🌍 Grid-X: Decentralized Compute Network

Grid-X is an open-source "Airbnb for Computing Power." It allows users to securely rent out their idle GPU/CPU resources to others for tasks like AI training, rendering, or heavy computations.

The system uses Docker Containers for isolation, Ngrok for tunneling (no port forwarding required), and a central Flask Registry to manage the marketplace.

---

📂 Repository Structure

This repository is divided into three main components:

Folder	Role	Description
GridX_Host	Provider	The software for people sharing their computer. Runs a secure Docker container and connects to the registry.
GridX_Registry	Server	The central marketplace. Tracks active nodes and provides the Web Dashboard.
GridX_User	Client	Tools for renters to connect (CLI Client + SSH scripts).

---

🚀 Getting Started

To run a complete network, you need at least one Registry (Server) and one Host (Provider).

Prerequisites

• Python 3.x
• Docker Desktop (Required for Hosts)
• Ngrok Account (Free tier works, but requires unique tokens for multiple nodes)

---

1️⃣ Setup the Registry (The Marketplace)

Run this first to create the network.

1. Navigate to the GridX_Registry folder.
2. Windows: Double-click Start_Registry.bat.
3. Linux/Mac: Run python3 registry.py.
4. Copy the URL: The terminal will display a public URL (e.g., http://5a2b.ngrok-free.app).
   • ⚠️ IMPORTANT: You must share this URL with your Hosts and Users.

---

2️⃣ Setup a Host (The Provider)

Run this on the machine with the powerful GPU/CPU.

1. Navigate to the GridX_Host folder.
2. Configuration:
   • Open gridx_service.py in a text editor.
   • Find REGISTRY_URL = "http://127.0.0.1:5000".
   • Replace it with the Public URL from the Registry (Step 1).
3. Launch:
   • Windows: Double-click Start_Host.bat.
   • Linux: Run ./Start_Host.sh (ensure permissions with chmod +x).
4. Verify:
   • A green icon will appear in your System Tray.
   • The node should now be visible on the Registry Dashboard.

Features:

• Auto-GPU Detection: Passes NVIDIA GPUs to the container automatically.
• Busy Status: Hides connection details when a user is logged in.
• Self-Healing: Automatically restarts the container if it crashes.

---

3️⃣ Connect as a User (The Renter)

Access the compute power and transfer files.

Method A: Web Dashboard (Easiest)

1. Open the Registry URL in your browser.

2. Find a node with status READY 🟢.

3. Connect (SSH):

   • Click [📋 COPY SSH].
   • Paste the command into your terminal:

     ssh renter@0.tcp.in.ngrok.io -p 12345

   • Enter the Password displayed on the dashboard card.

4. Transfer Files (SCP):

   • Click [📂 COPY SCP] (The orange button).
   • It copies a template command like:

     scp -P 12345 "YOUR_FILE" renter@0.tcp.in.ngrok.io:/home/renter/

   • Paste it into your terminal, replace "YOUR_FILE" with your actual file path (drag and drop the file into the terminal), and hit Enter.

Method B: CLI Tool (Advanced)

1. Navigate to GridX_User.
2. Update tenant.py with the REGISTRY_URL.
3. Run Start_Client.bat (Windows) or ./Start_Client.sh (Linux).
4. Use the arrow keys to select a node and connect instantly.

---

🛡️ Security Features

Grid-X is built with safety in mind for the Host:

1. Docker Isolation: Users are trapped inside a container. They cannot access your host files (C:\ or /home).
2. Time Bomb Containers: Every session has a 24-hour lease. The container self-destructs afterwards to wipe data.
3. One-Time Passwords (OTP): Every session generates a new random password.
4. Graceful Exit: If the Host quits the app, the node is immediately removed from the marketplace.

---

🧪 Testing GPU Capabilities

To verify that the Host's GPU is accessible inside the container:

1. Connect via SSH.
2. Upload the gpu_benchmark.py script (found in GridX_User).
3. Run it:

   python3 gpu_benchmark.py

4. It will install PyTorch (if missing) and run a matrix multiplication stress test.

---

⚠️ Known Limitations (Free Tier)

• Ngrok Limits: Free Ngrok accounts allow only 1 active tunnel. You cannot run the Registry AND a Host on the same computer using the same Ngrok Auth Token simultaneously.
• Persistence: Files created inside the container are lost when the session ends (Container is ephemeral).

---