Display Card GPU Texture Calculation Offloading
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
As data processing demands increase, existing systems rely heavily on high-efficiency CPUs, but the powerful calculation capabilities of display cards, equipped with GPUs, are often underutilized during idle times, leading to inefficient CPU usage.
Innovation Solution
A method is introduced to transform CPU-executed calculation tasks into GPU-executable data and programs, allowing the display card to handle system tasks, thereby reducing CPU usage by converting input data into texture data, performing texture calculations, and outputting results to system memory.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Productivity
If CPU is used to execute all calculation tasks, then calculation completeness is ensured, but CPU usage rate becomes too high and system bottleneck occurs
Solution Approach 1:
The patent divides calculation tasks into two segments: those executed by CPU and those executed by GPU. The system selectively transfers calculation tasks to the GPU based on CPU usage rate monitoring, segmenting the computational workload to balance CPU utilization and maintain calculation completion reliability.
Solution Approach 2:
The display card's GPU is made multi-functional by enabling it to execute not only traditional display-related calculations but also general system calculation tasks. This universality allows the GPU to serve dual purposes: display processing and computational offloading, thereby reducing CPU usage rate while maintaining system functionality.
2Productivity
If GPU of display card is used for calculation tasks, then CPU usage rate is reduced, but system complexity increases due to data transformation requirements
Solution Approach 1:
The patent introduces texture data as an intermediary format between CPU input data and GPU calculation requirements. The CPU transforms input data into texture data format, which serves as the intermediary representation that the GPU can process. This intermediary layer simplifies the interaction between CPU and GPU by providing a standardized data format for transfer and processing.
Solution Approach 2:
The system changes data parameters by transforming input data into texture data with specific format requirements (such as pixel format, data type, and memory arrangement). This parameter transformation enables the GPU to process data efficiently while the transformation overhead is managed through optimized conversion routines, balancing complexity reduction with processing efficiency.
3Speed
If all data processing is done by CPU, then processing control is simple, but processing speed becomes insufficient for large data volumes
Solution Approach 1:
The patent adds a new dimension to the processing architecture by introducing GPU parallel processing capability. Instead of relying solely on sequential CPU processing, the system leverages the GPU's parallel architecture to handle large volumes of data simultaneously. This dimensional shift from sequential to parallel processing dramatically increases data processing speed while the control mechanism remains managed through standardized API calls.
Data Source
AI summary
A method for assisting in data calculation by using a display card: In the present method, input data stored in a system memory is transformed into texture data, which is then stored in a display memory of the display card. Then, a Graphic processing unit (GPU) of the display card is used for executing a texture calculation to the texture data, and a result of the texture calculation is stored in a display target of the display memory. Finally, the display target is outputted to the system memory as the output data. Accordingly, a part of calculation tasks of a central processing unit (CPU) can be given to the GPU of the display card when the CPU is in a high usage rate, so as to reduce a calculation burden of the CPU.