Multi-GPU Console Framebuffer Layout for Uniform Memory Access

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Existing game console SoC designs face challenges in scaling performance and memory access, particularly in high-end systems with multiple GPUs, leading to issues like non-uniform memory access (NUMA) and thread management, framebuffer management, and HDMI output control.

Innovation Solution

Implementing a system with at least two GPUs, where each GPU renders different frames or portions of frames, and combines them for output, with options for shared or separate memory controllers and using scanout units or multiplexers to manage video data, allowing for efficient video rendering and HDMI output.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Productivity

If multiple GPUs are used to provide high-end performance, then processing capability is improved, but memory access uniformity deteriorates due to NUMA issues

Engineering Contradiction:

Improveprocessing capabilityVSAvoidmemory access uniformity

Core Design Contradiction:

ProductivityVSReliability

Solution Approach 1:

The system divides the video rendering workload into separate portions, with each GPU rendering different frames or different portions of frames. This segmentation allows each GPU to operate independently with its own memory controller, avoiding NUMA issues while maintaining high processing capability through parallel operation of multiple GPUs.

Inventive Principle:

Principle #1Segmentation

2Productivity

If multiple GPUs render different portions of video, then rendering efficiency is improved, but system complexity increases due to framebuffer management

Engineering Contradiction:

Improverendering efficiencyVSAvoidframebuffer management

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The framebuffer is segmented into separate portions, with each GPU responsible for rendering specific portions to dedicated buffers. This segmentation simplifies management by allowing each GPU to operate with its own memory resources without complex coordination, while still achieving efficient parallel rendering of the complete video output.

Inventive Principle:

Principle #1Segmentation

3Productivity

If multiple GPUs are implemented, then video rendering performance is improved, but HDMI output control becomes more complex

Engineering Contradiction:

Improvevideo rendering performanceVSAvoidHDMI output control

Core Design Contradiction:

ProductivityVSDevice complexity

Solution Approach 1:

The HDMI output is segmented into separate channels, with each GPU connected to its own HDMI output. This allows independent control of each GPU's video output without complex multiplexing or switching logic, simplifying the overall control architecture while maintaining high rendering performance through parallel processing.

Inventive Principle:

Principle #1Segmentation

Data Source

PatentEP3918567B1Scalable game console CPU / GPU design for home console and cloud gaming

Publication Date: 2026.04.15 SONY INTERACTIVE ENTERTAINMENT LLC

Figure 1
Figure 2
Figure 3~5

AI summary

In a multi-GPU simulation environment, frame buffer management may be implemented by multiple GPUs (306, 402, 504, 600, 704) rendering respective frames of video, or by rendering respective portions of each frame of video (900/902; 1000/1002; 1100/1102). One of the GPUs controls HDMI frame output by virtue of receiving frame information from the other GPU(s) and reading out complete frames through a physically connected HDMI output port (1200). Or, the outputs of the GPUs can be multiplexed together (1302).