Crossbar Array Signal Modification for Accurate Dot-Product Computing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Memristor crossbar arrays face limitations such as one polarity input signal and positive matrix value constraints, leading to inaccuracies and inefficiencies in processing digital signals, particularly in handling signals with negative values and large dynamic ranges.
Innovation Solution
An arithmetic logic unit (ALU) is introduced to modify digital input signals by applying functions like shifting, normalization, noise filtering, and thermal compensation, ensuring accurate analog output signals are generated, which are then converted back to digital outputs, thereby overcoming the limitations of the crossbar array.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If a memristor crossbar array is used to process dot-product operations, then computational speed is improved, but accuracy deteriorates due to one polarity input signal limitation and positive matrix value constraints
Solution Approach 1:
The ALU performs preliminary modification of digital input signals before they are converted to analog signals and processed by the crossbar array. This includes applying offset values to handle one-polarity constraints and scaling to manage dynamic range limitations, thereby preparing the signals in advance to overcome the crossbar array's inherent limitations and maintain computational accuracy
Solution Approach 2:
The ALU acts as an intermediary component between the digital input signals and the analog crossbar array processing. It modifies digital signals with appropriate transformations (offset addition, scaling) before analog conversion, and also processes the analog output signals after ADC conversion, serving as a mediator that compensates for the crossbar array's limitations without sacrificing its high-speed computational capability
2Adaptability or versatility
If the crossbar array processes signals with large dynamic ranges, then processing capability is improved, but accuracy deteriorates due to signal limitations
Solution Approach 1:
The ALU dynamically changes signal parameters including applying offset values to shift signal ranges and applying scaling factors to adjust dynamic ranges. These parameter transformations enable the crossbar array to process signals with large dynamic ranges by mapping them into the acceptable input range, thereby maintaining both processing capability and accuracy
3Device complexity
If the crossbar array is used without signal modification, then device complexity is reduced, but accuracy deteriorates due to uncorrected limitations
Solution Approach 1:
The ALU provides self-service functionality by automatically detecting and correcting for crossbar array limitations through built-in compensation mechanisms. It applies predefined correction functions and compensation algorithms to offset errors introduced by the crossbar array's one-polarity constraint and positive matrix value limitation, thereby maintaining accuracy without requiring external correction systems
Data Source
AI summary
According to examples, an apparatus may include an arithmetic logic unit (ALU) to apply a modification function to a digital input signal to generate a modified digital input signal, a digital-to-analog converter (DAC) to convert the modified digital input signal to an analog input signal, a crossbar array to apply an operation on the analog input signal to generate an analog output signal, and an analog-to-digital converter (ADC). The ADC may modify the analog output signal to compensate for application of the modification function to the digital input signal, may convert the modified analog output signal to a digital output signal, and may output the digital output signal.