RxNN: A Framework for Evaluating Deep Neural Networks on Resistive Crossbars

TL;DR

RxNN is a fast simulation framework that accurately evaluates the impact of resistive crossbar non-idealities on large-scale DNN accuracy, revealing significant potential degradations in real hardware implementations.

Contribution

This work introduces RxNN, the first efficient simulation tool for assessing large-scale DNN accuracy on resistive crossbar hardware with non-idealities.

Findings

Resistive crossbar non-idealities cause 9.6%-32% accuracy loss in large DNNs.

RxNN is 10,000 to 100,000 times faster than circuit-level simulation.

Large-scale DNNs are significantly affected by hardware non-idealities, impacting their deployment.

Abstract

Resistive crossbars designed with non-volatile memory devices have emerged as promising building blocks for Deep Neural Network (DNN) hardware, due to their ability to compactly and efficiently realize vector-matrix multiplication (VMM), the dominant computational kernel in DNNs. However, a key challenge with resistive crossbars is that they suffer from a range of device and circuit level non-idealities such as interconnect parasitics, peripheral circuits, sneak paths, and process variations. These non-idealities can lead to errors in VMMs, eventually degrading the DNN's accuracy. It is therefore critical to study the impact of crossbar non-idealities on the accuracy of large-scale DNNs. However, this is challenging because existing device and circuit models are too slow to use in application-level evaluations. We present RxNN, a fast and accurate simulation framework to evaluate large-scale DNNs on resistive crossbar systems. RxNN splits and maps the computations involved in each DNN layer into crossbar operations, and evaluates them using a Fast Crossbar Model (FCM) that accurately captures the errors arising due to crossbar non-idealities while being four-to-five orders of magnitude faster than circuit simulation. FCM models a crossbar-based VMM operation using three stages - non-linear models for the input and output peripheral circuits (DACs and ADCs), and an equivalent non-ideal conductance matrix for the core crossbar array. We implement RxNN by extending the Caffe machine learning framework and use it to evaluate a suite of six large-scale DNNs developed for the ImageNet Challenge. Our experiments reveal that resistive crossbar non-idealities can lead to significant accuracy degradations (9.6%-32%) for these large-scale DNNs. To the best of our knowledge, this work is the first quantitative evaluation of the accuracy of large-scale DNNs on resistive crossbar based hardware.