GPU cann't be used after deploy with Microsoft.ML.OnnxRuntime.Gpu

[evolcano]

System Information (please complete the following information):

• OS & Version: [Windows 11 23H2]
• Microsoft.ML.OnnxRuntime.Gpu Version: [Version="1.16.3]
• .NET Version: [NET 8]
• Visual studio 2022

Describe the bug I download a demo for object detection. My laptop has two GPU: 0 intel & 1 Nvidia, so I have set ""Environment.SetEnvironmentVariable("CUDA_VISIBLE_DEVICES", "1"); (Source code: yolov5-net-master.zip ) When in debug mode, GPU is used, it's seems good. But after deploy without any change just change to release model, GPU is not used.

To Reproduce Steps to reproduce the behavior:

1. Open sloution;
2. Debug, Gpu is used
3. Delopy to local file system in release win-x64, net8.0-windows
4. Click exe to execute, GPU cann't be used

Expected behavior GPU should be used after deploy and consist with debug.

Screenshots, Code, Sample Projects 1、When debug, GPU is used

2、Deploy setting

3、When deployed, GPU cann't use

Additional context NVIDIA-SMI 546.33 Driver Version: 546.33 cuda_11.6.2_511.65_windows cudnn-windows-x86_64-8.9.6.50_cuda11

[evolcano]

Maybe the GPU is actually used because the temperature is increasing...

I can see the usage of GPU now after reinstall.