. Learned-Miller, Labeled faces in the wild: a database for studying face recognition in unconstrained environments. Technical report, Technical Report 07-49, University of Massachusetts, Amherst, 2007Google Scholar
  9. 9.
    J. Huang, J. Lach, G. Robins, A methodology for energy-quality tradeoff using imprecise hardware, in Proceedings of the 49th Annual Design Automation Conference (ACM, New York, 2012), pp. 504–509Google Scholar
  10. 10.
    A. Krizhevsky, I. Sutskever, G.E. Hinton, ImageNet classification with deep convolutional neural networks, in Advances in Neural Information Processing Systems, ed. by F. Pereira, C.J.C. Burges, L. Bottou, K.Q. Weinberger, vol. 25, pp. 1097–1105 (Curran Associates, Inc., Red Hook, 2012)Google Scholar
  11. 11.
    P. Kulkarni, P. Gupta, M. Ercegovac, Trading accuracy for power with an underdesigned multiplier architecture, in International Conference on VLSI Design, 2011Google Scholar
  12. 12.
    K.Y. Kyaw et al., Low-power high-speed multiplier for error-tolerant application, in Electron Devices and Solid-State Circuits (EDSSC), 2011Google Scholar
  13. 13.
    Y. LeCun, C. Cortes, The MNIST database of handwritten digits, 1998Google Scholar
  14. 14.
    Y. LeCun, L. Bottou, Y. Bengio, P. Haffner, Gradient-based learning applied to document recognition. Proc. IEEE 86(11), 2278–2324 (1998)CrossRefGoogle Scholar
  15. 15.
    Y. LeCun, Y. Bengio, G. Hinton, Deep learning. Nature 521(7553), 436–444 (2015)CrossRefGoogle Scholar
  16. 16.
    C. Liu, J. Han, F. Lombardi, A low-power, high performance approximate multiplier with configurable partial error recovery, in Design, Automation and Test in Europe (DATE), 2014Google Scholar
  17. 17.
    B. Moons, M. Verhelst, DVAS: dynamic voltage accuracy scaling for increased energy-efficiency in approximate computing, in International Symposium on Low Power Electronics and Design (ISLPED), 2015Google Scholar
  18. 18.
    B. Moons, M. Verhelst, A 0.3-2.6 tops/w precision-scalable processor for real-time large-scale ConvNets, in Proceedings of the IEEE Symposium on VLSI Circuits, pp. 178–179, July 2016Google Scholar
  19. 19.
    B. Moons, B. De Brabandere, L. Van Gool, M. Verhelst, Energy-efficient convnets through approximate computing, in Proceedings of the IEEE Winter Conference on Applications of Computer Vision (WACV), pp. 1–8, 2016Google Scholar
  20. 20.
    B. Moons, R. Uytterhoeven, W. Dehaene, M. Verhelst, DVAFS: trading computational accuracy for energy through dynamic-voltage-accuracy-frequency-scaling, in 2017 Design, Automation & Test in Europe Conference & Exhibition (DATE) (IEEE, Piscataway, 2017), pp. 488–493CrossRefGoogle Scholar
  21. 21.
    B. Moons, K. Goetschalckx, N. Van Berckelaer, M. Verhelst, Minimum energy quantized neural networks, in Asilomar Conference on Signals, Systems and Computers, 2017Google Scholar
  22. 22.
    B. Moons, R. Uytterhoeven, M. Dehaene, W. Verhelst, Envision: a 0.26-to-10 tops/w subword-parallel dynamic-voltage-accuracy-frequency-scalable convolutional neural network processor in 28nm FDSOI, in International Solid-State Circuits Conference (ISSCC), 2017Google Scholar
  23. 23.
    O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. Bernstein et al., ImageNet large scale visual recognition challenge. Int. J. Comput. Vis. 115(3), 211–252 (2015)MathSciNetCrossRefGoogle Scholar
  24. 24.
    S. Venkataramani et al., Quality programmable vector processors for approximate computing, in 46th Annual IEEE/ACM International Symposium on Microarchitecture (MICRO), 2013Google Scholar
  25. 25.
    B. Widrow, Statistical analysis of amplitude-quantized sampled-data systems. Trans. Am. Inst. Electr. Eng. II: Appl. Ind. 79(6), 555–568 (1961)Google Scholar
  26. 26.
    B. Wu, M. Willems, Rapid architectural exploration in designing application-specific processors, in ASIP Designer Whitepaper, 2015Google Scholar

Copyright information

© Springer Nature Switzerland AG 2020

Authors and Affiliations

  1. 1.ESAT/MICASKU LeuvenLeuvenBelgium

Personalised recommendations