i use script to instance 30000 instance and add them to the scene.it just cost 2-3 seconds
but if i quit() the scene or queue_free() the scene it is really slow!!
the test example was uploaded
As the profiler show the performance is lost in the destructor of Object.

I'm not entirely sure why the performance should be bad with disconnect since maps should be pretty efficient in terms of performance. It's worth checking this out to see if there's some way to improve performance of our disconnection.
@chenjie199234 Why closing this issue? It is not solved yet.
@Geequlim just closed all my issue,just disappointed,i will go back to use urho3d
@chenjie199234 Reopening for now, feel free to ignore and have fun & success with Urho, but we still need to fix this. Thanks for reporting!
if you need 30k instances, using servers directly or multimesh seems like a better bet, but will give it a check anyway
This seems like a non-issue to me. We should just be documenting how to achieve 1000+ instances using particles, multimesh, server.
This was actually a design problem which was just fixed.