Project G-Assist: an AI assistant that uses voice and text to optimize computer performance

General Introduction

Project G-Assist is an AI assistant from NVIDIA designed for GeForce RTX users. It helps users optimize PC performance, adjust game settings, and monitor hardware status through voice or text commands. The tool runs locally on the RTX GPU, requires no Internet connection, is responsive, and is free. Using the Small Language Model (SLM), it understands simple commands and performs tasks such as boosting frame rates, lowering power consumption, or setting peripheral lights.G-Assist is currently in an experimental version for desktop RTX users, with support for laptops coming in the future, and NVIDIA expects to continue to improve the tool through user feedback to make PC operation easier.

Function List

• Optimize PC performance: Improve running speed or save power by adjusting GPU settings with a single click.
• Adjust game settings: automatically configure the optimal game screen and frame rate according to your hardware.
• Real-time hardware diagnostics: Monitor frame rate, latency, GPU utilization, and temperature to provide recommendations for improvement.
• Chart performance data: Generate and export visual charts of FPS, power consumption and other data.
• Control peripherals: Support for adjusting lights and fan speeds on Logitech G, Corsair, and more.
• Answer hardware questions: Provide information about RTX GPUs and NVIDIA software.
• Support for plugin extensions: Users can create plugins for customized features via GitHub.

Using Help

Installation process

1. Download NVIDIA App
   Open your browser and visit the NVIDIA website to download the latest version of the NVIDIA App. Make sure your computer is equipped with a GeForce RTX GPU (minimum 12GB of video memory).
2. Installation of applications
   Double-click the downloaded installation file and follow the prompts to complete the installation. After installation, launch the NVIDIA App.
3. Enable G-Assist
   On the NVIDIA App main screen, go to the "Discover" tab and locate Project G-Assist, click "Install" and wait for the download to complete. Once installed, G-Assist will be automatically integrated into the app.
4. activation tool
   Open the NVIDIA App and press Alt+G shortcut, or click the "Overlay" button in the interface to enable G-Assist.

How to use the main features

Optimize computer performance

• procedure
  1. After launching G-Assist, say or type "Optimize my computer's performance".
  2. G-Assist scans your GPU, CPU, and monitor configurations.
  3. After a few seconds, it will suggest tweaking options, such as enabling one-click overclocking or reducing power consumption.
  4. After confirming, click the "Apply" button and the settings will take effect immediately.
• caveat
  If you want to adjust it manually, you can say "Show current GPU settings" and G-Assist will list the details for you to modify.

Adjusting game settings

• procedure
  1. With the game open, launch G-Assist.
  2. Type or say "Optimize the settings of this game".
  3. G-Assist will recommend the best resolution, image quality and frame rate settings for your hardware.
  4. After selecting "Apply automatically" or confirming manually, the game will reload the new configuration.
• Featured Functions
  It recognizes the contents of the game window and provides targeted optimization recommendations. In Cyberpunk 2077, for example, it might suggest turning on DLSS and adjusting lighting and shadow details.

Real-time hardware diagnostics

• procedure
  1. Say "Check my hardware status" in the G-Assist interface.
  2. It displays FPS, GPU utilization, temperature and other data in real time.
  3. If a bottleneck is detected (e.g. high temperatures), it will suggest lowering the image quality or cleaning the fan.
• Export data
  Enter "Generate Performance Chart" and G-Assist will create a chart with frame rate and latency and save it as a local file.

Control Peripherals

• procedure
  1. Ensure that a peripheral, such as a Logitech G keyboard, is connected and supports NVIDIA control.
  2. Say "Adjust keyboard lights to blue" or "Increase fan speed" to G-Assist.
  3. It will execute commands directly and give feedback on the results.
• Supported Devices
  Some models of Logitech G, Corsair, MSI and Nanoleaf are currently supported, see the official website for a list.

Creating custom plug-ins

• procedure
  1. Visit the NVIDIA GitHub repository to download the G-Assist plug-in development documentation.
  2. utilization ChatGPT or other tools to generate JSON and Python files.
  3. Place the file in the plugin directory of G-Assist (see documentation for path).
  4. Restart the NVIDIA App and the customization will be available.
• typical example
  For example, when you create a Twitch plugin, you can say "Check if an anchor is online" and G-Assist will return real-time information.

Handling Precautions

• G-Assist is an experimental version, there may be errors in recognizing commands, it is recommended to operate in clear language.
• It is not suitable for complex conversations and only handles hardware and game-related tasks.
• Make sure the RTX driver is up to date before use, it can be updated in the NVIDIA App.

application scenario

1. Gamer Optimization Experience
   You're playing Eldenfalcon and you feel that the frame rate is unstable. Launch G-Assist, say "Increase Game Framerate" and it will automatically adjust the settings to make the game smoother.
2. Computer hardware monitoring
   You want to know how the newly installed RTX 4080 is performing. Ask G-Assist to "Show GPU Usage" and it will give you real-time feedback and generate graphs.
3. Simplify peripheral management
   You have multiple RGB devices and you want to standardize your light colors. Say "set all lights to red" to G-Assist and the adjustment is done in seconds.
4. Developer Extensions
   You're a programmer and you want G-Assist to check the status of a live stream. After creating the plugin via GitHub, it will tell you directly if the Twitch host is online or not.

QA

1. Does G-Assist require an internet connection?
   Not required. It runs on the local RTX GPU and is available completely offline.
2. What GPUs are supported?
   Currently supports RTX 30, 40, and 50 series desktop GPUs with at least 12GB of RAM; laptop support to be updated.
3. Can you optimize all games?
   Can be optimized for most games, but results vary by game and hardware.
4. How do I give feedback on a problem?
   Click "Send Feedback" in the upper right corner of the NVIDIA App and select "Project G-Assist" to submit a suggestion or question.