CPU+GPU Stress in Room Temperature

Overview

1. Please ensure the system fan is 100%

2. Please ensure the GPU fan is 100%

3. gpu_burn is a easy accessible tool to for thermal testing

4. superposition is a good tool for testing transient power consumption

---

Installation

# GPU Driver
## Check the Recommended GPU Driver based on current installed GPU cards
sudo ubuntu-drivers devices
## Install the recommended GPU Driver (which might be different due to GPU generation)
sudo apt-get install nvidia-driver-575


# Add the necessary permissions to allow changes to the GPU fan speed.
cd /etc/X11/
sudo gedit Xwrapper.config
## Then modify the file, make sure the following line is added before allowed_users=console:
needs_root_rights=yes
## To restart the Desktop service 
sudo systemctl restart gdm

## Quick test commands 
## sudo nvidia-smi 
## sudo nvidia-settings

## Install CUDA (not necessary for gpu burn)
# wget https://developer.download.nvidia.com/compute/cuda/repos/ubuntu2404/x86_64/cuda-keyring_1.1-1_all.deb
# sudo dpkg -i cuda-keyring_1.1-1_all.deb
# sudo apt update
# sudo apt install -y cuda

# Install CUDA tool kit (for gpu_burn)
sudo apt install nvidia-cuda-toolkit

# Install gpu burn
cd ~/Desktop
sudo apt-get install git
git clone https://github.com/wilicc/gpu-burn.git
cd gpu-burn
make

# (TBC) Make system fan controllable
sudo apt install lm-sensors
sudo sensors-detect
sudo apt install psensor

# Reboot the whole machine if needed
sudo reboot

Run (gpu-burn)

# (TBC) Make system fan 100%

# Change GPU Fan Speed (GUI)
sudo nvidia-settings
## (GUI) Traverse to the fan control page to manually change the GPU fan speed

# Change GPU Fan Speed (Command Line)
## (TBD) There should be a command line version

# Run gpu_burn
cd ~/Desktop/gpu-burn
## Stress GPU 0 for 3600 sec
sudo ./gpu-burn -i 0 3600
## Stress GPU 1 for 3600 sec
sudo ./gpu-burn -i 1 3600

# Check GPU Temperature for every second
watch -n 1 nvidia-smi