GPUs

A real-time web interface for monitoring NVIDIA GPUs.

pip install gpus
gpus start

Features

• Real-time monitoring of NVIDIA GPU statistics
• Clean, modern, responsive web interface
• Historical utilization graphs
• Process monitoring
• Command-line interface
• Background server mode

Requirements

• Python 3.6+
• NVIDIA GPU with installed drivers
• NVIDIA Management Library (NVML)

Installation

pip install gpus

Usage

Starting the Web Interface

# Start the web interface in the foreground (default port 5000)
gpus

# Specify a different port
gpus --port 8080
# or with the short option
gpus -P 8080

# Specify update interval (in seconds)
gpus --update-interval 2.0
# or with the short option
gpus -U 2.0

# Specify history length (in seconds)
gpus --history-length 600
# or with the short option
gpus -L 600

# Specify history resolution (in seconds)
gpus --history-resolution 1.0
# or with the short option
gpus -R 1.0

Then open your web browser and navigate to http://localhost:5000 (or the port you specified).

Managing Background Server

You can run the server in the background and manage it with subcommands:

# Start the server in the background
gpus start

# Check if the server is running
gpus status

# Stop the background server
gpus stop

The background server uses the same command-line options as the foreground server:

# Start the background server on a specific port
gpus -P 8080 start

Using as a Python Package

from gpus.app import GPUMonitorApp

# Create the application
app = GPUMonitorApp(
    update_interval=2.0,
    history_length=300,
    history_resolution=1.0
)

# Run the application
app.run(host='0.0.0.0', port=5000)

Cloud

GPUs Cloud is a free cloud service that allows you to easily manage your GPUs on-the-go. Remotely monitor your GPUs from anywhere & get notified when your training runs fail (coming soon).

Warning

GPUs Cloud is currently in a limited beta stage. If you want early access (highly unstable), please DM me on X for an invite code. Expect some bugs, limited features, and high downtime.

Getting Started

Sign up for a free account (using Hugging Face login) on the GPUs Cloud website. Enter your invite code and add a device. Then run the following command to login:

gpus cloud login

To start sending GPU metrics to the cloud (in the background), run the following command:

gpus cloud start

To stop sending GPU metrics to the cloud, run the following command:

gpus cloud stop

To check the status of the cloud client, run the following command:

gpus cloud status

To send metrics to the cloud (in the foreground), run the following command:

gpus cloud

To logout of the cloud, run the following command:

gpus cloud logout

Development

Setup Development Environment

# Clone the repository
git clone https://github.com/fakerybakery/gpus
cd gpus

# Create a virtual environment
python -m venv venv
source venv/bin/activate  # On Windows: venv\Scripts\activate

# Install in development mode
pip install -e .

Debugging

If you encounter issues, please check the logs in the ~/.gpus directory. If the problem persists, please create an issue on GitHub.

Note that GPUs has not been tested on Windows. Use at your own risk.

License

BSD-3-Clause