Can’t able run more than 12 stream in deepstream python

Accelerated Computing Intelligent Video Analytics DeepStream SDK
1

Please provide complete information as applicable to your setup.

• Hardware Platform (Jetson / GPU) : Jetson Orin Nano 4GB
• DeepStream Version : 6.1.1
• JetPack Version (valid for Jetson only) : 5.0.2
• TensorRT Version: 5.0.2
• NVIDIA GPU Driver Version (valid for GPU only): 11.4
• Issue Type( questions, new requirements, bugs)
I am recording and inference rtsp streams using deepstream python. For inference I am using yolov7-tiny. I can only able to achieve 12 rtsp streams. if I increase to 13 or 14, fps is low like 18 to 19.
I also tried with Jetson Orin Nano 8gb module, in that I can achieve 14 streams, but can’t able to increase further.

• How to reproduce the issue ? (This is for bugs. Including which sample app is using, the configuration files content, the command line used and other details for reproducing)
create pipeline in the following order, nvstreammux->nvinfer->nvvideoconvert->capsfilter->nvstreamdemux->queue->nvvideoconvert->nvv4l2h264enc->h264parse->splitmuxsink

• Requirement details( This is for new requirement. Including the module name-for which plugin or for which sample application, the function description)
is the above pipeline is correct or not, if correct, why I can’t able to run more than 12 streams, is there a any limitation
kindly please help me to resolve this issue

3

Can you check check the output of command “tegrastats”? It looks to be a bottleneck on some hardware component.
As there is no hardware encoder in Orin Nano, I’m wondering what the behavior is for plugin “nvv4l2h264enc” (sorry that I don’t have a Orin Nano to test it)

4

In which platform?

Your pipeline will not work in Orin Nano since there is no HW encoder in Orin Nano.

6

I am using Orin Nano Emulator in AGX Orin. Can you help me how to record or store rtsp streams in effective way in deeepstream python.

Screenshot from 2023-04-10 11-15-58
Screenshot from 2023-04-10 11-15-581593×565 312 KB

Screenshot from 2023-04-10 11-20-21
Screenshot from 2023-04-10 11-20-211566×1079 258 KB

8

There is no HW encoder in Orin Nano. Your pipeline does not work with Orin Nano. If you insist to encode the 14 streams, Orin Nano is not recommended.

For AGX Orin, have you enable maximum power mode and lock the VIC frequency?
https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_Performance.html#id19
https://docs.nvidia.com/vpi/algo_performance.html#maxout_clocks

9

but its working for 12 streams.

10

You are working with AGX Orin but not Orin Nano.

11

which is the correct way to record or store(.mp4) rtsp streams?

12

For AGX Orin, have you enabled maximum power mode and lock the VIC frequency?
https://docs.nvidia.com/metropolis/deepstream/dev-guide/text/DS_Performance.html#id19
https://docs.nvidia.com/vpi/algo_performance.html#maxout_clocks

13

no I don’t want for agx orin, only jetson orin nano or jetson nano

I am using emulator to convert agx orin to orin nano, if this is not suitable to test orin nano performance.

15

There is no HW encoder in Orin Nano. Your pipeline does not work with Orin Nano. To record with software encoder will take a lot of CPU resource, so the performance will be very poor.

If the poor performance is OK, you can use software encoder in GStreamer to generate RTSP streams. GStreamer: open source multimedia framework

16

is HW encoder is available in jetson nano?

17

There is HW encoder in Jetson Nano. But the memory and GPU are poor.

18

ok. can you guide which is the correct deepstream pipeline to record(.mp4) rtsp streams?

19

For generate RTSP streams, please refer to GStreamer document and community. GStreamer: open source
There is also RTSP output sample in pyds sample, please replace the HW encoder with SW encoder if you want to use in Orin Nano.

20

not rtsp out,

input rtsp and output .mp4

21

For mp4 mux, please refer to GStreamer. It is a must to have basic GStreamer knwoledge and coding skills before you start with DeepStream.

22

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