#8663 CUDA GPU acceleration unavailable. Verify NVIDIA GPU & CUDA installation (Help -> CUDA Help).

This error means the software's CUDA device detection failed. It is usually caused by one of three things: no NVIDIA discrete GPU, an outdated driver, or CUDA 12.8/cuDNN 9.x not installed or not in the environment variables. Note that AMD/Intel GPUs and integrated graphics cannot enable CUDA.

Troubleshooting in this order:

1. Confirm the hardware: check whether the computer has an NVIDIA discrete GPU (integrated AMD/Intel graphics do not support CUDA).
2. Update the NVIDIA driver to the latest version (CUDA 12.8+ requires the newer driver).
3. Install CUDA 12.8 and cuDNN 9.11, and add the bin and lib directories of cuDNN to the system Path environment variable.
4. Restart the software, then check CUDA Acceleration on the main interface and start the task again.
5. Check whether the bottom progress region still reports that GPU acceleration is not enabled; you can verify with Menu -> Help -> CUDA Help.

Additionally: the current software directory C:/Users/jeffy/Downloads/win-pyvideotrans-v4.17-1028 (1) contains spaces and parentheses and is located in the system user directory, which can easily cause loading failures of underlying libraries (including CUDA-related dlls). It is recommended to extract the software to a short path containing only English letters and digits, for example D:\pyVideoTrans, and run it with administrator privileges after adding it to the security software trust list.

If all the above are confirmed correct and the error persists, provide the last 30 lines of the latest log file in the logs/ folder under the software installation directory (which will record the specific reason for CUDA detection failure, such as driver version or missing dll).

Please refer to the related documentation:

• Configuring CUDA and cuDNN on Windows