Fragmented GPU Cores in Smart Displays for Deep Learning

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional display technologies lack the ability to manage power efficiency and perform intelligent processing, as they simply present input information without any thought or management, leading to inefficient energy usage and limited functionality.

Innovation Solution

Integration of fragmented graphic cores and smart pixels within the display panel, utilizing micro-LEDs and embedded memory, allows for local processing and analytics, enabling features like object identification through deep learning and image enhancement, reducing the need for external processing and enhancing energy efficiency.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If display devices perform only basic presentation functions without intelligent processing, then device complexity is low, but power efficiency and functionality are limited

Engineering Contradiction:
Improveintelligent processing capabilityVSAvoiddisplay device complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The patent divides the display device into multiple functional segments: traditional display elements for presentation and embedded fragmented graphic cores for intelligent processing. Each segment performs specialized functions, allowing the display to gain advanced capabilities while maintaining the simplicity of individual components. The fragmented graphic cores are distributed throughout the display structure, enabling localized processing without requiring a centralized complex processing unit.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

The display device is designed to perform multiple functions: traditional display presentation and intelligent image processing. The fragmented graphic cores enable the display to act as both a presentation device and a processing unit, eliminating the need for separate external processing equipment. This multi-functionality allows the display to adapt to various applications including image enhancement, object identification, and real-time analytics.

Inventive Principle:
Principle #6Universality (Multi-functionality)

2Use of energy by moving object

If display devices integrate intelligent processing capabilities, then power efficiency improves, but device complexity increases

Engineering Contradiction:
Improvepower efficiencyVSAvoiddisplay device complexity
Core Design Contradiction:
Use of energy by moving objectVSDevice complexity

Solution Approach 1:

The patent implements local processing capabilities by embedding fragmented graphic cores at specific locations within the display structure. Rather than adding a centralized complex processing unit that would consume significant power, the display performs intelligent processing locally at the pixel or pixel-block level. This local quality approach enables power-efficient processing by handling tasks close to where the data resides, reducing energy consumption for data transmission and processing.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The patent transitions from traditional two-dimensional display presentation to three-dimensional functional integration by embedding processing cores within the display structure. This dimensional change allows the display to perform intelligent processing in addition to presentation, creating a multi-functional device that improves power efficiency without requiring separate external processing systems.

Inventive Principle:
Principle #17Another dimension (Dimensionality change)

3Power

If external processing is used for image analysis, then processing power is sufficient, but signal transmission requirements increase energy consumption

Engineering Contradiction:
Improveprocessing powerVSAvoidenergy consumption
Core Design Contradiction:
PowerVSUse of energy by moving object

Solution Approach 1:

The patent extracts the essential processing functionality from external systems and embeds it directly within the display device through fragmented graphic cores. By taking out only the necessary processing capabilities needed for image analysis and embedding them locally, the system maintains sufficient processing power while eliminating the need for high-energy signal transmissions to external processors. This extraction approach keeps the most critical processing functions within the display structure.

Inventive Principle:
Principle #2Taking out (Extraction)

Solution Approach 2:

The fragmented graphic cores act as intermediaries between the display elements and external processing systems. Instead of directly transmitting raw pixel data to external processors, the embedded cores perform preliminary intelligent processing locally, reducing the data volume and complexity of signals that need to be transmitted externally. This intermediary function significantly reduces energy consumption while maintaining adequate processing power for complex image analysis tasks.

Inventive Principle:
Principle #24Intermediary (Mediator)

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

This solution enables displays to perform intelligent processing and analytics, such as object identification and image enhancement, while reducing energy consumption by minimizing signal transmissions and processing power, thereby making displays smarter and more efficient.

Implementation Method 1

Intensifying the light from the light emitting diodes (LEDs) to enhance the image

Methodology Applied
Scientific EffectLight emitting diode: Light Emitting Diode

Data Source

PatentUS11010861B2Fragmented graphic cores for deep learning using LED displays
Publication Date: 2021.05.18 INTEL CORP
  • US11010861B2 patent drawing
  • US11010861B2 patent drawing
  • US11010861B2 patent drawing

AI summary

A smart display including one or more groups of smart pixels and at least one graphics engine. The at least one graphics engine is fragmented into GPU (graphics processing unit) minute cores. The GPU minute cores are distributed throughout the smart display. The smart pixels with distributed graphics within the smart display perform deep learning. Libraries stored on GPU minute core embedded memory are used to perform object identification using deep learning. The smart display monitors for pixel degradation and, if necessary, performs pixel enhancement.