A Non-Idealities Aware Software–Hardware Co-Design Framework for Edge-AI Deep Neural Network Implemented on Memristive Crossbar

Cao, Tiancheng; Liu, Chen; Wang, Weijie; Zhang, Tantan; Lee, Hock Koon; Li, Ming Hua; Song, Wen‐Dong; Chen, Zhixian; Zhuo, Victor Yi-Qian; Wang, Nan; Zhu, Yao; Gao, Yuan; Goh, Wang Ling

doi:10.1109/jetcas.2022.3214334

Cited by 7 publications

(1 citation statement)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Nonidealities, including wire resistances, cause variations in the actual voltage across the memristor and memcapacitor VMM accelerator, resulting in a lower voltage than the theoretical value. This reduction is due to the accumulated voltage drop on the connecting traces and sneak pathways [29]. The presence of line resistance and sneak paths impacts the training accuracy of the model, as depicted from Fig.…”

Section: A Crossbar and Device Nonidealitiesmentioning

confidence: 99%

Framework for In-Memory Computing Based on Memristor and Memcapacitor for On-Chip Training

Singh,

Lee

2023

IEEE Access

View full text Add to dashboard Cite

Memristive crossbar arrays have gained considerable attention from researchers to perform analog in-memory vector-matrix multiplications in machine learning accelerators with low power and constant computational time. This work introduces a comprehensive framework for co-designing the software and hardware for deep neural networks (DNN) based on memristive and memcapacitive crossbars while considering various non-idealities. The model takes into account device-level factors, including conductance variation, cycle-to-cycle variation, device-to-device variation, peripheral circuits for error/weight gradient computation, and high tolerance. The overall neural network performance is thoroughly assessed by integrating these elements into a unified DNN training process. The proposed framework is implemented using a hybrid approach with Python and PyTorch. Performance evaluation was conducted using a simplified 8-layer VGG network on a measured 128×128 array with weight resolution. Remarkably, the memristive and memcapacitive crossbar arrays achieved outstanding training accuracies of 90.02% and 91.03%, respectively, for the CIFAR-10 dataset. Additionally, detailed hardware estimation for both mem-elements devices is provided, enabling meaningful comparisons with prior works.

show abstract

Section: A Crossbar and Device Nonidealitiesmentioning

confidence: 99%