#7079 CUDA GPU acceleration unavailable. Verify NVIDIA GPU & CUDA installation (Help -> CUDA Help).

This error indicates the software cannot detect a valid CUDA environment for GPU acceleration. Since you are running a source deployment (frozen:False), the issue is almost certainly one of the following: NVIDIA driver incompatibility, missing/incomplete CUDA Toolkit installation, missing cuDNN, or incorrect system environment variables.

Prerequisites for CUDA acceleration:

• NVIDIA GPU only (AMD/Intel GPUs are NOT supported for CUDA)
• CUDA Toolkit ≥ 12.8
• cuDNN ≥ 9.11

Diagnosis steps:

1. Open Command Prompt (CMD) and run:

   nvcc -V

   This shows the CUDA compiler version. If it returns "not recognized", CUDA Toolkit is not installed or not in PATH.

2. Run:

   nvidia-smi

   This shows your GPU model, driver version, and the maximum CUDA version supported by the driver. The driver must support CUDA 12.8+.

3. Verify your system PATH environment variables contain both:

   • C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.8\bin
   • C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.8\lib

   And that cuDNN files (cudnn*.dll) are placed in the bin directory of the CUDA installation.

Common fixes:

If nvcc -V is not found:

1. Download and install CUDA Toolkit 12.8+ from NVIDIA.
2. Download cuDNN 9.11+ matching your CUDA version (requires an NVIDIA developer account).
3. Extract cuDNN and copy the contents into the CUDA installation directory.

If nvidia-smi shows a driver version too old:
Update your NVIDIA driver to the latest version from the NVIDIA website.

If both tools work but the software still reports CUDA unavailable:

1. Make sure the environment variables are set system-wide, then restart the software (and ideally the computer).
2. Check that you are not using an MSVC build of Python that conflicts with CUDA libraries — use the official Python 3.10 from python.org.

Important note for source deployment:

• The pyvideotrans source code requires the CUDA toolkit and cuDNN to be properly accessible to the running Python process. If you installed CUDA to a custom path, ensure that path is in your system PATH, not just the user PATH.
• Some older driver versions may not support CUDA 12.8. If your GPU is older, you may need to fall back to CPU processing temporarily.

Reference documents:

• Windows CUDA & cuDNN configuration
• Windows CUDA 12.6 & cuDNN 9.8 installation guide