How to make four pipelines or more?

Accelerated Computing Intelligent Video Analytics DeepStream SDK
#1

• Hardware Platform GPU
• DeepStream 5.0
• TensorRT Version7.1
• Gst-python
• NVIDIA GPU Driver Version 450.57
• nvcr.io/nvidia/deepstream:5.0-dp-20.04-base

Hi
I am using tee-function to make parallel processing pipeline.
I have successfully created 3-stream pipeline performing 3-class object detection. But it was not working with 4-stream pipeline to detect 4-class object each.
Is this resource problem?

Cpu used 5%
Gpu used 9%
memory used 4%

Error: gst-resource-error-quark: Device ‘/dev/nvhost-msenc’ failed during initialization (1): gstv4l2object.c(4052): gst_v4l2_object_set_format_full (): /GstPipeline:pipeline0/nvv4l2h264enc:nvv4l2h264enc1:
Call to S_FMT failed for YM12 @ 640x480: Unknown error -1

Regards Chan

#3

Which platform are you working on?

#4

RTX 2080ti

#5

Please check https://developer.nvidia.com/video-encode-and-decode-gpu-support-matrix-new.
RTX 2080ti only supports at most 3 streams HW encoding sessions. It is a hardware limitation. So you need to change to software encoder for more encoding sessions.