In-Memory Implication Gate Using Magnetic Tunnel Junctions

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Current von-Neumann architecture faces bottlenecks in data transfer rate and energy consumption, particularly in data-intensive applications like big-data and AI, due to the decoupling of memory and processor, and existing in-memory technologies suffer from poor throughput, high energy consumption, and limited non-volatility and endurance.

Innovation Solution

The introduction of an in-situ in-memory implication gate using two-terminal magnetic tunnel junction devices with voltage-controlled magnetic anisotropy (VCMA) and precessional dynamics, enabling high-throughput, energy-efficient, and non-volatile operations by selectively switching between parallel and anti-parallel states without the need for high switching currents.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Power

If von-Neumann architecture with decoupled memory and processor is used, then computing power can be improved, but data transfer rate and energy efficiency deteriorate due to the bottleneck between memory and processor

Engineering Contradiction:

Improvecomputing powerVSAvoiddata transfer rate

Core Design Contradiction:

PowerVSProductivity

Solution Approach 1:

The patent merges memory storage and logic computation functions into a single integrated structure. Magnetic tunnel junction devices serve dual purposes as both memory elements and logic gate components, eliminating the need for separate memory and processor units. This integration allows data to be processed in-situ without transfer between decoupled components, directly resolving the von-Neumann bottleneck.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The magnetic tunnel junction devices perform multiple functions: they store data as memory elements and simultaneously execute logic operations as computational units. The same physical structure serves as both the storage medium and the processing element, enabling the system to overcome the limitations of specialized separate components.

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Adaptability or versatility

If STT-MRAM with multi-row activation is used for in-memory computing, then logic operations can be performed, but sense-margin and throughput deteriorate

Engineering Contradiction:

Improvelogic operation capabilityVSAvoidsense-margin

Core Design Contradiction:

Adaptability or versatilityVSMeasurement precision

Solution Approach 1:

The patent implements logic operations at the local level within individual memory cell units rather than requiring multi-row activation. Each memory cell structure contains the necessary components to perform logic operations locally, improving sense-margin by avoiding the signal interference and complexity associated with activating multiple rows simultaneously.

Inventive Principle:

Principle #3Local quality

3Ease of operation

If resistive devices are used for in-situ in-memory operations, then logic operations can be performed, but lifecycle and material endurance deteriorate

Engineering Contradiction:

Improvein-situ operation capabilityVSAvoidlifecycle and endurance

Core Design Contradiction:

Ease of operationVSReliability

Solution Approach 1:

The patent replaces resistive devices with magnetic tunnel junction devices that utilize magnetic field effects rather than resistive switching. This substitution improves reliability and endurance because magnetic states are more stable and less prone to degradation from repeated write operations compared to resistive states, while still enabling in-situ logic operations.

Inventive Principle:

Principle #28Mechanics substitution (Replace mechanical system)

Applied Scientific Principles

This section explains which scientific principles are used to turn an abstract innovation direction into a practical engineering solution.

Function Achieved in This Case

This approach provides improved throughput, energy efficiency, and enhanced non-volatility, overcoming the limitations of existing technologies by allowing for high-speed, low-energy in-memory computations and parallel logical operations, such as NOT and XOR, while maintaining data integrity during power cycles.

Implementation Method 1

two-terminal magnetic tunnel junction devices with voltage-controlled magnetic anisotropy (VCMA) and precessional dynamics

Methodology Applied

Scientific EffectVoltage-controlled magnetic anisotropy (VCMA):

Implementation Method 2

first and second magnetic tunneling junction devices (MTJ) selectively juxtaposed in a series

Methodology Applied

Scientific EffectMagnetic tunneling:

Implementation Method 3

voltage-controlled magnetic anisotropy (VCMA) and precessional dynamics

Methodology Applied

Scientific EffectPrecessional dynamics: Precession

Data Source

PatentUS10802827B2Memory device having in-situ in-memory stateful vector logic operation

Publication Date: 2020.10.13 PURDUE RES FOUND

AI summary

An in-situ in-memory implication gate is disclosed. The gate include a memory cell. The cell includes a first voltage source, a second voltage source lower in value than the first voltage source, a first and second magnetic tunneling junction devices (MTJ) selectively juxtaposed in a series and mirror imaged relationship between the first and second sources, each having a pinned layer (PL) in a first direction and a free layer (FL) having a polarity that can be switched from the first direction in which case the MTJ is in a parallel configuration presenting an electrical resistance to current flow below a first resistance threshold to a second direction in which case the MTJ is in an anti-parallel configuration presenting an electrical resistance to current flow higher than a second resistance threshold, and further each having a non-magnetic layer (NML) separating the PL from the FL.