eF2lowSim: System-Level Simulator of eFlash-Based Compute-in-Memory Accelerators for Convolutional Neural Networks

被引：2

作者：

Wang, Jooho ^{[1
,2
]}

Kim, Sunwoo ^{[1
,2
]}

Heo, Junsu ^{[1
]}

Park, Chester Sungchung ^{[1
]}

机构：

[1] Konkuk Univ, Dept Elect & Elect Engn, Seoul, South Korea

[2] Samsung Elect Inc, Memory Business, Suwon, South Korea

来源：

2023 DESIGN, AUTOMATION & TEST IN EUROPE CONFERENCE & EXHIBITION, DATE | 2023年

关键词：

Compute-in-memory (CIM); convolutional neural network (CNN); dataflow; embedded flash (eFlash); hardware accelerators; system-level simulators;

D O I：

10.23919/DATE56975.2023.10137200

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

A new system-level simulator, eF(2)lowSim, is proposed to estimate the bit-accurate and cycle-accurate performance of eFlash compute-in-memory (CIM) accelerators for convolutional neural networks. The eF(2)lowSim can predict the inference accuracy by considering the impact of circuit nonideality such as program disturbance. Moreover, the eF(2)lowSim can also evaluate the system-level performance of dataflow strategies that have a significant impact on hardware area and performance of eFlash CIM accelerators. The simulator helps to find the optimal dataflow strategy of an eFlash CIM accelerator for each convolutional layer. It is shown that the improvement of area efficiency amounts to 26.8%, 21.2% and 17.9% in the case of LeNet-5, VGG-9 and ResNet-18, respectively.

引用

页数：6

共 5 条

[1] A System-Level Exploration of Binary Neural Network Accelerators with Monolithic 3D Based Compute-in-Memory SRAM
Choi, Jeong Hwan
Gong, Young-Ho
Chung, Sung Woo
ELECTRONICS, 2021, 10 (05) : 1 - 11
[2] CiMComp: An Energy Efficient Compute-in-Memory based Comparator for Convolutional Neural Networks
Kavitha, S.
Kailath, Binsu J.
Reniwal, B. S.
2024 DESIGN, AUTOMATION & TEST IN EUROPE CONFERENCE & EXHIBITION, DATE, 2024,
[3] CMOS-compatible compute-in-memory accelerators based on integrated ferroelectric synaptic arrays for convolution neural networks
Kim, Min-Kyu
Kim, Ik-Jyae
Lee, Jang-Sik
SCIENCE ADVANCES, 2022, 8 (14)
[4] Heterogeneous integration of 2D memristor arrays and silicon selectors for compute-in-memory hardware in convolutional neural networks
Samarth Jain
Sifan Li
Haofei Zheng
Lingqi Li
Xuanyao Fong
Kah-Wee Ang
Nature Communications, 16 (1)
[5] APIM: An Antiferromagnetic MRAM-Based Processing-In-Memory System for Efficient Bit-Level Operations of Quantized Convolutional Neural Networks
Li, Yueting
Wang, Jinkai
Zhu, Daoqian
Li, Jinhao
Du, Ao
Wang, Xueyan
Zhang, Yue
Zhao, Weisheng
IEEE TRANSACTIONS ON COMPUTER-AIDED DESIGN OF INTEGRATED CIRCUITS AND SYSTEMS, 2024, 43 (08) : 2405 - 2410

← 1 →