Iterating all static objects in compact array is super fast. Around 1000 objects in our scene cost less than 0.1ms. Renderer always do normal frustum culling, so if engine would use this optimization then it wouldn’t cost anything extra.
Clip plane will cull objects but this culling happen after vertex shader. So all objects that are rendered that are behind clip plane will still cost extra draw calls and vertex shader performance. This could be fixed by easily by replacing frustum near plane with clipping plane.