There arent any tips here really 🙂 Every Parallel node is updated per frame while it is running of course. Their update loop per-se is only as performance heavy as the conditions it checks or the actions it performs. Also note that a Parallel node that has finished (not set to Repeat) will not have any impact.
In the end, I think that you should use as many as you think are required to achieve what you are after 🙂
Gavalakis
