How to profile Beluga

Generating a flamegraph from recorded perf events

A flamegraph is a convenient tool for understanding how CPU time is being used.

To start, run the docker container in privileged mode:
```
docker/run.sh -p
```

Your code needs to be built with debug info and may also need frame pointers. For that, use:

colcon build --symlink-install --cmake-args -DCMAKE_BUILD_TYPE=RelWithDebInfo -DCMAKE_CXX_FLAGS="-fno-omit-frame-pointer"

Run the following script to generate the profiling data:
```
source /ws/install/setup.bash
ros2 run beluga_benchmark profile_amcl_with_bagfile
```
perf will generate a perf.data file in the folder it was run.
To generate a flamegraph from the recorded data, run:
```
source /ws/install/setup.bash
ros2 run beluga_benchmark flamegraph  # This may take a long time
```
To visualize the flamegraph and be able to zoom it in or out, open the generated svg file in a web-browser.

References