Please provide complete information as applicable to your setup. <a name="p-1534680-hardware-platform-jetson-gpu-jetson-deepstream-version-62-jetpack-version-valid-for-jetson-only-502-tensorrt-version-8415-cuda-114239-cudnn-84150-issue-type-questions-new-requirements-bugs-questions-how-to-reproduce-the-issue-pull-the-camera-rtsp-stream-in-the-pipeline-of-deepstream-test3-app-the-camera-is-directly-connected-to-the-stream12-channels-of-camera-data-take-the-src-pad-in-nvosd-and-add-the-latency-probe-to-get-the-latency-of-each-plug-in-1the-results-of-the-two-latency-tests-differ-considerably-first-nvds_enable_latency_measurement-is-1-1" class="anchor" href="#p-1534680-hardware-platform-jetson-gpu-jetson-deepstream-version-62-jetpack-version-valid-for-jetson-only-502-tensorrt-version-8415-cuda-114239-cudnn-84150-issue-type-questions-new-requirements-bugs-questions-how-to-reproduce-the-issue-pull-the-camera-rtsp-stream-in-the-pipeline-of-deepstream-test3-app-the-camera-is-directly-connected-to-the-stream12-channels-of-camera-data-take-the-src-pad-in-nvosd-and-add-the-latency-probe-to-get-the-latency-of-each-plug-in-1the-results-of-the-two-latency-tests-differ-considerably-first-nvds_enable_latency_measurement-is-1-1" aria-label="Heading link"> **• Hardware Platform (Jetson / GPU) Jetson **• DeepStream Version 6.2 **• JetPack Version (valid for Jetson only) 5.0.2 **• TensorRT Version 8.4.1.5 **• CUDA 11.4.239 **• cuDNN 8.4.1.50 **• Issue Type( questions, new require…

[image] orin_test: The total latency of the three plugins is 163ms. The latency is just a pipeline status measurement, we can’t say it is reasonable or not reasonable. There are many factors which will impact the latency data, E.G. the network bandwidth and the transferring efficiency will al…

[image] orin_test: nvv4l2decoder3 Dear Foina.Chen Thanks for you reply. regarding the Nvinfer latency in “FakeSink”. can I understand that Nvinfer infers how many frames are grouped in a batch at the moment in Nvstreammux, and the latency of Nvinfer is the latency of processing this batch. n…

For “fakesink”. If we want to measure nvv4l2decoder latency/nvinfer latency/nvtracker latency of multi-cameras, how should we measure it?

[image] orin_test: for example, in the BATCH-NUM = 2852, is the Nvinfer latency of 10.031982ms the latency of processing two frames? Comp name = primary-nvinference-engine in_system_timestamp = 1733740556277.585938 out_system_timestamp = 1733740556287.617920 component latency= 10.031982 It…

[image] orin_test: For “fakesink”. If we want to measure nvv4l2decoder latency/nvinfer latency/nvtracker latency of multi-cameras, how should we measure it? The video data before nvstreammux is frames in separated streams. After nvstreammux the video data are combined into batches. [image…

hi Fiona.Chen: I read your reply above carefully, it is very helpful, and now we want to see the output frame rate of each camera after pipeline processing, similar to the following: **PERF : FPS_0 (14.99) FPS_1 (14.99) FPS_2 (14.99) FPS_3 (14.96) FPS_4 (14.99) FPS_5 (14.99) FPS_6 (14.99) FPS_7 (1…

[image] orin_test: How is this FPS_x printed in the pipeline? How to print this information if a new pipeline is created？ Please refer to DeepStream SDK FAQ - Intelligent Video Analytics / DeepStream SDK - NVIDIA Developer Forums

If the pipeline consists only of nvurisrcbin, nvstreammux, and fakesink, and the timeout setting in nvstreammux is 0.04(ms), can I understand that the data flow of the entire pipeline is basically unblocked? thanks Fiona.Chen

[image] orin_test: can I understand that the data flow of the entire pipeline is basically unblocked? What do you mean by “unblocked”?

the pipeline consists only of nvurisrcbin, nvstreammux, and nv3dsink -->data block? the pipeline consists only of nvurisrcbin, nvstreammux, and fakesink-->data ubblock? Taking the above pipeline as an example, I found that if the pipeline is last with nv3dsink, the latency of decode and mux will …

nv3dsink will not block the pipeline if you set “sync=0” and “async=0” properties with it. The same case to fakesink.

An understanding of the delay result produced by latency_measurement_buf_probe

Accelerated Computing Intelligent Video Analytics DeepStream SDK

Fiona.Chen December 11, 2024, 6:51am 15

Please refer to DeepStream SDK FAQ - Intelligent Video Analytics / DeepStream SDK - NVIDIA Developer Forums

Topic		Replies	Views
Latency_measurement_buf_prob The results of the two latency tests differed greatly DeepStream SDK cudnn , deepstream	2	115	December 10, 2024
Can the Speed of nvvideoconvert Vary? DeepStream SDK gstreamer , deepstream	5	247	August 27, 2024
RTSP latency does not work with NVSTREAMMUX DeepStream SDK nvbugs	36	5940	January 9, 2022
How to messure actual nvv4l2decoder0 latancy? DeepStream SDK	5	355	July 13, 2023
Multisession inference, segmentation DeepStream SDK jira	50	2333	July 3, 2024
How to test the pipeline latency of deepstream test3? DeepStream SDK deepstream	10	277	May 23, 2025
Slow decoding T4 DS 5.0.1 DeepStream SDK	10	720	March 2, 2021
There is a question between streammux and nvinfer DeepStream SDK	17	319	December 9, 2025
Delay increase in rtsp over time DeepStream SDK deepstream	34	683	December 5, 2024
How to improve the performance of pipelines in deepstream DeepStream SDK deepstream	25	775	October 21, 2025

An understanding of the delay result produced by latency_measurement_buf_probe

Related topics