Integrated Circuits and Systems >

2025 , Vol. 2 >Issue 4: 243 - 254

DOI: https://doi.org/10.23919/ICS.2025.3612817

Regular Papers

Simulation Framework and Design Exploration of in-Situ Error Correction for Multi-Bit Computation-in-Memory Circuits

  • TING-AN LIN ,
  • TOURANGBAM HARISHORE SINGH ,
  • PO-TSANG HUANG
Expand
  • International College of Semiconductor Technology, National Yang Ming Chiao Tung University, Hsinchu 30010, Taiwan
PO-TSANG HUANG (e-mail: ).

PO-TSANG HUANG(Member, IEEE)

Received date: 2025-04-12

  Revised date: 2025-08-23

  Accepted date: 2025-09-04

  Online published: 2025-12-24

Supported by

NSTC, Taiwan(NSTC 114-2218-E-A49-031-MBK)

NSTC, Taiwan(112-2628-E-A49-021-MY3)

NSTC, Taiwan(114-2634-FA49-001-)

“Center for Advanced Semiconductor Technology Research” from MOE in Taiwan

Fold

Abstract

As computational complexity continues to increase, effectively designing a computation-inmemory (CIM) architecture has become a crucial task. In such an architecture, errors may occur due to factors such as voltage drift. This work focuses on designing a simulation framework for In-Situ error correction of multi-bit memory-in-computing circuits. The research concentrates on In-Situ error correction techniques, allowing the system to instantly detect and correct errors during memory or computational operations at the same location where data is being processed and stored. The primary goal of this work is to explore how to minimize the impact of these errors on model accuracy. In constructing the simulation environment, multi-bit weights are decomposed, and 2D convolutions are decomposed into matrix multiplications, then mapped onto the CIM architecture. Based on this framework, this work further analyzes hardware errors in CIM, including the causes of errors, statistical characteristics, and the impact of extreme error values on accuracy. Furthermore, we introduce and deeply analyze clamping as an error correction technique. Through a series of simulations, we came to the following clear conclusion: To maximize hardware efficiency and accuracy correction effects, special attention must be paid to high-bit weights and the protection of sensitive convolutional layers. In addition, reasonable setting of clamping threshold and appropriate array-based output grouping strategy are also indispensable. These strategies provide clear optimization directions for neural networks in specific application scenarios. After considering the above strategies and optimizing, the model accuracy can reach a maximum of 73.8%, which is close to the baseline of 75.8%. Considering that the protection circuit area is reduced by 50%, this result shows excellent benefits.

Key words： In-situ error correction; multi-bit computation-in-memory

Cite this article

TING-AN LIN , TOURANGBAM HARISHORE SINGH , PO-TSANG HUANG . Simulation Framework and Design Exploration of in-Situ Error Correction for Multi-Bit Computation-in-Memory Circuits[J]. Integrated Circuits and Systems, 2025 , 2(4) : 243 -254 . DOI: 10.23919/ICS.2025.3612817

References

Publishing order | Descend order by publishing year | Descend order by cited within
[1]
K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proc. IEEE Conf. Comput. Vis. Pattern Recognit., 2016, pp. 770-778.

[2]
Z.-Q. Zhao, P. Zheng,S.-T. Xu, and X. Wu, “Object detection with deep learning: A review,” IEEE Trans. Neural Netw. Learn. Syst., vol. 30, no. 11, pp. 3212-3232, Nov. 2019.

[3]
J.-H. Yoon, M. Chang, W.-S. Khwa, Y.-D. Chih,M.-F. Chang, and A. Raychowdhury, “A 40-nm 118.44-TOPS/W voltage-sensing computein-memory RRAM macro with write verification and multi-bit encoding,” IEEE J. Solid-State Circuits, vol. 57, no. 3, pp. 845-857, Mar. 2022.

[4]
E. Choi et al., “A 333TOPS/W logic-compatible multi-level embedded flash compute-In-memory macro with dual-slope computation,” in Proc. IEEE Custom Integr. Circuits Conf., 2023, pp. 1-2.

[5]
J. M. Correll et al., “An 8-bit 20.7 TOPS/W multi-level cell ReRAMbased compute engine,” in Proc. IEEE Symp. VLSI Technol. Circuits (VLSI Technol. Circuits), 2022, pp. 264-265.

