Magnetoresistive Circuits and Systems: Embedded Non-Volatile Memory to Crossbar Arrays

Abstract

This overview article describes Magnetoresistive Random Access Memory (MRAM) from a circuits and systems perspective. We discuss various tradeoffs and design challenges of MRAM in three broad application areas: 1) embedded nonvolatile memory (eNVMs), 2) crossbar-based analog in-memory computing, and 3) stochastic computing. Certain MRAM characteristics, such as high retention, high endurance and fast read and write operations, make them ideal for replacing the standard CMOS memories for last-level cache applications with future scaling. However, various tradeoffs in power, performance and area pose conflicting requirements on MRAM design. We explore these challenges and various circuit techniques that have been developed to mitigate them. Further, we present various requirements of memristive crossbar arrays for accelerating matrixvector-multiplication (MVM) operations in light of MRAM devices, and highlight various challenges, design considerations, and applicability of MRAM as crossbar arrays. Finally, we will elaborate on how inherent stochasticity of MRAM devices can be leveraged for implementing energy-efficient true random number generators (TRNGs) and stochastic units for performing certain tasks, such as developing fast solvers for combinatorial optimization, and stochastic neural networks.

Magnetic tunnel junction random number generators applied to dynamically tuned probability trees driven by spin orbit torque

“…In this mode of operation, the user is 'agnostic' as to where the number comes from; they do not know and do not care ahead of time which MTJ device will generate the number. This operation strategy may be preferrable as while an MTJ can theoretically have unlimited endurance [27,28], in practice MTJ devices are expected to have a cycle life of ~10 15 cycles before degradation [29], and thus distributing the work can let devices last longer. Further, each device can generate numbers in parallel, instead of needing to sequentially be sampled millions or billions of times.…”

DHMM-Based Asynchronous Finite-Time Sliding Mode Control for Markovian Jumping Lur’e Systems

“…In this case, the hidden Markov model (HMM) is introduced to describe such phenomenon. As a statistical analysis model, HMM has became a hot topic over the last few years, and its application including but not limited to speech processing [15], target tracking [16], digital communication [17], biomedical engineering [18] and finance [19]. It's worth mentioning that the aforementioned HMM includes two random processes, i.e., the random process caused by the Markov chain, and the other process brought from the observed variables.…”

On the Design of 7/9-Rate Sparse Code for Spin-Torque Transfer Magnetic Random Access Memory

“…With fascinating features such as high density, low power consumption, and suitable portable device, the NAND flash memory is dominant in the current NVM market [3], [4]. Emerging memory technologies such as magnetic random access memory (MRAM), resistive random access memory (ReRAM) are expected to replace the conventional memories for the consumer as well as data center applications [5]. Spintronics is also one of the most exciting topics in today's nanotechnology.…”

“…Because of its main characteristics of nonvolatility, nanosecond write/read speeds with low latency, endurance, and reliability, STT-MRAM is very useful in many applications. Significantly, the STT-MRAM is expected to replace both static RAM (SRAM) and dynamic RAM (DRAM) that are commonly used for the cache memories and CPU main memory, respectively [5]. Moreover, the MRAM technology provides lower power consumption, higher write/read speeds, better reliability, and longer endurance than Flash memory technology.…”