IN nvidia-smi dmon for my rtsp stream all time enc column is show "0" What can be the problem

Please provide complete information as applicable to your setup.

**• Hardware Platform ------> GPU
**• DeepStream Version ----> 6.4
• JetPack Version (valid for Jetson only)
**• TensorRT Version ----> 8.6
**• NVIDIA GPU Driver Version ---------> 545
**• Issue Type -----> questions

I’m using " NVurisrcbin " and whole pipeline is idle. When I am checking for nvidia-smi dmon
enc is always “0” dec is doing things up-down between some range and depends on how many camera I’m adding …

My question is why the encode is showing always 0. If you can give more insides that can help me !

gpu pwr gtemp mtemp sm mem enc dec jpg ofa mclk pclk

Idx W C C % % % % % % MHz MHz

0     41     52      -     18      7      0      0      0      0   7000   1560 
0     39     52      -     11      5      0      0      0      0   7000   1560 
0     41     53      -     15      7      0     12      0      0   7000   1560 
0     43     53      -     15      6      0      4      0      0   7000   1560 
0     42     53      -     17      6      0      8      0      0   7000   1560 
0     42     53      -     19      6      0      7      0      0   7000   1560 
0     42     53      -     22      8      0      8      0      0   7000   1560 
0     42     53      -     17      6      0      6      0      0   7000   1560 
0     42     53      -     15      6      0      4      0      0   7000   1560 
0     42     53      -     17      6      0      7      0      0   7000   1560 
0     42     54      -     16      6      0      8      0      0   7000   1560 
0     42     54      -     23      8      0     11      0      0   7000   1560 
0     42     54      -     16      6      0      7      0      0   7000   1560 
0     42     54      -     15      6      0      3      0      0   7000   1560 
0     42     54      -     17      7      0      8      0      0   7000   1560 
0     42     54      -     19      7      0      8      0      0   7000   1560 
0     41     54      -     13      5      0      5      0      0   7000   1560 
0     43     54      -     17      6      0      7      0      0   7000   1560 
0     42     54      -     16      6      0      4      0      0   7000   1560 
0     42     54      -     18      7      0     10      0      0   7000   1560 
0     42     54      -     15      6      0      8      0      0   7000   1560 
0     42     54      -     18      7      0      7      0      0   7000   1560 
0     43     54      -     20      8      0      7      0      0   7000   1560 
0     43     55      -     17      6      0     15      0      0   7000   1560 
0     43     55      -     18      7      0      6      0      0   7000   1560 
0     42     55      -     18      7      0      7      0      0   7000   1560 
0     43     55      -     23      7      0      7      0      0   7000   1560 
0     43     55      -     22      7      0      5      0      0   7000   1560 
0     42     55      -     18      6      0      5      0      0   7000   1560 
0     43     55      -     17      6      0      7      0      0   7000   1560 
0     43     55      -     18      7      0      7      0      0   7000   1560 
0     43     55      -     18      7      0      7      0      0   7000   1560

How many encoders are you using at the same time? On my L4 GPU, I encoded three rtsp streams(all of them are 1920x1080) at the same time, and the encoder occupied only 2%

# gpu    pwr  gtemp  mtemp     sm    mem    enc    dec    jpg    ofa   mclk   pclk 
# Idx      W      C      C      %      %      %      %      %      %    MHz    MHz 
    2     30     43      -      2      1      2      1      0      0   6250   2040 
    2     30     43      -      2      1      1      0      0      0   6250   2040 
    2     30     43      -      2      1      2      0      0      0   6250   2040 
    2     30     43      -      2      1      1      1      0      0   6250   2040 
    2     30     43      -      2      1      2      1      0      0   6250   2040 
    2     31     43      -      2      1      1      0      0      0   6250   2040 
    2     31     43      -      2      1      2      0      0      0   6250   2040 
    2     32     43      -      2      1      2      1      0      0   6250   2040 
    2     31     43      -      2      1      1      0      0      0   6250   2040 
    2     32     43      -      3      1      2      0      0      0   6250   2040 
    2     31     43      -      2      1      1      1      0      0   6250   2040 
    2     31     43      -      2      1      2      0      0      0   6250   2040 
    2     31     43      -      2      1      1      0      0      0   6250   2040 
    2     31     43      -      2      1      2      1      0      0   6250   2040 
    2     30     43      -      2      1      1      1      0      0   6250   2040 
    2     30     43      -      3      1      1      0      0      0   6250   2040 

@junshengy Thanks for replay ! I’m using 10 RTSP streams ! Where I can understand How many encoder I’m using ?.
my laptop GPU is 3070.

Do you mean 10 inputs?

I ran three deepstream_test1_rtsp_in_rtsp_out.py instances, nvidia-smi cannot know the number of encoder instances.

Also, which example did you use? The number of instances in the code depends on this line of code

encoder = Gst.ElementFactory.make("nvv4l2h264enc", "encoder")

I am running with my custom pipeline Where I’m not using “nvv4l2h264enc”/“nvv4l2h265enc” directly
to get the frame from camera I am using “nvurisrcbin” which internally doing encoding H265/H264 by itself right ?
In the pipeline I’m not using ‘’‘’‘‘Gst.ElementFactory.make(“nvv4l2h264enc”, “encoder”) ‘’’’ this one directly !

can you give me more clear picture for better understanding !

This plugin is only responsible for decoding and does not use encoding.

What kind of sink does your program use? 0% is normal if you are not using encoding

I’m using appsink as sink element. When I need to use encoder ?
and one more question is NVjpeg encoder can able to using GPU encoder core ?

No need.

I don’t think it should be used. nvjpeg is implemented by cuda and should only use the gpu core.

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