Multi-GPU System Segmentation for Rendering Quality

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Current graphics chip technologies face challenges in achieving high-quality rendering while ensuring compatibility with older-generation programming languages and various video output formats, requiring advanced hardware and software integration with flexible adaptability.

Innovation Solution

A multi-graphics processor system comprising a first and second graphics processor, where data processed by the first processor is sent to the second for further processing, returned, and output as a video signal through a single channel, with the first processor handling security and format conversions, and the second processor integrated with high-speed memory access.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If multiple graphics processors are used to enhance rendering quality and performance, then rendering quality and performance are improved, but system configuration complexity increases

Engineering Contradiction:
Improverendering quality and performanceVSAvoidsystem configuration complexity
Core Design Contradiction:
ProductivityVSDevice complexity

Solution Approach 1:

The rendering workload is segmented between two graphics processors: the first GPU handles security processing and format conversion, while the second GPU performs high-speed rendering operations. This segmentation allows each processor to specialize in specific tasks, improving overall rendering performance while maintaining a manageable system configuration through clear functional division.

Inventive Principle:
Principle #1Segmentation

2Manufacturing precision

If data is processed through multiple graphics processors, then rendering quality is enhanced, but data processing time may increase

Engineering Contradiction:
Improverendering qualityVSAvoiddata processing time
Core Design Contradiction:
Manufacturing precisionVSLoss of time

Solution Approach 1:

Security processing and format conversion are performed as preliminary actions by the first graphics processor before the main rendering operation. This preliminary processing prepares the data in advance, allowing the second GPU to immediately process the pre-processed data without waiting, thereby reducing overall processing time while maintaining high rendering quality.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The system maintains continuous useful action by having the first GPU prepare data while the second GPU processes rendering operations simultaneously. The data flow between the two processors is optimized to ensure continuous processing without idle time, maintaining high rendering quality while minimizing processing delays.

Inventive Principle:
Principle #20Continuity of useful action

3Device complexity

If a single video output channel is used, then system configuration is simplified, but video output flexibility is reduced

Engineering Contradiction:
Improvesystem configuration simplicityVSAvoidvideo output flexibility
Core Design Contradiction:
Device complexityVSAdaptability or versatility

Solution Approach 1:

The first graphics processor is designed with multi-functionality, handling both security processing and format conversion tasks. This universal processor can adapt to different video output requirements by changing the format conversion parameters, providing video output flexibility through a single channel while keeping the system configuration simple.

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

4Speed

If the second graphics processor is integrated with video memory, then memory access speed is improved, but chip complexity increases

Engineering Contradiction:
Improvememory access speedVSAvoidchip complexity
Core Design Contradiction:
SpeedVSDevice complexity

Solution Approach 1:

The second graphics processor integrates video memory directly onto the same chip, merging storage and processing functions. This integration eliminates external memory interface complexities and achieves high-speed memory access through short internal connections, while the increased chip complexity is offset by the elimination of separate memory components and their interconnections.

Inventive Principle:
Principle #5Merging (Combining)

Data Source

PatentEP1746538B1Multi-graphics processor system, graphics processor and rendering method
Publication Date: 2019.06.26 SONY INTERACTIVE ENTERTAINMENT LLC
  • EP1746538B1 patent drawingFigure 1
  • EP1746538B1 patent drawingFigure 2
  • EP1746538B1 patent drawingFigure 3

AI summary

A first GPU is provided with a digital video output terminal (134) (Vout terminal) for connection to an external source. A digital video signal output from the Vout terminal is provided to a display device via a HDMI. The first GPU (100) and a second GPU (200) are connected to each other via a data bus (140) for bidirectional data exchange. The second GPU (200) applies a predetermined rendering process on data provided from the first GPU (100) via a data input and output interface. The rendered data is returned to the first GPU (100) via the data input and output interface. The first GPU (100) processes the data returned from the second GPU (200) as necessary and outputs a digital video signal via the Vout terminal and the HDMI.