Geometry efficiency?

I'm wondering if a polygon takes as long to draw depending on whether it is partly occluded or not, or whether its size affects drawing time.

To give a really simple example: Suppose I have two identical cubes A & B in the same position, now I make cube B thinner, but also taller so it is protruding out of the top of cube A. Now, is it beneficial to also bring the bottom edges of cube B up to the top of cube A, i.e so there are no hidden parts of the polygon?

My thinking is that the end result looks the same, but because the side faces of B will now be smaller, Unity won't waste time drawing/filling unnecessary parts that wouldn't be visible anyway. But I hope I'm wrong... I'm making a lot of blocky architecture, and it's much easier to just let some blocks overlap than resize them. The poly count is the same but I'm worried those partly hidden faces might add to drawing time significantly.

As the polycount (and more importantly, the vertexcount) stays the same, it will make no difference to the vertexshader. But, in general, it will increase the rendertime for the pixelshader.

In your example, if blok a is rendered first, then B, it won't make much difference, as the part of B that is behind A will be culled. But if it's the other way, B will be drawn completely and then A will be rendered on top of it. But normally, you can't really decide in which order it will render, so the answer is, yes, it will take more time, but how much depends on the scene, the shaders and the computer.

(yeah, old question, but I just happened to stumble upon it)