Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Skip to main content

Advertisement

Springer Nature Link
Log in

LayCO: Achieving Least Lossy Accuracy for Most Efficient RRAM-Based Deep Neural Network Accelerator via Layer-Centric Co-Optimization

  • Regular Paper
  • Published:
Journal of Computer Science and Technology Aims and scope Submit manuscript

Abstract

Resistive random access memory (RRAM) enables the functionality of operating massively parallel dot products and accumulations. RRAM-based accelerator is such an effective approach to bridging the gap between Internet of Things devices’ constrained resources and deep neural networks’ tremendous cost. Due to the huge overhead of Analog to Digital (A/D) and digital accumulations, analog RRAM buffer is introduced to extend the processing in analog and in approximation. Although analog RRAM buffer offers potential solutions to A/D conversion issues, the energy consumption is still challenging in resource-constrained environments, especially with enormous intermediate data volume. Besides, critical concerns over endurance must also be resolved before the RRAM buffer could be frequently used in reality for DNN inference tasks. Then we propose LayCO, a layer-centric co-optimizing scheme to address the energy and endurance concerns altogether while strictly providing an inference accuracy guarantee. LayCO relies on two key ideas: 1) co-optimizing with reduced supply voltage and reduced bit-width of accelerator architectures to increase the DNN’s error tolerance and achieve the accelerator’s energy efficiency, and 2) efficiently mapping and swapping individual DNN data to a corresponding RRAM partition in a way that meets the endurance requirements. The evaluation with representative DNN models demonstrates that LayCO outperforms the baseline RRAM buffer based accelerator by 27x improvement in energy efficiency (over TIMELY-like configuration), 308x in lifetime prolongation and 6x in area reduction (over RAQ) while maintaining the DNN accuracy loss less than 1%.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Subscribe and save

Springer+ Basic
$34.99 /Month
  • Get 10 units per month
  • Download Article/Chapter or eBook
  • 1 Unit = 1 Article or 1 Chapter
  • Cancel anytime
Subscribe now

Buy Now

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Explore related subjects

Discover the latest articles, news and stories from top researchers in related subjects.

References

  1. Jin H, Liu B, Jiang W B, Ma Y, Shi X H, He B S, Zhao S F. Layer-centric memory reuse and data migration for extreme-scale deep learning on many-core architectures. ACM Trans. Architecture and Code Optimization, 2018, 15(3): Article No. 37. https://doi.org/10.1145/3243904.

  2. Yang T J, Chen Y H, Sze V. Designing energy-efficient convolutional neural networks using energy-aware pruning. In Proc. the 2017 IEEE Conference on Computer Vision and Pattern Recognition, Jul. 2017, pp.6071–6079. https://doi.org/10.1109/CVPR.2017.643.

  3. He K M, Zhang X Y, Ren S Q, Sun J. Deep residual learning for image recognition. In Proc. the 2016 IEEE Conference on Computer Vision and Pattern Recognition, Jun. 2016, pp.770–778. https://doi.org/10.1109/CVPR.2016.90.

  4. Xie S N, Girshick R, Dollár P, Tu Z W, He K M. Aggregated residual transformations for deep neural networks. In Proc. the 2017 IEEE Conference on Computer Vision and Pattern Recognition, Jul. 2017, pp.5987–5995. https://doi.org/10.1109/CVPR.2017.634.

  5. Chen Y H, Emer J, Sze V. Eyeriss: A spatial architecture for energy-efficient dataflow for convolutional neural networks. ACM SIGARCH Computer Architecture News, 2016, 44(3): 367–379. https://doi.org/10.1145/3007787.3001177.

    Article  Google Scholar 

  6. Gao M Y, Yang X, Pu J, Horowitz M, Kozyrakis C. TANGRAM: Optimized coarse-grained dataflow for scalable NN accelerators. In Proc. the 24th International Conference on Architectural Support for Programming Languages and Operating Systems, Apr. 2019, pp.807–820. https://doi.org/10.1145/3297858.3304014.

