Game Performance Prediction via Collaborative Filtering

Resolve Bottlenecks,
Find Innovative SolutionsGenerate Solutions

Solution Overview

Problem

Gaming applications face performance variability across different computing devices due to differences in CPU, GPU, and memory performance, making it challenging to maintain consistent frame rates and rendering quality, and developers struggle to test and optimize their applications on the vast array of devices without excessive storage requirements.

Innovation Solution

A computing system uses collaborative filtering to predict the performance of a gaming application on a given device by identifying similar applications and devices, allowing for adaptive adjustment of fidelity parameters to optimize performance without extensive testing or storage needs.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If gaming applications are developed to execute on a variety of computing devices within a device ecosystem, then the application can be deployed across multiple devices, but the performance varies due to differences in device configurations

Engineering Contradiction:

Improvedeployment across devicesVSAvoidperformance consistency

Core Design Contradiction:

Adaptability or versatilityVSReliability

Solution Approach 1:

The system performs preliminary actions by pre-calculating and storing performance predictions for gaming applications on different device configurations before actual execution. The machine learning model proactively determines predicted frame rates and performance metrics based on device characteristics and application requirements, allowing developers to optimize applications in advance without needing to test on every specific device.

Inventive Principle:

Principle #10Preliminary action

Solution Approach 2:

The patent introduces a machine learning-based performance prediction system as an intermediary between the gaming application and the diverse device ecosystem. This intermediary component receives application requirements and device specifications, processes them through the trained model, and outputs optimized performance parameters, thereby mediating the performance variability across different devices.

Inventive Principle:

Principle #24Intermediary (Mediator)

2Manufacturing precision

If developers test and optimize applications on a vast array of devices, then performance can be optimized for each device, but the testing process becomes time-consuming and storage requirements increase

Engineering Contradiction:

Improvedevice-specific optimizationVSAvoidtesting time

Core Design Contradiction:

Manufacturing precisionVSLoss of time

Solution Approach 1:

Instead of physically testing applications on every possible device configuration, the system creates a virtual copy of device performance characteristics through the machine learning model. The model stores and processes performance data from representative devices, creating a digital replica that can predict performance for any device configuration without requiring actual hardware testing for each scenario.

Inventive Principle:

Principle #26Copying

Solution Approach 2:

The system changes the approach from testing physical device configurations to analyzing parameter variations through the machine learning model. By inputting different device parameters (CPU performance, GPU performance, display characteristics) into the trained model, the system can predict performance outcomes without physically testing each configuration, thereby reducing testing time while maintaining optimization precision.

Inventive Principle:

Principle #35Parameter changes

3Reliability

If developers store device-specific settings for all possible devices, then performance can be optimized for each device, but storage requirements become excessive

Engineering Contradiction:

Improvedevice-specific performanceVSAvoidstorage space

Core Design Contradiction:

ReliabilityVSQuantity of substance

Solution Approach 1:

The patent extracts the essential performance optimization data from the vast amount of possible device-specific settings by using the machine learning model to identify and store only the critical performance parameters and relationships. The model extracts key patterns from device characteristics and application requirements, storing condensed performance predictions rather than complete device-specific configuration data for all possible devices.

Inventive Principle:

Principle #2Taking out (Extraction)

Solution Approach 2:

The machine learning model serves as a universal performance prediction system that can handle multiple device types and configurations through a single integrated model. Rather than storing separate device-specific settings for each device type, the universal model processes various device parameters and application requirements to provide optimized performance predictions across the entire device ecosystem, reducing storage requirements while maintaining device-specific optimization.

Inventive Principle:

Principle #6Universality (Multi-functionality)

Data Source

PatentUS20240152440A1Game performance prediction across a device ecosystem

Publication Date: 2024.05.09 GOOGLE LLC

AI summary

A computing system may receive an indication of a gaming application and an indication of a computing device. The computing system may determine, using collaborative filtering of performance scores of a plurality of gaming applications executing at a plurality of computing devices, a predicted performance of the gaming application executing at the computing device. The computing system may send an indication of the predicted performance of the gaming application executing at the computing device.