Resistive Memory Crossbar Circuits for Low-Power Dot-Product Computing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional microprocessor technology is limited by its chronological operation execution, leading to inefficiencies in computation power, size, and power consumption, making it unsuitable for applications requiring high computational power like image recognition, and large-scale neuromorphic computing networks are physically large and power-intensive, limiting their implementation in industries such as biomedical, military, and mobile devices.
Innovation Solution
Analog neuromorphic circuits utilizing resistive memories with variable resistance values and a controller to perform parallel multiplication and addition operations, enabling efficient computation with minimal power consumption and compact size by implementing resistive memories in a crossbar configuration.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If conventional microprocessor technology is used, then operations can be executed in chronological order, but computation power, size, and power consumption are limited
Solution Approach 1:
The patent replaces conventional microprocessor architecture with analog neuromorphic circuits that use resistive memory crossbars to perform matrix-vector multiplication. This substitution enables parallel computation through analog signal processing, achieving significantly higher computational power while consuming much less power compared to digital microprocessors.
Solution Approach 2:
The patent changes the fundamental operating parameter from sequential digital operations to parallel analog operations. By using continuous voltage signals representing vector elements and exploiting the ohmic law (I=V/R), the system performs computations in parallel across multiple resistive memory cells, transforming the computation paradigm to achieve both high productivity and low power consumption.
2Productivity
If large scale computer clusters are used to implement conventional neuromorphic computing networks, then computing power increases, but physical space and power requirements become significant
Solution Approach 1:
The patent merges multiple computational functions into a single integrated circuit. By implementing matrix-vector multiplication operations within a compact crossbar architecture using resistive memory cells, the system combines what would traditionally require large-scale computer clusters into a single chip, dramatically reducing physical space while maintaining high computing power.
Solution Approach 2:
The patent transitions from three-dimensional physical stacking of computing components to two-dimensional planar integration. The crossbar architecture lays out resistive memory cells in a grid pattern, enabling efficient use of space and allowing for compact scaling that would be impossible with conventional vertical stacking approaches.
3Productivity
If conventional neuromorphic computing networks are implemented, then multiple operations can be executed simultaneously, but space and power requirements severely limit implementation
Solution Approach 1:
The resistive memory cells inherently perform the multiplication operation through their physical property of resistance. When voltage signals are applied, the current flowing through each cell automatically equals V/R, performing the multiplication without requiring additional active circuitry. This self-service characteristic eliminates the need for power-intensive digital logic, achieving high computational efficiency with minimal power consumption.
4Productivity
If resistive memories with variable resistance values are used, then parallel multiplication and addition operations can be performed, but device complexity increases
Solution Approach 1:
The patent extracts the computational function from complex digital logic and concentrates it into the simple resistive memory cells. By removing the need for digital signal processing, control logic, and data buses, the system achieves parallel computation using only passive resistive elements, significantly reducing overall device complexity despite the variable resistance requirement.
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
The solution provides significantly more computational power with minimal power and space requirements, allowing for applications like image recognition and learning algorithms, suitable for industries with limited space and power resources.
Implementation Method 1
Each resistive memory is configured to provide a resistance value to each corresponding input voltage. Each resistance value is a positive resistance value selected from a finite range of resistance values.
Data Source
AI summary
An analog neuromorphic circuit is disclosed having resistive memories that provide a resistance to each corresponding input voltage signal. Input voltages are applied to the analog neuromorphic circuit. Each input voltage represents a vector value that is a non-binary value included in a vector that is incorporated into a dot-product operation with weighted matrix values included in a weighted matrix. A controller pairs each resistive memory with another resistive memory. The controller converts each pair of resistance values to a single non-binary value. Each single non-binary value is mapped to a weighted matrix value included in the weighted matrix that is incorporated into the dot-product operation with the vector values included in the vector. The controller generates dot-product operation values from the dot-product operation with the vector and the weighted matrix where each dot-product operation is a non-binary value.