#6669 CUDA GPU acceleration unavailable. Verify NVIDIA GPU & CUDA installation (Help -> CUDA Help).

CUDA GPU acceleration is unavailable because the software requires three conditions to be met simultaneously: an NVIDIA GPU, an up-to-date NVIDIA driver supporting CUDA 12.8+, and correctly installed CUDA Toolkit 12.8+ / cuDNN 9.11+ with proper PATH environment variables.

Please check the following in order:

1. Confirm NVIDIA GPU and driver support
   Open Command Prompt (CMD) and run:

   • nvidia-smi
     If this fails or shows no GPU, your machine does not have a usable NVIDIA GPU (AMD/Intel GPUs do not support CUDA here).
     If it succeeds, check the "Driver Version" and "CUDA Version" shown at the top. The driver must support CUDA 12.8 or higher; if it is lower, update your NVIDIA driver from the official NVIDIA website.

2. Confirm CUDA Toolkit is installed
   Run in CMD:

   • nvcc -V
     If this fails, CUDA Toolkit is not installed or not on PATH. Install CUDA Toolkit 12.8 or newer.
3. Install cuDNN 9.11+
   CUDA alone is not enough — cuDNN 9.11 or later must also be installed.
4. Verify environment variables
   Ensure the system PATH variable contains the CUDA bin and lib directories (e.g., C:\Program Files\NVIDIA GPU Computing Toolkit\CUDA\v12.8\bin). After adding them, restart the software.
5. Restart and retry
   After all installations, fully close and reopen pyVideoTrans, then enable the CUDA option again.

Note: CUDA acceleration only affects local ASR/TTS channels (e.g., faster-whisper, openai-whisper, Qwen3-TTS). Online cloud APIs do not use local GPU.

If the issue persists after the above steps, please provide the latest log file from the logs/ folder under the software directory (copy the last ~30 lines of error output), plus confirm your GPU model and driver version.

请查阅相关文档：

• Windows上配置CUDA和cuDNN