#8041 CUDA GPU acceleration unavailable. Verify NVIDIA GPU & CUDA installation (Help -> CUDA Help).

This error indicates that the NVIDIA CUDA environment is not properly installed or detected by the software. pyVideoTrans relies on NVIDIA's CUDA 12.8+ and cuDNN 9.11+ for GPU acceleration; without them, the software falls back to CPU or reports this warning.

Follow these steps to enable CUDA acceleration:

1. Verify NVIDIA GPU

   • Open Command Prompt and run nvidia-smi to confirm you have an NVIDIA GPU and check the driver version.
   • AMD or Intel GPUs are not supported for CUDA acceleration.

2. Install CUDA Toolkit 12.8 or newer

   • Download and install from NVIDIA's official site.
   • After installation, run nvcc -V in Command Prompt to confirm the compiler version is 12.8 or above.

3. Install cuDNN 9.11 or newer

   • Download matching cuDNN for CUDA 12.8 from NVIDIA.
   • Add the bin and lib directories of cuDNN to your system Path environment variable.

4. Check environment variables

   • Ensure both CUDA's bin and lib directories are present in Path. Restart the software after modifying environment variables.

5. Enable CUDA in the software

   • In the main interface, check the box "Enable CUDA?" (or "CUDA加速?") before starting a task.
   • Some channels (e.g., faster-whisper, openai-whisper, Qwen-ASR) require this checkbox to attempt GPU usage; others (e.g., Qwen-ASR, HuggingFace_ASR) automatically use GPU if CUDA is available.

If you have completed all steps and the warning persists, run the following in Command Prompt and compare the outputs:

• nvcc -V → should show "release 12.8" or higher
• nvidia-smi → driver version should be recent; check the "CUDA Version" at the top-right (this is the maximum supported CUDA version, which must be at least 12.8)

Also ensure the software installation path is simple English (e.g., C:\pyVideoTrans), as unusual paths can conflict with loading CUDA libraries.

请查阅相关文档：

• Windows上配置CUDA和cuDNN