External Graphics System GPU Load Redistribution
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Existing Thunderbolt graphics card external enclosures for notebook computers waste the built-in graphics processing unit's computing performance by automatically ignoring the video stream from the internal GPU due to power saving and bandwidth limitations, only utilizing the external GPU for graphics processing.
Innovation Solution
An external graphics system and method that utilizes a Thunderbolt interface to enable joint graphics processing between a notebook computer's built-in GPU and an external Thunderbolt graphics card enclosure, with a software-driven load redistribution mechanism to optimize graphics performance by determining when to engage each GPU based on a load-carrying capacity threshold.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Power
If an external Thunderbolt graphics card enclosure is used to replace the built-in GPU, then graphics computing performance is improved, but the built-in GPU's computing performance is wasted
Solution Approach 1:
The patent merges the built-in GPU and external Thunderbolt GPU into a collaborative system where both GPUs work simultaneously on graphics processing tasks. The host device receives rendering requests and distributes them to both the built-in GPU and external GPU, combining their outputs to achieve enhanced overall graphics performance while utilizing the computing capabilities of both devices.
Solution Approach 2:
The system enables the built-in GPU to serve dual purposes: it continues to handle graphics rendering tasks alongside the external GPU, making the system more versatile. The graphics processing system can dynamically switch between using only the external GPU, only the built-in GPU, or both together depending on the workload requirements.
2Power
If the external GPU is used exclusively, then maximum graphics performance is achieved, but power consumption and interface bandwidth are overloaded
Solution Approach 1:
The patent segments the graphics processing workload between the built-in GPU and external GPU based on a load-carrying capacity threshold. Rendering requests are divided into first request commands (handled by built-in GPU) and second request commands (handled by external GPU), allowing the system to distribute computational load and prevent overloading the external GPU and Thunderbolt interface.
Solution Approach 2:
The system dynamically adjusts the distribution of graphics processing tasks based on real-time conditions. The host device determines whether to use the built-in GPU, external GPU, or both together based on the current workload and load-carrying capacity threshold, creating a flexible and adaptive graphics processing system that optimizes power consumption and bandwidth usage.
Data Source
AI summary
An external graphics method for an external graphics system is provided. The external graphics system includes a host device at least having a first graphics processing unit that generates a first video stream signal, and an external graphics device for connecting to the host device. The external graphics device at least includes a second graphics processing unit that generates a second video stream signal, and a TBT interface unit electrically connected to the host device. The method mainly includes: when the host device determining a graphics processing performed by the second graphics processing unit not exceeding a load-carrying capacity threshold, displaying an image frame corresponding to a TBT signal; and when determining the graphics processing performed by the second graphics processing unit exceeding the load-carrying capacity threshold, computing the TBT signal and the first video stream signal in sum, and displaying the corresponding image frame accordingly.