Code to Test GPU

Run a premade tensorflow 2 estimator to test gpu

Nvidia Driver and Cuda libraries are needed to be pre-installed. If not refer this link.

Use Python version 3 (code tested on 3.7.6)

Make sure current directory contains following files Readme.txt, requirements.txt and run.py

Create and activate a virtual environment

Use pip install package_name, pip3 install package_name or python3 -m pip install package_name whichever suitable for install packages

pip install virtualenv
python -m venv gpu-check-env
source gpu-check-env/bin/activate

Install required packages inside the environment

pip install -r requirements.txt

Run code inside the environment

python run.py --batch-size 256 --train-steps 5000 --device gpu

Change values of batch size and train steps as per requirement

To check GPU utilization, on another terminal

watch -n 1 nvidia-smi

Leave/Deactivate the virtual enviroment

deactivate