!!!
1.) Im sorry I didnt try out my test level thoroughly, but I was using entities that where alpha dependent on camera distance. Thats probobly why I was reading alot of entities behind a wall, which is why I asked this question in the first place. When I did try non-transparent entities, you were exactly right! The visible entities count dropped when I was behind the wall.
However I did notice something... While I was behind the wall, and the entities were being culled, I noticed that the SRV was 2.0. But when I moved into view of the models (they were unnassigned flapping birds), the SRV went down to 1.0. So from this information is it safe to say that for this isolated incident that the engine used more CPU in hiding the entities rather than showing them? I know there MANY more factors in factorizing this, but is that a general conclusion for this scenario?
Well, its good to know that entities will be culled. Because of this I hope to start a little tutorial on culling in large open areas...
2.) So its generally safe to say that a hilly and mountainous terrain is much more efficient than a reletivly flat terrain.