Low Power Neural Network by Reducing SRAM Operating Voltage

Low Power Neural Network by Reducing SRAM Operating Voltage
复制标题

通过降低 SRAM 工作电压实现低功耗神经网络

DOI:
10.1109/access.2022.3219208
复制
发表时间:
2022
期刊:
影响因子:
3.9
通讯作者:
Namba Kazuteru
Namba Kazuteru
中科院分区:
计算机科学3区
文献类型:
--
作者:
Kozu Keisuke;Tanabe Yuya;Kitakami Masato;Namba Kazuteru

文献摘要

参考文献

被引文献

相似文献

随着机器学习技术的进步,网络变得越来越复杂,所涉及的计算范围也越来越大。因此,增加了学习过程的计算时间和功耗。作为解决这一问题的一种方法,神经网络的容错性引起了人们的关注。由于神经网络可以容忍很小的误差,因此有可能以牺牲精度为代价来降低计算速度和功耗。在这项研究中,我们提出了一种方法来降低电路的功耗,通过降低的静态随机存取存储器(SRAM),用于存储权重的工作电压。在所提出的方法中,使用两种不同的SRAM的工作电压,我们使用不同的误码率(BER)的容错和非容错。我们展示了BER和识别率之间的关系,以及保持高识别率的BER和电路配置的适当组合。
With advancements in machine learning technology, networks are becoming increasingly complex, and the extent of the computation involved is increasing. Consequently, the computation time and power consumption of the learning process are increased. The error tolerance of neural networks has attracted attention as an approach to solving this problem. Because neural networks can tolerate small errors, it is possible to reduce the calculation speed and power consumption at the expense of accuracy. In this study, we propose a method to reduce the power consumption of the circuit by lowering the operating voltage of the static random-access memory (SRAM) that is utilized to store the weights. In the proposed method, using two different operating voltages of SRAM, we used different bit error rates (BERs) for error-tolerant and non-error-tolerant. We demonstrated the relationship between the BER and recognition rate, and the appropriate combination of the BER and circuit configuration that maintains a high recognition rate.
DOI: 10.1147/jrd.2017.2716579
发表时间: 2017-07
期刊: IBM J. Res. Dev.
影响因子: --
作者:
P. Narayanan;Alessandro Fumarola;Lucas L. Sanches;K. Hosokawa;S. Lewis;R. Shelby;G. Burr
通讯作者: P. Narayanan;Alessandro Fumarola;Lucas L. Sanches;K. Hosokawa;S. Lewis;R. Shelby;G. Burr
使用模拟存储设备加速深度神经网络
DOI: --
发表时间: 2019
期刊: China Semiconductor Technology International Conference
影响因子: --
作者:
G. Burr;S. Ambrogio;P. Narayanan;H. Tsai;C. Mackin;An Chen
通讯作者: An Chen
用于实现三元神经网络的 SRAM 突触阵列的低 VDD 操作
DOI: 10.1109/tvlsi.2017.2727528
发表时间: 2017
影响因子: 2.8
作者:
Xiaoyu Sun;Rui Liu;Yi;Hsiao;Wei;Meng;Shimeng Yu
通讯作者: Shimeng Yu
采用 16 nm 技术在亚阈值区域设计低功耗、容变单位线 9T SRAM 单元
DOI: 10.1016/j.microrel.2021.114126
发表时间: 2021
影响因子: 1.6
作者:
Chandramauleshwar Roy;A. Islam
通讯作者: A. Islam
CIFAR-10 二值化卷积神经网络处理器的误码容限
DOI: 10.1109/iscas.2018.8351255
发表时间: 2018
期刊: International Symposium on Circuits and Systems
影响因子: --
作者:
Lita Yang;Daniel Bankman;Bert Moons;M. Verhelst;B. Murmann
通讯作者: B. Murmann