A Fully Hardware Implemented Accelerator Design in ReRAM Analog Computing without ADCs

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Dang, Peng, Li, Huawei, Wang, Wei
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866913627952381952
author Dang, Peng
Li, Huawei
Wang, Wei
author_facet Dang, Peng
Li, Huawei
Wang, Wei
contents Emerging ReRAM-based accelerators process neural networks via analog Computing-in-Memory (CiM) for ultra-high energy efficiency. However, significant overhead in peripheral circuits and complex nonlinear activation modes constrain system energy efficiency improvements. This work explores the hardware implementation of the Sigmoid and SoftMax activation functions of neural networks with stochastically binarized neurons by utilizing sampled noise signals from ReRAM devices to achieve a stochastic effect. We propose a complete ReRAM-based Analog Computing Accelerator (RACA) that accelerates neural network computation by leveraging stochastically binarized neurons in combination with ReRAM crossbars. The novel circuit design removes significant sources of energy/area efficiency degradation, i.e., the Digital-to-Analog and Analog-to-Digital Converters (DACs and ADCs) as well as the components to explicitly calculate the activation functions. Experimental results show that our proposed design outperforms traditional architectures across all overall performance metrics without compromising inference accuracy.
format Preprint
id arxiv_https___arxiv_org_abs_2412_19869
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle A Fully Hardware Implemented Accelerator Design in ReRAM Analog Computing without ADCs
Dang, Peng
Li, Huawei
Wang, Wei
Hardware Architecture
Artificial Intelligence
Emerging ReRAM-based accelerators process neural networks via analog Computing-in-Memory (CiM) for ultra-high energy efficiency. However, significant overhead in peripheral circuits and complex nonlinear activation modes constrain system energy efficiency improvements. This work explores the hardware implementation of the Sigmoid and SoftMax activation functions of neural networks with stochastically binarized neurons by utilizing sampled noise signals from ReRAM devices to achieve a stochastic effect. We propose a complete ReRAM-based Analog Computing Accelerator (RACA) that accelerates neural network computation by leveraging stochastically binarized neurons in combination with ReRAM crossbars. The novel circuit design removes significant sources of energy/area efficiency degradation, i.e., the Digital-to-Analog and Analog-to-Digital Converters (DACs and ADCs) as well as the components to explicitly calculate the activation functions. Experimental results show that our proposed design outperforms traditional architectures across all overall performance metrics without compromising inference accuracy.
title A Fully Hardware Implemented Accelerator Design in ReRAM Analog Computing without ADCs
topic Hardware Architecture
Artificial Intelligence
url https://arxiv.org/abs/2412.19869