[6]
W. Song et al., “Programming memristor arrays with arbitrarily high precision for analog computing,” Science, vol. 383, no. 6685, pp. 903-910, 2024, doi:10.1126/science.adi9405.

[7]
M. J. Rasch et al., “Hardware-aware training for large-scale and diverse deep learning inference workloads using in-memory computingbased accelerators,” Nature Commun, vol. 14, 2023, Art. no. 5282, doi: 10.1038/s41467-023-40770-4.

[8]
D. Gao et al., “Bayesian inference based robust computing on memristor crossbar,” in Proc. ACM/IEEE Des. Automat. Conf., 2021, pp. 121-126.

[9]
Z. Yan,X. S. Hu, and Y. Shi, “SWIM: Selective write-Verify for computing-in-memory neural accelerators,” in Proc. ACM/IEEE Des. Automat. Conf., 2022, pp. 277-282.

[10]
M. Bernard et al., “DL-RSIM: A simulation framework to enable reliable ReRAM-based accelerators for deep learning,” in Proc. IEEE/ACM Int. Conf. Comput.-Aided Des., San Diego, CA, USA, 2018, pp. 1-8, doi: 10.1145/3240765.3240800.

[11]
Z. Xiao et al., “Adapting magnetoresistive memory devices for accurate and on-chip-training-free in-memory computing,” Sci. Adv., vol. 10, 2024, Art. no. eadp3710, doi: 10.1126/sciadv.adp3710.

[12]
H. Wei and R. M. Roth, “Multiple-error-correcting codes for analog computing on resistive crossbars,” IEEE Trans. Inf. Theory, vol. 70, no. 12, pp. 8647-8658, Dec. 2024, doi: 10.1109/TIT.2024.3439674.

[13]
S. Ambrogio et al., “Equivalent-accuracy accelerated neural-network training using analogue memory,” Nature, vol. 558, pp. 60-67, Jun. 2018.

[14]
B. Govoreanu et al., “10 × 10nm2 Hf/HfOx crossbar resistive RAM with excellent performance, reliability and low-energy operation,” in Proc. IEDM Tech. Dig., 2011, pp. 729-732.

[15]
C. E. Shannon, “A mathematical theory of communication,” Bell System Tech. J., vol. 27, no. 3, pp. 379-423, 1948.

[16]
C. Li, R. M. Roth, C. Graves,X. Sheng, and J. P. Strachan, “Analog error correcting codes for defect tolerant matrix multiplication in crossbars,” in Proc. IEEE Int. Electron Devices Meeting, 2020, pp. 36.6. 1-36.6.4.

[17]
R. M. Roth, “Analog error-correcting codes,” IEEE Trans. Inf. Theory, vol. 66, no. no. 7, pp. 4075-4088, Jul. 2020.

[18]
S. Jeong, J. Kim,M. Jeong, and Y. Lee, “Variation-tolerant and low Rratio compute-in-memory ReRAM macro with capacitive ternary MAC operation,” IEEE Trans. Circuits Syst. I: Reg. Papers, vol. 69, no. 7, pp. 2845-2856, Jul. 2022.

[19]
Z. He, J. Lin, R. Ewetz,J.-S. Yuan, and D. Fan, “Noise injection adaption: End-to-end ReRAM crossbar non-ideal effect adaption for neural network mapping,” in Proc. ACM/IEEE Des. Automat. Conf., 2019, pp. 1-6.

[20]
W. Jiang et al., “Device-circuit-architecture co-exploration for computing-in-memory neural accelerators,” IEEE Trans. Comput., vol. 70, no. 4, pp. 595-605, Apr. 2021.

[21]
Z. Yan, W. Jiang,X. S. Hu, and Y. Shi, “RADARS: Memory efficient reinforcement learning aided differentiable neural architecture search,” in Proc. IEEE Asia South Pacific Des. Automat. Conf., 2022, pp. 128-133.

[22]
K. Zhong et al., “Towards lower bit multiplication for convolutional neural network training,” 2020, arXiv:2006.02804.

