Why is the maximized radar view so much more performant?

[Master-Athmos]

I wondered why there is this huge performance difference between using a standard camera view and a maximized radar screen. Of course using the maximized radar screen should be way more performant as it has to render just a couple of sprites - nevertheless I wonder why the performance is so much better even when you move to the corner of the map and zoom right on top of the ground so you have nothing in your view except a small piece of terrain (and maybe the skyboxes)...

This actually gives next to no performance improvement. I recently set some extremely high settings with a high view distance thing to do some eyecandy screenshots and let some AIs fight. When viewing like half of the entire map with a low height rotatable camera it gave me bad fps around 5-7. Going into a corner of the map and as said having nothing serious in the view made this go up to like 10-12 fps (which wasn't that much different from a standard top-down not too much zoomed out view you'd have when actually playing). Moving to the maximized radar screen made my fps max out immediately to 60fps (vsync lock)...

Does Spring lack advanced culling algorithms or why do scenes with half the map and hundreds of units visible against those with virtually nothing than a textured plane and a skybox make so little difference performance wise while clearly not being CPU limited (at least I couldn't see any projectiles or particles not being simulated while in radar view - pausing the game there and going back to a normal camera view had everything visible as normal)?

[Beherith]

FYI: terrain is rendered in vertical strips. So if you see top right corner. Whole top 129 heightmap pixel wide strip is pushed to gpu.
Fixes under way in ROAM, but its going slowly. All help is appreciated.

[Master-Athmos]

Hmmm ok. I might try it again with very low terrain settings and see if it really is just about that...