Neeed clarity for batch-size and batched-push-timeout for rtsp source

Accelerated Computing Intelligent Video Analytics DeepStream SDK
, , ,
1

• Hardware Platform (Jetson / GPU) GPU (1080ti)
• DeepStream Version 5.0
• JetPack Version (valid for Jetson only)
• TensorRT Version 7.2.1
• NVIDIA GPU Driver Version (valid for GPU only) 450.102.04
• Issue Type( questions, new requirements, bugs) Question
• 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)

Hi,
I want to know few things (specially for RTSP source)-

  1. If i want to use 10 rtsp cameras with 20 fps , 1080p, 4096 constant bitrate, h264 decoder so what should be my batch-size and batched-push-timeout values?

As in this post (“DS 5.0.1 nvstreammux batch-size bug? - #12 by Fiona.Chen”), i can see max batch-size i can set is 4 due to HW decoder limitation. But it is generally recommended that to use batch-size same as number of sources.

  1. If i set it 4, then output gets delayed and memory is constantly goes to increase (in around 3-4 hrs) like it shows sync behaviour. And when i set it to 10, output is realtime but glitchy images starts appear.
2
  1. Batch size of nvstreammux should be the same to stream number. For 10 streams, nvstreammux batch size should be 10 too. The post DS 5.0.1 nvstreammux batch-size bug? - #12 by Fiona.Chen wants to set nvstreammux batch size larger than stream number, this is not recommended. batched-push-timeout can be calculated by 1000000(us)/FPS = 50000
  2. Batch size of nvstreammux should be the same to stream number. Frequently Asked Questions — DeepStream 6.1.1 Release documentation
3

One more thing is does bitrate from rtsp camera affects the decoder performance?
Like if i set it 1024, then 6 cameras runs fine, if i set it to 4096, then glitch appears.
Also what should be the ideal iframeinterval for 20fps?

Note-
Network Bandwidth is not the bottleneck.

4

The high bitrate will cause more network delay, deepstream is only a rtsp client. You can monitor the GPU loading to check whether video decoder is the bottleneck when you run your case.

5

ok thanks for the support