Error when I use cuda_runtime.h and OpenVINO EP at the same time

[ddrepkk89]

I have custom build the onnxruntime 1.20.0 with TensorRT, CUDA and OpenVINO. Build is done successfully without error. I have two sessions to initialize. I have tried the following:
Option 1: Initialize both sessions with OpenVINO -> WORKS!
Option 2: Initialize both sessions with TensorRT -> WORKS!
Option 3: Initialize one sessions with CUDA, another with OpenVINO-> WORKS!
Option 4: Initialize one session with OpenVINO, another with TensorRT without including cuda_runtime.h -> WORKS!
Option 5: Initialize one session with OpenVINO, another with TensorRT with including cuda_runtime.h -> Not Work.

I need to include cuda_runtime.h because I have other kernel functions, I need to use.

The error message at option 5 is: OpenVINO "No available backend". Why there are conflict between cuda_runtime.h and OpenVINO provider? It seems, once I load the cuda dlls, there are conflict, and OpenVINO recognize the CPU as NPU (But I don't have NPU on my PC)

My CUDA Version is 12.4, CuDNN Version 9.4, TensorRT 10.5, OpenVINO 2024.4.0.

My operating system is Windows 11. Visual Studio 2022.

[github-actions]

This issue has been automatically marked as stale due to inactivity and will be closed in 30 days if no further activity occurs. If further support is needed, please provide an update and/or more details.

[snnn]

This issue has been automatically closed because it has been marked as 'stale' for more than 30 days without activity. If you believe this is still an issue, please feel free to reopen it.