Fast and Low-Cost Mitigation of ReRAM Variability for Deep Learning Applications

Sugil Lee, Mohammed Fouda, Jongeun Lee, Ahmed Eltawil, Fadi Kurdahi

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Scopus citations

Abstract

To overcome the programming variability (PV) of ReRAM crossbar arrays (RCAs), the most common method is program-verify, which, however, has high energy and latency overhead. In this paper we propose a very fast and low-cost method to mitigate the effect of PV and other variability for RCA-based DNN (Deep Neural Network) accelerators. Leveraging the statistical properties of DNN output, our method called Online Batch-Norm Correction (OBNC) can compensate for the effect of programming and other variability on RCA output without using on-chip training or an iterative procedure, and is thus very fast. Also our method does not require a nonideality model or a training dataset, hence very easy to apply. Our experimental results using ternary neural networks with binary and 4-bit activations demonstrate that our OBNC can recover the baseline performance in many variability settings and that our method outperforms a previously known method (VCAM) by large margins when input distribution is asymmetric or activation is multi-bit.
Original languageEnglish (US)
Title of host publication2021 IEEE 39th International Conference on Computer Design (ICCD)
PublisherIEEE
Pages269-276
Number of pages8
ISBN (Print)9781665432191
DOIs
StatePublished - Oct 2021

Fingerprint

Dive into the research topics of 'Fast and Low-Cost Mitigation of ReRAM Variability for Deep Learning Applications'. Together they form a unique fingerprint.

Cite this