What performance when using `load_yml_dags` to loading entire dag folder.

[raisinbl]

As I understand, load_yml_dags functions accept dag folder as an argument, it will loop through every .yaml and parsing it to generate DAG.
A core concept of Airflow is Scheduler will parsing entire dag folder in specific time interval, that means, Airflow will execute every .py file.
What I concern is: with every specific time interval like that, Airflow will rerun code and calling load_yml_dags to parse entire dag config, which dosen't like a pure py dag file, which've been compiled and cached.
have you've been testing parsing time of that function on large dag folder, e.x: 1000+ file