A 2941-TOPS/W Charge-Domain 10T SRAM Compute-in-Memory for Ternary Neural Network
IEEE Transactions on Circuits and Systems I: Regular Papers · 2023-05 · 70 (5) · 2085-2097
DOI: 10.1109/tcsi.2023.3241385 ↗Kurz gefasst
Fachartikel in der Zeitschrift IEEE Transactions on Circuits and Systems I: Regular Papers, 2023-05. Verfasst von Sungsoo Cheon und 2 weiteren Autor:innen. Die Arbeit wurde bisher 39-mal zitiert (Crossref) und nennt 36 Literaturangaben. Gefördert von: National Research Foundation of Korea, Korea Government, Institute of Information and Communications Technology Planning and Evaluation (IITP), Korea Government (MSIT), EDA Tool, IC Design Education Center (IDEC), South Korea.
Abstract (Originalsprache)
In this paper, we present a 10T SRAM compute-in memory (CiM) macro to process the multiplication-accumulation (MAC) operations between ternary-inputs and binary-weights. In the proposed 10T SRAM bitcell, the charge-domain analog computations are employed to improve the noise tolerance of bit-line (BL) signals where the MAC results are represented in CiM. Parallel processing of 3 different analog levels for ternary input activations is also performed in the proposed single 10T bitcell. To reduce the analog-to-digital converter (ADC) bit-resolutions without sacrificing deep neural network (DNN) accuracies, a confined-slope non-uniform integration (CS-NUI) ADC is proposed, which can provide layer-wise adaptive quantization for multiple different layers with different MAC distributions. In addition, by sharing the ADC reference voltage generator in every single column of SRAM array, the ADC area is effectively reduced with improved energy efficiencies of CiM. The $256\times 64.10\text{T}$ SRAM CiM macro with the proposed charge-sharing scheme and CS-NUI ADCs has been implemented using 28nm CMOS process. The silicon measurement results show that the proposed CiM shows the accuracies of 98.66% and 88.48% with MNIST dataset on MLP, and CIFAR-10 dataset on VGGNet-7, respectively, with the energy efficiency of 2941-TOPS/W and the area efficiency of 59.584-TOPS/mm2.
Förderung
- National Research Foundation of Korea, Korea Government
- Institute of Information and Communications Technology Planning and Evaluation (IITP), Korea Government (MSIT)
- EDA Tool, IC Design Education Center (IDEC), South Korea
Zitieren
APA
Cheon, S., Lee, K., & Park, J. (2023). A 2941-TOPS/W Charge-Domain 10T SRAM Compute-in-Memory for Ternary Neural Network. IEEE Transactions on Circuits and Systems I: Regular Papers, 70(5), 2085-2097. https://doi.org/10.1109/tcsi.2023.3241385
Vancouver
Cheon S, Lee K, Park J. A 2941-TOPS/W Charge-Domain 10T SRAM Compute-in-Memory for Ternary Neural Network. IEEE Transactions on Circuits and Systems I: Regular Papers. 2023;70(5):2085-2097. doi:10.1109/tcsi.2023.3241385
BibTeX
@article{Cheon2023,
author = {Cheon, Sungsoo and Lee, Kyeongho and Park, Jongsun},
title = {A 2941-TOPS/W Charge-Domain 10T SRAM Compute-in-Memory for Ternary Neural Network},
journal = {IEEE Transactions on Circuits and Systems I: Regular Papers},
year = {2023},
volume = {70},
number = {5},
pages = {2085--2097},
doi = {10.1109/tcsi.2023.3241385}
}
EUVIMED