Mixed-Signal Analog MAC Circuit With Coarse-Fine A/D Reconstruction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing digital processing units for matrix multiplication in machine learning algorithms and edge computing devices face inefficiencies in power consumption, particularly when performing analog-to-digital conversions.
Innovation Solution
A mixed-signal circuit system that splits integer values into coarse and fine values, performs analog multiply and accumulate operations, and then combines the analog output signals through analog-to-digital conversion to produce a reconstructed digital output signal, while optionally employing MSB skipping and LSB truncation for improved power efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Use of energy by moving object
If digital processing units are used for matrix multiplication, then computational accuracy is maintained, but power consumption increases
Solution Approach 1:
The patent segments the integer values into coarse values and fine values, processes them through separate analog MAC units, and combines the results. This segmentation allows the system to use power-efficient analog processing while maintaining accuracy through the combination of coarse and fine value results.
Solution Approach 2:
The patent introduces an intermediate representation where integer values are split into coarse and fine components. These intermediate values are processed analogously and then recombined, serving as a mediator between digital input and digital output while enabling power-efficient analog processing.
2Measurement precision
If analog MAC operations are performed on full precision values, then computational accuracy is maintained, but power efficiency is lost during A/D conversion
Solution Approach 1:
By segmenting the precision requirements into coarse and fine value components, the patent enables analog processing at reduced precision levels. The coarse values capture the majority of the computational result with fewer bits, while fine values provide the necessary precision correction, allowing efficient analog processing without full precision A/D conversion.
Solution Approach 2:
The patent applies partial action by processing only the necessary precision levels in analog domain. The coarse values are processed analogously to capture the bulk of the result, and fine values are used to correct precision, avoiding the need for full precision analog processing and subsequent full precision A/D conversion.
3Use of energy by moving object
If MSB skipping is performed during A/D conversion, then power efficiency increases, but measurement precision decreases
Solution Approach 1:
The patent segments the precision requirements so that coarse values (processed analogously) cover the most significant bits, reducing the precision burden on A/D conversion. This segmentation allows MSB skipping in the A/D converter while the coarse analog processing preserves the critical high-order information.
Solution Approach 2:
The patent performs preliminary analog processing of coarse values before A/D conversion. This preliminary action captures the most significant computational results in analog form, allowing the A/D converter to skip MSBs while still maintaining overall computational accuracy through the combination with fine value results.
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 system achieves increased power efficiency by preserving the efficiency gains of analog multiply and accumulate operations during analog-to-digital conversion, while maintaining accurate digital output signals, particularly beneficial for edge computing devices.
Implementation Method 1
MAC units based on switched capacitors offer greater power efficiency than digital processing units
Data Source
AI summary
A first circuit is configured to split a first integer value into a first coarse value and a first fine value, and split a second integer value into a second coarse value and a second fine value. A second circuit performs an analog multiply and accumulate (MAC) operation on the first and second coarse values to produce a first analog output, perform an analog MAC operation on the first coarse value and the second fine value to produce a second analog output, perform an analog MAC operation on the first fine value and the second coarse value to produce a third analog output, and perform an analog MAC operation on the first and second fine values together to produce a fourth analog output. A third circuit is configured to perform analog-to-digital (A/D) conversion on and combine the analog output signals to produce a reconstructed digital output signal.