[23]
N. Lepri, A. Glukhov, L. Cattaneo, M. Farronato,P. Mannocci, and D. Ielmini, “In-memory computing for Machine learning and Deep learning,” IEEE J. Electron Devices Soc., vol. 11, pp. 587-601, 2023.

[24]
R. W. Hamming, “Error detecting and error correcting codes,” Bell System Tech. J., vol. 29, no. 2, pp. 147-160, Apr. 1950, doi: 10.1002/j.1538-7305.1950.tb00463.x.

[25]
I. S. Reed and G. Solomon, “Polynomial codes over certain finite fields,” J. Soc. Ind. Appl. Math., vol. 8, no. 2, pp. 300-304, 1960.

[26]
C. Berrou, A. Glavieux, and P. Thitimajshima, “Near Shannon limit error-correcting coding and decoding: Turbo-codes. 1,” in Proc. ICC’93 IEEE Int. Conf. Commun., 1993, pp. 1064-1070.

[27]
R. Gallager, “Low-density parity-check codes,” IRE Trans. Inf. Theory, vol. 8, no. 1, pp. 21-28, 1962.

[28]
C. Li, R. M. Roth, C. Graves,X. Sheng, and J. P. Strachan, “Analog error correcting codes for defect tolerant matrix multiplication in crossbars,” in Proc. IEEE Int. Electron Devices Meeting, San Francisco, CA, USA, 2020, pp. 36.6. 1-36.6.4, doi: 10.1109/IEDM13553.2020.9371978.

[29]
K. Simonyan and A. Zisserman, “Very deep convolutional networks for large-scale image recognition,” 2014, arXiv:1409.1556.

[30]
K. Zhong et al., “Towards lower bit multiplication for convolutional neural network training,” 2020, arXiv:2006.02804.

[31]
N. Lepri, A. Glukhov, L. Cattaneo, M. Farronato,P. Mannocci, and D. Ielmini, “In-memory computing for machine learning and deep learning,” IEEE J. Electron Devices Soc., vol. 11, pp. 587-601, 2023.

[32]
K. Zhou, Y. He, R. Xiao,J. Liu, and K. Huang, “A customized NoC architecture to enable highly localized computing-on-the-move DNN dataflow,” IEEE Trans. Circuits Syst. II: Exp. Briefs, vol. 69, no. 3, pp. 1692-1696, Mar. 2022.

[33]
C. Liu et al., “Special session: Fault-tolerant deep learning: A hierarchical perspective,” in Proc. IEEE 40th VLSI Test Symp., 2022, pp. 1-12.

[34]
T. Andrulis, J. S. Emer, and V. Sze, “RAELLA: Reforming the arithmetic for efficient, low-resolution, and low-loss analog PIM: No retraining required!,” in Proc. 50th Annu. Int. Symp. Comput. Archit., 2023, pp. 1-16.

[35]
A. Lu et al., “A runtime reconfigurable design of compute-in-memorybased hardware accelerator for deep learning inference,” ACM Trans. Des. Automat. Electron. Syst., vol. 26, no. 6, pp. 1-18, 2021.

[36]
R. Xiao,W. Jiang, and P. Y. Chee, “An energy efficient timemultiplexing computing-in-memory architecture for edge intelligence,” IEEE J. Explor. Solid-State Comput. Devices Circuits, vol. 8, no. 2, pp. 111-118, Dec. 2022.

[37]
H. Cılasun et al., “On error correction for nonvolatile processing-Inmemory,” in Proc. ACM/IEEE 51st Annu. Int. Symp. Comput. Archit., Buenos Aires, Argentina, 2024, pp. 678-692, doi: 10.1109/ISCA59077. 2024.00055.

[38]
D. Shi, Y. Zhu, A. Fan, Y. Tao, Y. Yang, and B. Yan, “Nonbinary LDPC arithmetic error correction for processing-in-memory,” 2025, arXiv:2502.11487.

[39]
J.-W. Jang,T.-H. Nguyen, and J.-S. Yang, “VECOM: Variation resilient encoding and offset compensation schemes for reliable ReRAM based DNN accelerator,” in Proc. IEEE/ACM Int. Conf. Comput. Aided Des., San Francisco, CA, USA, 2023, pp. 1-9.

Options
Outlines

/

〈 〉