  7. Chi P, Li S C, Xu C, Zhang T, Zhao J S, Liu Y P, Wang Y, Xie Y. PRIME: A novel processing-in-memory architecture for neural network computation in ReRAM-based main memory. ACM SIGARCH Computer Architecture News, 2016, 44(3): 27–39. https://doi.org/10.1145/3007787.3001140.

    Article  Google Scholar 

  8. Shafiee A, Nag A, Muralimanohar N, Balasubramonian R, Strachan J P, Hu M, Williams R S, Srikumar V. ISAAC: A convolutional neural network accelerator with in-situ analog arithmetic in crossbars. ACM SIGARCH Computer Architecture News, 2016, 44(3): 14–26. https://doi.org/10.1145/3007787.3001139.

    Article  Google Scholar 

  9. Song L H, Qian X H, Li H, Chen Y R. PipeLayer: A pipelined ReRAM-based accelerator for deep learning. In Proc. the 2017 IEEE International Symposium on High Performance Computer Architecture, Feb. 2017, pp.541–552. https://doi.org/10.1109/HPCA.2017.55.

  10. Zhu Z H, Sun H B, Lin Y J, Dai G H, Xia L X, Han S, Wang Y, Yang H Z. A configurable multi-precision CNN computing framework based on single bit RRAM. In Proc. the 56th Annual Design Automation Conference, Jun. 2019, Article No. 56. https://doi.org/10.1145/3316781.3317739.

  11. Chou T, Tang W, Botimer J, Zhang Z Y. CASCADE: Connecting RRAMs to extend analog dataflow in an endto-end in-memory processing paradigm. In Proc. the 52nd Annual IEEE/ACM International Symposium on Microarchitecture, Oct. 2019, pp.114–125. https://doi.org/10.1145/3352460.3358328.

  12. Li W T, Xu P F, Zhao Y, Li H T, Xie Y, Lin Y Y. Timely: Pushing data movements and interfaces in PIM accelerators towards local and in time domain. In Proc. the 2020 ACM/IEEE Annual International Symposium on Computer Architecture, May 30–June 3, 2020, pp.832–845. https://doi.org/10.1109/ISCA45697.2020.00073.

  13. Waser R, Dittmann R, Staikov G, Szot K. Redox-based resistive switching memories-nanoionic mechanisms, prospects, and challenges. Advanced Materials, 2009, 21(25/26): 2632–2663. https://doi.org/10.1002/adma.200900375.

    Article  Google Scholar 

  14. Wong H S P, Lee H Y, Yu S M, Chen Y S, Wu Y, Chen P S, Lee B, Chen F T, Tsai M J. Metal-oxide RRAM. Proceedings of the IEEE, 2012, 100(6): 1951–1970. https://doi.org/10.1109/JPROC.2012.2190369.

    Article  Google Scholar 

  15. Chou C C, Lin Z J, Tseng P L, Li C F, Chang C Y, Chen W C, Chih Y D, Chang T Y J. An N40 256K×44 embedded RRAM macro with SL-precharge SA and low-voltage current limiter to improve read and write performance. In Proc. the 2018 IEEE International Solid-State Circuits Conference, Feb. 2018, pp.478–480. https://doi.org/10.1109/ISSCC.2018.8310392.

  16. Yang J G, Xue X Y, Xu X X, Wang Q, Jiang H J, Yu J, Dong D N, Zhang F, Lv H B, Liu M. 24.2 A 14nm-Fin-FET 1Mb embedded 1T1R RRAM with a 0.022μm2 cell size using self-adaptive delayed termination and multi-cell reference. In Proc. the 2021 IEEE International Solid-State Circuits Conference, Feb. 2021, pp.336–338. https://doi.org/10.1109/ISSCC42613.2021.9365945.

  17. Yao P, Wu H Q, Gao B, Eryilmaz S B, Huang X Y, Zhang W Q, Zhang Q T, Deng N, Shi L P, Wong H S P, Qian H. Face classification using electronic synapses. Nature Communications, 2017, 8: Article No. 15199. https://doi.org/10.1038/ncomms15199.

