Multiple engine files created for deepstream-rtsp-in-rtsp-out

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU)- GPU
• DeepStream Version- 6.0
• TensorRT Version- 8.2.1.8
• NVIDIA GPU Driver Version (valid for GPU only)- 470

Hello,
Whenever i am running deepstream-rtsp-in-rtsp-out for multiple inputs then multiple engines files are created with giving this warning (deserialized backend context :resnet10.caffemodel_b1_gpu0_int8.engine failed to match config params, trying rebuild). For my project i don’t want to create multiple engine files, can you please help to find solution for this.
I would really appreciate your help. Thanks in advance.

Sorry for the late response, our team will do the investigation and provide suggestions soon. Thanks

Have you changed anything in this sample? source code or config file?

No haven’t changed anything in sample app/config file and source code. Only running sample apps for deepstream-rtsp-in-rtsp-out with multiple inputs.

Awaiting for solution

The sample apps should be run under root privilege.

Hie i tired with your solution but it didn’t work. Whenever input is increasing then multiple engine files are created. For me i don’t want to create multiple engine file.

Can you list the engine files’ names?

resnet10.caffemodel_b3_gpu0_int8.engine,resnet10.caffemodel_b2_gpu0_int8.engine,resnet10.caffemodel_b1_gpu0_int8.engine

Which GPU are you working on?

NVIDIA GeForce GTX 1080/PCIe/SSE2

You got several engine files with different batch sizes. But in the batch size in the config file is 1

deepstream_python_apps/dstest1_pgie_config.txt at master · NVIDIA-AI-IOT/deepstream_python_apps (github.com)

Seems you have changed something in the sample.

When you input one stream, “resnet10.caffemodel_b1_gpu0_int8.engine” will be generated. When you input two streams, “resnet10.caffemodel_b2_gpu0_int8.engine” will be generated.
When you input 3 streams, “resnet10.caffemodel_b3_gpu0_int8.engine” will be generated.

So seems you have run the sample at least three times. And you input different streams for each time. It is correct.

When you input 3 streams for the second time, the “resnet10.caffemodel_b3_gpu0_int8.engine” is already there, no new engine will be generated.

Nothing wrong.

This topic was automatically closed 14 days after the last reply. New replies are no longer allowed.