Memcapacitive Cross-Bar Array for Dot Product Computation
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing systems for determining a dot product are inefficient in terms of power consumption and suffer from sneak path leakage, and they lack flexible output measurement resolution.
Innovation Solution
A memcapacitive cross-bar array that utilizes memcapacitors to perform dot product calculations by applying programming and vector signals, eliminating sneak path leakage and enabling low power consumption, with output measurement resolution tunable through integration of current over time.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of energy
If existing systems are used to determine a dot product, then the calculation can be performed, but power consumption is high and sneak path leakage occurs
Solution Approach 1:
The patent replaces conventional electronic circuit systems with a memcapacitive crossbar array system. Memcapacitors are used as the fundamental computing element instead of traditional transistors and capacitors, enabling direct analog computation of dot products through physical charge accumulation. This substitution eliminates the need for complex control circuits and reduces sneak path leakage inherent in conventional systems.
Solution Approach 2:
The patent introduces charge as an intermediary physical quantity to perform computation. Instead of using voltage or current directly for computation, the system uses charge accumulation on memcapacitor arrays to represent and compute dot products. This intermediary charge representation enables precise measurement while eliminating sneak path effects that plague conventional resistive or transistor-based systems.
2Measurement precision
If existing systems are used to determine a dot product, then the calculation can be performed, but output measurement resolution is fixed and inflexible
Solution Approach 1:
The patent implements dynamic measurement resolution by allowing the system to integrate current over variable time periods. The measurement resolution is not fixed but can be adjusted by changing the integration time, enabling the same hardware to provide different levels of precision based on application requirements. This dynamic capability is achieved through the charge accumulation nature of memcapacitors.
Solution Approach 2:
The patent changes the measurement parameter from fixed voltage levels to integrable current over time. By measuring the accumulated charge (integral of current) rather than instantaneous voltage, the system achieves flexible resolution. The measurement resolution can be adjusted by varying the integration time constant, providing adaptability without requiring multiple fixed-resolution channels.
3Productivity
If memristor systems are used instead of memcapacitors, then dot product calculation is possible, but manufacturing costs are higher
Solution Approach 1:
The patent employs memcapacitors which can be manufactured using standard semiconductor fabrication processes, making them cheaper and more scalable than memristors. Memcapacitors can be implemented as simple capacitor structures with memory effects, requiring fewer material layers and simpler fabrication steps compared to the complex oxide stack structures needed for memristors. This enables lower manufacturing costs while maintaining dot product calculation functionality.
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 memcapacitive cross-bar array achieves efficient dot product calculations with reduced power consumption and eliminates sneak path leakage, offering flexible output measurement resolution and potentially lower manufacturing costs compared to memristor systems.
Implementation Method 1
capacitive memory devices coupled between the row lines and the column lines at the junctions. The capacitive memory devices receive a number of ground signals. The ground signals clear all charges from the capacitive memory devices.
Data Source
AI summary
A method of obtaining a dot product includes applying a programming signal to a number of capacitive memory devices coupled at a number of junctions formed between a number of row lines and a number of column lines. The programming signal defines a number of values within a matrix. The method further includes applying a vector signal. The vector signal defines a number of vector values to be applied to the capacitive memory devices.