  18. Strukov D B. Endurance-write-speed tradeoffs in nonvolatile memories. Applied Physics A, 2016, 122(4): Article No. 302. https://doi.org/10.1007/s00339-016-9841-0.

  19. Vogelsang T. Understanding the energy consumption of dynamic random access memories. In Proc. the 43rd Annual IEEE/ACM International Symposium on Microarchitecture, Dec. 2010, pp.363–374. https://doi.org/10.1109/MICRO.2010.42.

  20. Koppula S, Orosa L, Yağlıkçı A G, Azizi R, Shahroodi T, Kanellopoulos K, Mutlu O. EDEN: Enabling energy-efficient, high-performance deep neural network inference using approximate DRAM. In Proc. the 52nd Annual IEEE/ACM International Symposium on Microarchitecture, Oct. 2019, pp.166–181. https://doi.org/10.1145/3352460.3358280.

  21. Indiveri G, Linn E, Ambrogio S. RERAM-based neuromorphic computing. In Resistive Switching: From Fundamentals of Nanoionic Redox Processes to Memristive Device Applications, Ielmini D, Waser R (eds.), Wiley-VCH, 2016, pp.715–736. https://doi.org/10.1002/9783527680870.ch25.

  22. Chandramoorthy N, Swaminathan K, Cochet M, Paidimarri A, Eldridge S, Joshi R V, Ziegler M M, Buyuktosunoglu A, Bose P. Resilient low voltage accelerators for high energy efficiency. In Proc. the 2019 IEEE International Symposium on High Performance Computer Architecture, Feb. 2019, pp.147–158. https://doi.org/10.1109/HPCA.2019.00034.

  23. Sandrini J. Fabrication, characterization and integration of resistive random access memories [Ph. D. Thesis]. École Polytechnique Fédérale De Lausanne, Switzerland, 2017. https://doi.org/10.5075/epfl-thesis-8097.

  24. Hirtzlin T, Bocquet M, Klein J O, Nowak E, Vianello E, Portal J M, Querlioz D. Outstanding bit error tolerance of resistive RAM-based binarized neural networks. In Proc. the 2019 IEEE International Conference on Artificial Intelligence Circuits and Systems, Mar. 2019, pp. 288–292. https://doi.org/10.1109/AICAS.2019.8771544.

  25. Li G P, Hari S K S, Sullivan M, Tsai T, Pattabiraman K, Emer J, Keckler S W. Understanding error propagation in deep learning neural network (DNN) accelerators and applications. In Proc. the International Conference for High Performance Computing, Networking, Storage and Analysis, Nov. 2017, Article No. 8. https://doi.org/10.1145/3126908.3126964.

  26. Geng Q H, Zhou Z, Cao X C. Survey of recent progress in semantic image segmentation with CNNs. SCIENCE CHINA Information Sciences, 2018, 61(5): Article No. 051101. https://doi.org/10.1007/s11432-017-9189-6.

  27. Krizhevsky A. Learning multiple layers of features from tiny images. Technical Report, TR-2009, University of Toronto, 2009. http://www.cs.toronto.edu/~kriz/learningfeatures-2009-TR.pdf, Mar. 2023.

  28. Xia L X, Liu M Y, Ning X F, Chakrabarty K, Wang Y. Fault-tolerant training with on-line fault detection for RRAM-based neural computing systems. In Proc. the 54th Annual Design Automation Conference 2017, Jun. 2017, Article No. 33. https://doi.org/10.1145/3061639.3062248.

  29. Ketkar N, Moolayil J. Introduction to PyTorch. In Deep Learning with Python: Learn Best Practices of Deep Learning Models with PyTorch, Ketkar N, Moolayil J (eds.), Apress, 2021, pp.27–91. https://doi.org/10.1007/978-1-4842-5364-9_2.

  30. Krizhevsky A, Sutskever I, Hinton G E. ImageNet classification with deep convolutional neural networks. In Proc. the 25th International Conference on Neural Information Processing Systems, Dec. 2012, pp.1097–1105. https://doi.org/10.5555/2999134.2999257.

  31. Simonyan K, Zisserman A. Very deep convolutional networks for large-scale image recognition. In Proc. the 3rd International Conference on Learning Representations, May 2015.

