Real-time profiler plugin for Gstreamer

• GstShark

https://github.com/RidgeRun/gst-shark

• For further information, visit GstShark wiki

https://developer.ridgerun.com/wiki/index.php?title=GstShark

• plotly (for Log Visualizer)

https://plotly.com/

• On Ubuntu x64

$ sudo apt install gtk-doc-tools libgraphviz-dev libncurses5-dev libncursesw5-dev
$ ./autogen_ubuntux64.sh
$ make
$ sudo make install

For other environment, visit GstShark wiki

Since gst-shark have gst-indent, you should run gst-indent before commit

common/gst-indent <file name>

Set environment variable as below.

• GST_DEBUG = "GST_TRACER:7"
• GST_TRACERS = "live"

$ GST_DEBUG="GST_TRACER:7" GST_TRACERS="live"\
     gst-launch-1.0 videotestsrc ! videorate max-rate=15 ! fakesink

• python 3.x
• plotly
• numpy

If you want to record log, add LOG_ENABLED environment like below

$ GST_DEBUG="GST_TRACER:7" GST_TRACERS="live" LOG_ENABLED=TRUE\
     gst-launch-1.0 videotestsrc ! videorate max-rate=15 ! fakesink

You should know directory which have log (default path is /nnstreamer/bin/gstshark_<timestamp>)

python3 scripts/graphics/log_visualizer.py --dir=<log directory>

It will open 3 browsers.

Each browser have Two graph and below one is Buffer Timeline graph. User can see which buffers each pad is handling at a specific time.

User can check CPU usage over time.

User can check processing time for each element.

User can check buffer rate for each pad