Why the multi channel video input is so slow?(Deepstream Python Apps)

• Hardware Platform (Jetson / GPU) Jetson Nano

• DeepStream Version 5.0

• JetPack Version (valid for Jetson only) 4.4

• Sample application and the configuration file content deepstream python apps deepstream_test_3.py(Multi video input)

When I try to run the deepstream_test_3.py, it seems to be so slow in screen, and the terminal shows a 6 fps.
I run the python app with h264, 1080p*4.
The C code sample looks like well and can run smoothly with 8 sources input.
How can use python code to run 4 or 8 sources fluently?
Forgive my bad English.

Are you using same config and same video sources for c version and python version?

I use the original(default) config file in the corresponding folder no matter in Python version code and C version code.
I try it again today.I am sorry that the result is different from my question.
It looks like the speed is quite the same no matter in Python version code and C version code.
The result is that when I use 4 sources input to run deepstream_test_3.app(both C and Python version), it will be very slow in the screen.
But when I use the command below in /opt/nvidia/deepstream/deepstream-4.0/samples/configs/deepstream-app
deepstream-app -c source8_1080p_dec_infer-resnet_tracker_tiled_display_fp16_nano.txt
It will run fast and smoothly with 8 video input.
How can I use C or Python code to run 8 input sources samples fast like deepstream app?
I notice that the deepstream app use engine model but the C or Python use caffe model.
Do different models make speed difference?
It will be very helpful to my work because I want use an 8-channel nano to monitor 8 RTSP or USB cameras.
The benefits will be very greatful.

Are you using same input for deepstream-app and test3 sample? and please be noted, nano can reach 8 channels for local stream, but for rtsp stream, it can’t by now. but you can tune the performance, check this trouble shoot,
The DeepStream application is running slowly (Jetson only).