Binarized Neural Network Accelerator Macro Using Ultralow-Voltage Retention SRAM for Energy Minimum-Point Operation

Binarized Neural Network Accelerator Macro Using Ultralow-Voltage Retention SRAM for Energy Minimum-Point Operation
复制标题

使用超低压保持 SRAM 进行能量最小点操作的二值化神经网络加速器宏

DOI:
10.1109/jxcdc.2022.3225744
复制
发表时间:
2022
影响因子:
2.4
通讯作者:
Sugahara Satoshi
Sugahara Satoshi
中科院分区:
--
文献类型:
--
作者:
Shiotsu Yusaku;Sugahara Satoshi

文献摘要

参考文献

相似文献

提出了一种基于内存处理(PIM)/内存计算(CIM)架构的二值化神经网络(BNN)加速器,使用超低电压保留静态随机存取存储器(ULVR-SRAM)进行能量最小点(EMP)操作。 BNN 加速器 (BNA) 宏旨在在超低电压 (< EMP) 下使用 ULVR 在 EMP 和实质性功率门控 (PG) 上执行稳定的推理操作,可应用于任意形状和尺寸的全连接层 (FCL)。通过将 ULVR-SRAM 应用于宏来实现 BNA 宏的 EMP 操作,可以显着提高能源效率 (TOPS/W),并显着增加并行乘法累加 (MAC) 操作的数量。此外,BNA宏的ULVR模式也受益于ULVR-SRAM的使用,可有效降低待机功耗。所提出的 BNA 宏可以为 FCL 显示 65 TOPS/W 的高能效。这种使用 ULVR-SRAM 的 BNA 宏概念可以扩展到卷积层,其中 EMP 操作也有望提高卷积层的能量效率。
A binarized neural network (BNN) accelerator based on a processing-in-memory (PIM)/ computing-in-memory (CIM) architecture using ultralow-voltage retention static random access memory (ULVR-SRAM) is proposed for the energy minimum-point (EMP) operation. The BNN accelerator (BNA) macro is designed to perform stable inference operations at EMP and substantive power-gating (PG) using ULVR at an ultralow voltage (< EMP), which can be applied to fully connected layers (FCLs) with arbitrary shapes and sizes. The EMP operation of the BNA macro, which is enabled by applying the ULVR-SRAM to the macro, can dramatically improve the energy efficiency (TOPS/W) and significantly enlarge the number of parallelized multiply–accumulate (MAC) operations. In addition, the ULVR mode of the BNA macro, which also benefits from the usage of ULVR-SRAM, is effective at reducing the standby power. The proposed BNA macro can show a high energy efficiency of 65 TOPS/W for FCLs. This BNA macro concept using the ULVR-SRAM can be expanded to convolution layers, where the EMP operation is also expected to enhance the energy efficiency of convolution layers.
功能存储器发展趋势(LSI存储器特刊)
DOI: --
发表时间: 1993
影响因子: 0.5
作者:
K. Tamaru
通讯作者: K. Tamaru
具有使用页眉和页脚电源开关的电源门控单元架构的超低压保持 SRAM
DOI: 10.1109/ojcas.2021.3104945
发表时间: 2021
影响因子: 2.6
作者:
Yoshida Hayato;Shiotsu Yusaku;Kitagata Daiki;Yamamoto Shuu'ichirou;Sugahara Satoshi
通讯作者: Sugahara Satoshi
通过 XNOR 网络在边缘启用 AI
DOI: 10.1145/3429945
发表时间: 2020
影响因子: 22.7
作者:
Mohammad Rastegari;Vicente Ordonez;Joseph Redmon;Ali Farhadi
通讯作者: Ali Farhadi