Abstract:This work reports a prototypical implementation of in‐memory analog mathematical operations based on spin‐orbit torque devices. Several orders smaller area overhead and comparable power consumption are obtained compared to the complementary metal oxide semiconductor counterparts, but the data can be locally stored in the configuration. Edge detection and amplitude modulation are experimentally demonstrated while pattern recognition is simulated. Analog arithmetic operations are the most fundamental mathematical operations used in image and signal processing as well as artificial intelligence (AI). In‐memory computing (IMC) offers a high performance and energy‐efficient computing paradigm. To date, in‐memory analog arithmetic operations with emerging nonvolatile devices are usually implemented using discrete components, which limits the scalability and blocks large scale integration. Here, a prototypical implementation of in‐memory analog arithmetic operations (summation, subtraction and multiplication) is experimentally demonstrated, based on in‐memory electrical current sensing units using spin‐orbit torque (SOT) devices. The proposed structures for analog arithmetic operations are smaller than the state‐of‐the‐art complementary metal oxide semiconductor (CMOS) counterparts by several orders of magnitude. Moreover, data to be processed and computing results can be locally stored, or the analog computing can be done in the nonvolatile SOT devices, which are exploited to experimentally implement the image edge detection and signal amplitude modulation with a simple structure. Furthermore, an artificial neural network (ANN) with SOT devices based synapses is constructed to realize pattern recognition with high accuracy of ≈95%.

Parallel Computing in Memory Paradigm Based on Reconfigurable Spin-Orbit Torque.

PRISM: Energy-Efficient Polymorphic Operation Based on Spin-Orbit Torque Memory for Reconfigurable Computing

Reconfigurable Processing in Memory Architecture Based on Spin Orbit Torque

Parallel-Prefix Adder in Spin-Orbit Torque Magnetic RAM for High Bit-Width Non-Volatile Computation

Reconfigurable spintronic logic gate utilizing precessional magnetization switching

IMCS2: Novel Device-to-Architecture Co-Design for Low-Power In-Memory Computing Platform Using Coterminous Spin Switch

A Novel Computing-in-Memory Platform Based on Hybrid Spintronic/CMOS Memory

Spin Orbit Torques for Ultra-Low Power Computing.

Electrically Reconfigurable 3D Spin‐Orbitronics

Field-free spin–orbit devices via heavy-metal alloy with opposite spin Hall angles for in-memory computing

In-Memory Computing Paradigms by Exploiting the Intrinsic Interactions Between STT and SOT

High performance and energy-efficient in-memory computing architecture based on SOT-MRAM

Low Power In-Memory Computing Based On Dual-Mode Sot-Mram

Computing-in-Memory Paradigm Based on STT-MRAM with Synergetic Read/Write-Like Modes

An Energy-Efficient Computing-in-Memory (cim) Scheme Using Field-Free Spin-Orbit Torque (SOT) Magnetic RAMs

Spin-Transfer Versus Spin-Orbit Torque Mram

An STT-MRAM Based Reconfigurable Computing-in-memory Architecture for General Purpose Computing

Cascadable In-Memory Computing Based on Symmetric Writing and Read Out

Implementation of Highly Reliable and Energy‐Efficient Nonvolatile In‐Memory Computing Using Multistate Domain Wall Spin–Orbit Torque Device

In‐Memory Mathematical Operations with Spin‐Orbit Torque Devices

HXNOR-PBNN: A Scalable and Parallel Spintronics Synaptic Architecture for Probabilistic Binary Neural Networks