Configurable GPU With Dual I/O Interfaces For Mixed Signaling

Solution Overview

Problem

The transition to differential signaling in graphics processing units (GPUs) is hindered by the need for backward compatibility with single-ended frame buffer memory, requiring either the development of separate GPUs with differential I/O interfaces or increasing the size and complexity of existing single-ended I/O GPU designs to accommodate differential I/O interfaces, both of which are costly.

Innovation Solution

A GPU with both differential and single-ended I/O interfaces that allows for high-bandwidth differential communication with future devices while maintaining compatibility with single-ended frame buffer memory, implemented in a single integrated circuit device without increasing size, complexity, or cost, by using a dual configuration that includes a PCI-Express interface for differential communication and single-ended signaling for legacy compatibility.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Speed

If a separate GPU with differential I/O interface is developed, then differential communication performance is improved, but manufacturing cost and complexity increase

Engineering Contradiction:

Improvedifferential communication speedVSAvoidGPU design complexity

Core Design Contradiction:

SpeedVSDevice complexity

Solution Approach 1:

The GPU integrates both differential and single-ended I/O interfaces within a single device, allowing it to communicate with both differential and single-ended frame buffer memory. This multi-functional design eliminates the need for separate GPUs for different interface types, reducing manufacturing complexity while maintaining high-speed differential communication capability through the differential I/O interface.

Inventive Principle:

Principle #6Universality (Multi-functionality)

2Adaptability or versatility

If the size of existing single-ended I/O GPU designs is increased to accommodate differential I/O interface, then backward compatibility is maintained, but device size and complexity increase

Engineering Contradiction:

Improveinterface compatibilityVSAvoidGPU size

Core Design Contradiction:

Adaptability or versatilityVSVolume of moving object

Solution Approach 1:

The patent merges both differential and single-ended I/O interfaces into a single integrated GPU design. By combining these interface types in one device, the solution maintains backward compatibility with single-ended frame buffer memory while enabling forward compatibility with differential interfaces, without requiring separate devices or increasing overall device size.

Inventive Principle:

Principle #5Merging (Combining)

Solution Approach 2:

The GPU is designed with universal functionality to support multiple interface types (both differential and single-ended) within a single device. This allows the same GPU to communicate with legacy single-ended frame buffer memory while also supporting future differential frame buffer memory, maintaining adaptability without increasing device volume.

Inventive Principle:

Principle #6Universality (Multi-functionality)

3Adaptability or versatility

If multiple device designs are created for different interface types, then market coverage is improved, but production cost increases

Engineering Contradiction:

Improvemarket coverageVSAvoidproduction cost

Core Design Contradiction:

Adaptability or versatilityVSEase of manufacture

Solution Approach 1:

The GPU integrates both differential and single-ended I/O interfaces within a single device, allowing it to communicate with both differential and single-ended frame buffer memory. This multi-functional design eliminates the need for separate GPUs for different interface types, reducing manufacturing complexity while maintaining high-speed differential communication capability through the differential I/O interface.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Solution Approach 2:

The patent merges both differential and single-ended I/O interfaces into a single integrated GPU design. By combining these interface types in one device, the solution maintains backward compatibility with single-ended frame buffer memory while enabling forward compatibility with differential interfaces, without requiring separate devices or increasing overall device size.

Inventive Principle:

Principle #5Merging (Combining)

Data Source

PatentUS8412872B1Configurable GPU and method for graphics processing using a configurable GPU

Publication Date: 2013.04.02 NVIDIA CORP

AI summary

The present invention pertains to a graphics processing unit. The graphics processing unit includes a graphics processing core configured for graphics processing. A single-ended I/O interface configured to implement single-ended communication with a frame buffer memory is included in the graphics processing unit. The graphics processing unit further includes a differential I/O interface having a first portion and a second portion. In a first configuration, the first portion and the second portion implement a PCI-Express interface with a computer system. In a second configuration, the first portion implements a PCI-Express interface with the computer system and the second portion implements differential communication with a coupled device.