  32. Deng J, Dong W, Socher R, Li L J, Li K, Li F F. ImageNet: A large-scale hierarchical image database. In Proc. the 2009 IEEE Conference on Computer Vision and Pattern Recognition, Jun. 2009, pp.248–255. https://doi.org/10.1109/CVPR.2009.5206848.

  33. Cai Y, Lin Y J, Xia L X, Chen X M, Han S, Wang Y, Yang H Z. Long live TIME: Improving lifetime for training-in-memory engines by structured gradient sparsification. In Proc. the 55th Annual Design Automation Conference, Jun. 2018, Article No. 107. https://doi.org/10.1145/3195970.3196071.

  34. Wang F, Luo G J, Sun G Y, Wang Y H, Niu D M, Zheng H Z. Area efficient pattern representation of binary neural networks on RRAM. Journal of Computer Science and Technology, 2021, 36(5): 1155–1166. https://doi.org/10.1007/s11390-021-0906-y.

    Article  Google Scholar 

  35. Liu B, Cai H, Wang Z, Sun Y H, Shen Z Y, Zhu W T, Li Y, Gong Y, Ge W, Yang J, Shi L X. A 22nm, 10.8 μW/15.1 μW dual computing modes high power-performance-area efficiency domained background noise aware keyword-spotting processor. IEEE Trans. Circuits and Systems I: Regular Papers, 2020, 67(12): 4733–4746. https://doi.org/10.1109/TCSI.2020.2997913.

    Article  Google Scholar 

  36. Liu B, Cai H, Zhang Z L, Ding X L, Wang Z Y, Gong Y, Liu W Q, Yang J J, Wang Z, Yang J. More is less: Domain-specific speech recognition microprocessor using onedimensional convolutional recurrent neural network. IEEE Trans. Circuits and Systems I: Regular Papers, 2022, 69(4): 1571–1582. https://doi.org/10.1109/TCSI.2021.3134271.

    Article  Google Scholar 

  37. Xiao T P, Feinberg B, Bennett C H, Agrawal V, Saxena P, Prabhakar V, Ramkumar K, Medu H, Raghavan Chettuvetty R, Agarwal S, Marinella M J. An accurate, error-tolerant, and energy-efficient neural network inference engine based on SONOS analog memory. IEEE Trans. Circuits and Systems I: Regular Papers, 2022, 69(4): 1480–1493. https://doi.org/10.1109/TCSI.2021.3134313.

    Article  Google Scholar 

Download references

Author information

Authors and Affiliations

  1. Wuhan National Laboratory for Optoelectronics, Huazhong University of Science and Technology, Wuhan, 430074, China

    Shao-Feng Zhao, Fang Wang & Dan Feng

  2. School of Computer Science and Technology, Huazhong University of Science and Technology, Wuhan, 430074, China

    Shao-Feng Zhao, Fang Wang & Dan Feng

  3. Cloud Computing and Big Data Institute, Henan University of Economics and Law, Zhengzhou, 450001, China

    Shao-Feng Zhao & Yang Liu

  4. Research Institute of Huazhong University of Science and Technology in Shenzhen, Shenzhen, 518057, China

    Fang Wang

  5. School of Computer and Artificial Intelligence, Zhengzhou University, Zhengzhou, 450001, China

    Bo Liu

Authors
  1. Shao-Feng Zhao
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Fang Wang
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Bo Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Dan Feng
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Yang Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding authors

Correspondence to Fang Wang or Bo Liu.

Additional information

Fang Wang initiated the project and coordinated the research. Bo Liu worked on the conceptualization and methodology.

Supplementary Information

ESM 1

(PDF 101 kb)

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Zhao, SF., Wang, F., Liu, B. et al. LayCO: Achieving Least Lossy Accuracy for Most Efficient RRAM-Based Deep Neural Network Accelerator via Layer-Centric Co-Optimization. J. Comput. Sci. Technol. 38, 328–347 (2023). https://doi.org/10.1007/s11390-023-2545-y

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s11390-023-2545-y

Keywords

Search

Navigation