DeepStream 5.1 H265 Use Too Much Cpu

• Hardware Platform (Jetson / GPU) GPU
• DeepStream Version 5.1
• TensorRT Version 11.2
• NVIDIA GPU Driver Version (valid for GPU only) 460.32.03
• Issue Type( questions, new requirements, bugs)
• 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)

在项目配置中使用了rtsp流,视频流格式为h265,最小运行大概5小时之后,会出现CPU占用率高于95%的情况;
而在视频流格式为h264的情况下,未出现问题

主要的CPU使用出现在nvv4l2decoder插件中。

想请教一下这个问题怎么解决呢?

Can you share the pipeline you are using? Thanks.
For the high CPU usage issue, is the CPU usage normal in the first 5 hours?

Thanks Reply.
We use the sample_apps/deepstream-app , added properties “short-header” to “rtspsrc” with TRUE in deepstream_source_bin.c.

And set “drop-frame-interval=30” for all sources;
set “width=1920 height=1080” for streammux;
close “tiled-display” and “osd”.

By the way,the h265 source is 2560x1440 fps:25.

Yes,first 5 hours CPU usage is normal.Time varies in different environments.

Hi happo, Could you give us your rtsp src and configure file? Thanks

deepstream_app_person.txt (8.9 KB)
config_infer_primary_peoplenet.txt (1.8 KB)
dstest5_msgconv_sample_config.txt (5.4 KB)
labels_peoplenet.txt (16 Bytes)

that’s my total configure files.

Hi @happo , it seems that you have 30 h265 sources and mainly the decoder occupies the CPU.
Cause I do not have your stream, So could you help to have a simple test in your env?
1.Save your stream to a mp4 file(you can also give that file to us)
2.Modify the config file blow simply:

/opt/nvidia/deepstream/deepstream-XX/samples/configs/deepstream-app/source30_1080p_dec_infer-resnet_tiled_display_int8.txt

change the [source0] [source1] uri to your h265 file
change the file_loop=1
3.run the cli

deepstream-app source30_1080p_dec_infer-resnet_tiled_display_int8.txt

4.Observe the CPU utilizition
Thanks

thanks all

My config set “rtsp-reconnect-interval-sec=30”
I found too many of the 30 rtsp sources lost data, the “g_timeout_add" func will stock loop thread.
Then cpu usage increased.