Image Rendering Command Caching for GPU Load Reduction
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
The load on graphics processing units (GPUs) is high due to the need to generate drawing targets for image rendering, leading to increased power consumption and heat generation in terminals like computers and mobile phones.
Innovation Solution
The method involves comparing drawing commands in a to-be-rendered image frame with those of a previously rendered frame to identify reusable drawing targets, allowing the GPU to utilize pre-generated targets instead of recreating them, thereby reducing the load and power consumption.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If the GPU generates a rendering target based on each drawing command in the image command stream, then the image rendering function is implemented, but the load of the GPU becomes heavy and power consumption increases
Solution Approach 1:
The patent applies preliminary action by pre-generating rendering targets for drawing commands that are likely to be reused in future frames. The system analyzes historical drawing command streams to identify commands that have been executed before and pre-computes their rendering targets, so that when these commands appear again, the pre-generated targets can be directly reused without requiring real-time GPU computation, thereby reducing power consumption while maintaining rendering functionality
Solution Approach 2:
The patent changes the parameter of drawing command representation by introducing a simplified command structure that includes identification information and parameter information. This allows the system to efficiently compare and identify identical or similar drawing commands across different frames, enabling the reuse of rendering targets. The parameter change transforms the raw drawing commands into a format that facilitates caching and reuse decisions
2Reliability
If the GPU generates a rendering target based on each drawing command, then complete rendering is achieved, but the load of the GPU becomes heavy
Solution Approach 1:
The patent applies discarding and recovering by identifying rendering targets that can be discarded (not regenerated) when their corresponding drawing commands are detected in subsequent frames. The system recovers and reuses these previously generated rendering targets, thereby reducing the computational load on the GPU while ensuring that complete rendering is achieved through selective reuse of valid targets
Solution Approach 2:
The patent applies copying by creating and storing copies of rendering targets in a cache memory when drawing commands are executed. These copied targets are then retrieved and reused when identical drawing commands are detected, eliminating the need for the GPU to regenerate them. This copying mechanism reduces GPU load while maintaining rendering completeness through accurate target reuse
Data Source
AI summary
The present disclosure relates to image rendering methods apparatuses, and electronic devices. One example method includes obtaining, by a central processing unit (CPU), an image command stream used to render an image frame, comparing the image command stream with a previous image command stream, and, when a same drawing command exists in the two image command streams, instructing a graphics processing unit (GPU) to render the image frame based on a previously generated drawing target.