GPU Fan Control

A Bash script that automatically controls NVIDIA GPU fan speeds based on temperature thresholds. The script monitors GPU temperatures and adjusts fan speeds accordingly to maintain optimal cooling while minimizing noise. Supports multi-fan GPUs.

Features

• Automatic GPU-to-fan mapping using nvidia-settings verbose output
• Temperature-based fan curve with configurable thresholds
• Support for multiple GPUs and their associated fans
• Runs continuously in the background with 5-second intervals
• Minimal resource usage

Requirements

• NVIDIA GPU with proprietary drivers installed
• nvidia-settings utility (part of the NVIDIA driver package)
• Linux system with systemd (Ubuntu 20.04+, Kubuntu 24.04+, etc.)
• Root/sudo privileges for fan control

Installation

Step 1: Clone the Repository

git clone https://github.com/avtc/gpu-fan-control.git
cd gpu-fan-control

Step 2: Make the Script Executable

chmod +x gpu-fan-control.sh

Step 3: Create a Systemd Service

The most robust way to run the fan control script permanently is to create a systemd service. This ensures the script starts automatically on boot and restarts if it fails.

1. Create the service file:

sudo nano /etc/systemd/system/gpu-fan-control.service

2. Paste the following content into the file:

[Unit]
Description=NVIDIA GPU Fan Control Script
After=network.target multi-user.target graphical.target nvidia-persistenced.service

[Service]
Type=simple
ExecStart=/path/to/your/gpu-fan-control/gpu-fan-control.sh
Restart=on-failure
RestartSec=10
User=root
Group=root
StandardOutput=journal
StandardError=journal

[Install]
WantedBy=multi-user.target

Important: Replace /path/to/your/gpu-fan-control/gpu-fan-control.sh with the actual absolute path where you cloned the repository (e.g., /home/username/gpu-fan-control/gpu-fan-control.sh).

3. Save the file: Press Ctrl+O, then Enter, then Ctrl+X to exit nano.

Step 4: Enable and Start the Service

1. Reload the systemd daemon:

sudo systemctl daemon-reload

2. Enable the service to start on boot:

sudo systemctl enable gpu-fan-control.service

3. Start the service immediately:

sudo systemctl start gpu-fan-control.service

Step 5: Verify the Installation

1. Check the service status:

sudo systemctl status gpu-fan-control.service

You should see "Active: active (running)".

2. View the logs:

sudo journalctl -u gpu-fan-control.service -f

3. Test after reboot:

sudo reboot

After rebooting, verify the service is running:

sudo systemctl status gpu-fan-control.service

Setting GPU Power Limits on Startup

For systems with multiple GPUs (like 8x RTX 3090), you may want to set power limits automatically on system startup to ensure consistent performance and thermal management. This can be achieved using a systemd service.

Example: 8x RTX 3090 Power Limit Service

1. Create the power limit service file:

sudo nano /etc/systemd/system/gpu-power-limit.service

2. Paste the following content into the file:

[Unit]
Description=Set NVIDIA GPU Power Limit
After=network.target multi-user.target graphical.target
Wants=nvidia-persistenced.service
# Ensure nvidia-persistenced is started if you use it

[Service]
Type=oneshot
RemainAfterExit=yes
ExecStartPre=/usr/bin/nvidia-smi -pm 1
# Enable persistence mode for all GPUs
ExecStart=/usr/bin/nvidia-smi -i 0 -pl 280
ExecStart=/usr/bin/nvidia-smi -i 1 -pl 280
ExecStart=/usr/bin/nvidia-smi -i 2 -pl 280
ExecStart=/usr/bin/nvidia-smi -i 3 -pl 280
ExecStart=/usr/bin/nvidia-smi -i 4 -pl 280
ExecStart=/usr/bin/nvidia-smi -i 5 -pl 280
ExecStart=/usr/bin/nvidia-smi -i 6 -pl 280
ExecStart=/usr/bin/nvidia-smi -i 7 -pl 280
# Add more ExecStart lines for additional GPUs (e.g., -i 2, -i 3, etc.)

[Install]
WantedBy=multi-user.target

3. Enable and start the service:

sudo systemctl daemon-reload
sudo systemctl enable gpu-power-limit.service
sudo systemctl start gpu-power-limit.service

4. Verify the power limits:

nvidia-smi -q | grep "Power Limit"

Customizing for Your Setup

• Power Limit Value: The example uses 280 watts, which is typical for RTX 3090. Adjust this value based on your GPU model and requirements.
• GPU Count: Modify the number of ExecStart lines to match your GPU count.
• GPU Indices: Ensure the GPU indices (-i 0, -i 1, etc.) match your actual GPU configuration.

Important Notes

• This service sets power limits to 280W for each GPU, which is common for RTX 3090 cards
• The ExecStartPre=/usr/bin/nvidia-smi -pm 1 line enables persistence mode for better performance
• Power limits are applied before the fan control service starts, ensuring optimal thermal management
• Adjust the power limit value according to your specific GPU model and power supply capabilities

Fan Curve

The script uses the following temperature-to-fan-speed mapping:

Temperature Range	Fan Speed
< 35°C	20%
35-39°C	30%
40-49°C	40%
50-59°C	50%
60-64°C	60%
65-69°C	70%
70-79°C	75%
80-89°C	80%
≥ 90°C	90%

Troubleshooting

Common Issues

1. Permission Denied:

   • Ensure the script is executable: chmod +x gpu-fan-control.sh
   • Make sure the service runs as root (configured in the service file)

2. X Authority Error:

   • The script automatically detects the X authority file
   • If you encounter issues, ensure you're running in a graphical session

3. NVIDIA Driver Issues:

   • Verify nvidia-settings is working: nvidia-settings -q gpus
   • Check that the NVIDIA persistence daemon is running: systemctl status nvidia-persistenced.service

4. Service Not Starting:

   • Check the logs: sudo journalctl -u gpu-fan-control.service -b
   • Verify the path in ExecStart is correct
   • Ensure all dependencies are installed

Manual Testing

To test the script manually before setting up the service:

sudo ./gpu-fan-control.sh

Press Ctrl+C to stop the script.

Customization

Modifying the Fan Curve

Edit the temperature thresholds in gpu-fan-control.sh:

# Fan curve
if (( temp < 35 )); then
    speed=20
elif (( temp < 40 )); then
    speed=30
# ... add your custom thresholds here
else
    speed=90
fi

Changing the Update Interval

Modify the sleep duration at the end of the main loop (line 72):

sleep 5  # Change 5 to your preferred interval in seconds

Safety Notes

• This script enables manual fan control, overriding the default NVIDIA driver settings
• Monitor your GPU temperatures initially to ensure the fan curve is appropriate for your hardware
• The script includes safety limits (maximum 90% fan speed) to prevent excessive wear
• Always test in a controlled environment before deploying to production systems

License

This project is licensed under the MIT License - see the LICENSE file for details.

Contributing

1. Fork the repository
2. Create a feature branch (git checkout -b feature/amazing-feature)
3. Commit your changes (git commit -m 'Add some amazing feature')
4. Push to the branch (git push origin feature/amazing-feature)
5. Open a Pull Request

Disclaimer

This software comes with no warranty. Use at your own risk. The authors are not responsible for any damage to your hardware.