Performance question: Many layers containing few markers or few layers with many markers

[edwardribbit]

What is the runtime performance in terms for computational efficiency between having 1 map layer that contains 20 markers versus 20 layers containing 1 marker each? When panning/zooming around, does one structure somehow result in less cpu operations?

I assume the choice of grouping markers in layers depend on whether they are semantically relevant with the benefit of being able to remove all markers by removing the single layer rather than looping through all markers and removing each one individually.