Web Platform Rendering Reducing GPU Memory Usage
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Web-platform operations on electronic devices, such as smartphones and smart TVs, face performance deterioration due to high GPU memory usage and limited hardware structure, especially when executing web applications with direct rendering techniques, leading to increased CPU and GPU memory occupation.
Innovation Solution
A method and apparatus for processing web-platform operations by parsing and compressing the root layer of a web application's tree structure, storing it as a compressed file, and rendering the graphic layer in real time, using techniques like open vector graphics (OpenVG) and scalable vector graphics (SVG) to reduce data residency in GPU memory, thereby optimizing GPU bandwidth and performance.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Power
If direct rendering technique is applied with GPU acceleration, then web content can be displayed with hardware acceleration, but GPU memory usage increases significantly occupying about 8-16 MB of memory
Solution Approach 1:
The patent segments the web content rendering into two distinct parts: the root layer (background) and graphic layers (content). By separating these layers and applying different processing methods to each, the system can optimize memory usage while maintaining GPU acceleration benefits for the graphic layers.
Solution Approach 2:
The patent extracts the root layer from the GPU memory management system entirely. Instead of rendering the root layer through GPU acceleration, it uses CPU-based rendering and stores the result in a separate bitmap, removing this large memory consumer (8-16 MB) from the GPU memory allocation.
2Stability of the object's composition
If root layer is rendered through GPU acceleration, then consistent rendering pipeline is maintained, but memory bandwidth is consumed and performance deteriorates
Solution Approach 1:
The rendering pipeline is segmented into two independent paths: CPU-based rendering for the root layer and GPU-based rendering for graphic layers. This segmentation allows each path to be optimized independently, maintaining pipeline consistency where needed while improving overall productivity.
Solution Approach 2:
The patent introduces an intermediary mechanism (separate bitmap storage and composite rendering approach) between the root layer rendering and the final display. This intermediary allows the root layer to be rendered separately without consuming GPU memory bandwidth, while still being integrated into the final rendered output.
3Manufacturing precision
If texture buffer is used for root layer rendering, then direct rendering is achieved, but memory occupation increases to about 16 MB when including CPU memory
Solution Approach 1:
The patent extracts the root layer from the texture buffer system entirely. By rendering the root layer to a separate bitmap in CPU memory and then composing it with the GPU-rendered graphic layers, the system eliminates the need to allocate texture buffer space for the root layer, reducing total memory occupation from 16 MB to approximately 8 MB or less.
Data Source
AI summary
An electronic apparatus and an operating method of a web-platform, the method including: executing a web application in the electronic apparatus; parsing a root layer corresponding to a most significant node having a tree structure for a visual configuration of a web page displayed by the web application; processing the root layer to have less or a reduced amount of data resident in a graphic memory as the web application is executed; and rendering a graphic layer, except the root layer, from the nodes forming the tree structure in real time. Thus, the root layer is processed so that less or reduced data can be resident in the GPU memory, thereby improving the graphic processing performance of the GPU with even a low bandwidth.