Skip to content

CUDA device requested but falls back to CPU #10

Description

@woldelow

When attempting to process a video, the application requests cuda:0, but immediately triggers a fallback to CPU inference because no CUDA inference provider is available in the bundled environment (CUDA device 0 was requested but no CUDA inference provider is available; using CPU inference).

Even though an NVIDIA GPU is present and hardware encoder (h264_nvenc) is detected, all ONNX-based tasks (like RapidOCR) run strictly on the CPU, causing significantly slower processing times.

2026-08-31 19:34:19,320 [WARNING] CUDA device 0 was requested but no CUDA inference provider is available; using CPU inference.
2026-08-31 19:34:19,320 [WARNING] Inference device fallback: cuda:0 -> cpи
2026-08-31 19:54:58,719 [WARNING] CUDA device 0 was requested but no CUDA inference provider is available; using CPU inference.
2026-08-31 19:54:58,719 [WARNING] Inference device fallback: cuda:0 -> cpu
2026-08-31 19:54:59,028 [INFO] Hardware encoder available: h264_nvenc
2026-08-31 19:54:59,028 [INFO] Detector: RapidOCR | Inpainter: sttn | Device: cpu | HW encode: h264_nvenc

Windows

Activity

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions