English

A Single-Cycle MLP Classifier Using Analog MRAM-based Neurons and Synapses

Emerging Technologies 2020-12-07 v1 Hardware Architecture Machine Learning

Abstract

In this paper, spin-orbit torque (SOT) magnetoresistive random-access memory (MRAM) devices are leveraged to realize sigmoidal neurons and binarized synapses for a single-cycle analog in-memory computing (IMC) architecture. First, an analog SOT-MRAM-based neuron bitcell is proposed which achieves a 12x reduction in power-area-product compared to the previous most power- and area-efficient analog sigmoidal neuron design. Next, proposed neuron and synapse bit cells are used within memory subarrays to form an analog IMC-based multilayer perceptron (MLP) architecture for the MNIST pattern recognition application. The architecture-level results exhibit that our analog IMC architecture achieves at least two and four orders of magnitude performance improvement compared to a mixed-signal analog/digital IMC architecture and a digital GPU implementation, respectively while realizing a comparable classification accuracy.

Keywords

Cite

@article{arxiv.2012.02695,
  title  = {A Single-Cycle MLP Classifier Using Analog MRAM-based Neurons and Synapses},
  author = {Ramtin Zand},
  journal= {arXiv preprint arXiv:2012.02695},
  year   = {2020}
}

Comments

arXiv admin note: substantial text overlap with arXiv:2006.01238

R2 v1 2026-06-23T20:44:16.139Z