Programmable Threshold Logic Implementations in a Memristor Crossbar Array.

Area of Science:

• Materials Science
• Electrical Engineering
• Computer Engineering

Background:

• Memristor crossbar arrays offer potential for high-density, low-power computing.
• Implementing complex logic functions directly on-chip is crucial for advancing neuromorphic and in-memory computing.

Purpose of the Study:

• To demonstrate the implementation of programmable threshold logic functions using a memristor crossbar array.
• To verify the accurate programming characteristics and logic operation capabilities of the memristor array.

Main Methods:

• Utilized a 32x32 memristor crossbar array with forming-free characteristics achieved through annealing.
• Implemented 3-input and 4-input Boolean logic functions by simultaneous subtraction of weighted sums and threshold values.
• Verified full-adder circuit fidelity and implemented a 4-bit ripple carry adder using read-based logic operations.

Main Results:

• Accurate 256-level programming characteristics were achieved and presented via grayscale images.
• Boolean logic functions were implemented without additional reference bias.
• A 4-bit ripple carry adder demonstrated reliable operation through read-based parallel logic, outperforming stateful logic in reliability and steps.

Conclusions:

• Memristor crossbar arrays are viable for implementing complex digital logic functions.
• Read-based parallel logic operations on memristor crossbars offer advantages in reliability and efficiency over stateful approaches.
• The study validates the potential of memristor technology for next-generation computing architectures.