The only relevant consideration in real time rendering is speed right, so you’re saying this is a good approach?
I am rendering depth cubemaps for 8 point light sources, and storing each within a cubemap array. From those depth maps I’m calculating shadows for the 8 light sources. This is all updated dynamically every frame and this is working fine, but I want to speed it up by only rendering one shadow map per frame, the effect will be unnoticeable and the performance gain will be huge. The thing is I need to be clear only a subset of the cubemap array per update loop and that is exactly my question, what is the most efficient way to do this?