NVIDIA GPU Setup for Deep Learning (Windows)

Reference video: https://www.youtube.com/watch?v=nATRPPZ5dGE

1. Install the NVIDIA GPU Driver

Install the latest driver for your GPU: https://www.nvidia.com/Download/index.aspx

2. Install Visual Studio with C++ Tools

Install Visual Studio Community and include the C++ workload: https://visualstudio.microsoft.com/vs/community/

During installation, select:

• Desktop development with C++

3. Install Anaconda or Miniconda

Use Anaconda/Miniconda to manage your Python environment:

• Anaconda: https://www.anaconda.com/download
• Miniconda: https://docs.conda.io/en/latest/miniconda.html

4. Install CUDA Toolkit

Download and install CUDA Toolkit: https://developer.nvidia.com/cuda-downloads

5. Install cuDNN

Download cuDNN (NVIDIA Developer account required): https://developer.nvidia.com/cudnn

6. Install PyTorch

Use the official installer selector and pick the version that matches your CUDA setup: https://pytorch.org/get-started/locally/

Example (CUDA build):

pip install torch torchvision torchaudio --index-url https://download.pytorch.org/whl/cu121

7. Verify GPU and CUDA in This Project

Run the provided scripts:

python torch_device_check.py
python cuda_version_check.py

Expected checks:

• GPU count and GPU name
• Active device (cuda if available)
• CUDA version used by PyTorch
• cuDNN version used by PyTorch