NVM Die Bit-Flip Object Insertion for In-Memory Computing
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing technologies face challenges in efficiently inserting images of objects into other images stored within a non-volatile memory (NVM) array, particularly in the context of in-memory computing for computer vision applications.
Innovation Solution
The proposed solution involves a bit-flip object insertion technique, where an image of an object is inserted into a target image by inverting at least one bit within each of the identified pixels of the target image, using processing circuitry on an NVM die to perform the bit-flipping operation.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Loss of time
If traditional image processing methods are used with separate processors, then image manipulation accuracy is maintained, but data transfer time and system complexity increase
Solution Approach 1:
The patent combines the NVM array and processing circuitry into a single integrated device, eliminating the need for separate processors and reducing data transfer time. The processing circuitry is formed directly on the die containing the NVM array, enabling in-memory computing operations.
Solution Approach 2:
The NVM array serves multiple functions: it acts as both the storage medium for image data and the processing unit for bit-flip operations. This multi-functionality eliminates the need for dedicated separate processors while maintaining computational capabilities.
2Productivity
If bit-flip insertion method is used, then processing speed is improved, but image manipulation precision may be affected
Solution Approach 1:
The patent uses bit-flip operations to create copies of object images and insert them into background images. The bit-flip method efficiently creates and manipulates image copies at the bit level, enabling high-speed processing while maintaining the integrity of the original and copied image data.
Solution Approach 2:
The patent segments the image manipulation process into distinct operations: obtaining the background image, obtaining the object image, identifying target pixels, and performing bit-flip insertion. This segmentation allows each operation to be optimized independently, with bit-flip operations providing high-speed processing for the insertion step.
3Quantity of substance
If in-memory computing is implemented, then data transfer requirements are reduced, but device functionality complexity increases
Solution Approach 1:
The patent merges storage and processing functions into a single NVM die, enabling data to be processed in-place without transfer to separate processors. This eliminates the need for extensive data transfer while the integrated processing circuitry handles image manipulation operations.
Solution Approach 2:
The NVM array performs processing operations on its own stored data through integrated processing circuitry, eliminating the need for external processors. The device serves itself by conducting bit-flip operations and image manipulation directly within the memory structure.
Data Source
AI summary
Bit-flip object insertion techniques are provided for use with a non-volatile memory (NVM) wherein an object is inserted into a background image by flipping or inverting one or more bits within the pixels of the background image that correspond to the shape and insertion location of an object being inserted. In an illustrative example, pixels within the background image that correspond to the shape and insertion location of the object are XORed with binary 1s. This flips the bits of those pixels to change the color (hue) and/or intensity (brightness) of the pixels so the object appears in the background image. In other examples, only the most significant bits of pixels in the background image are inverted (flipped). Exemplary latch-based procedures are described herein for high-speed processing on an NVM die. Multiple plane NVM die implementations are also described for massive processing.