Wikipedia says that overhead cost in business means on going cost.
Does it mean the same thing in terms of processing power?
(Fyi, this is the passage that I was reading that raised this question in my mind:
The performance gained from using extremely low-poly meshes (like under 500 polygons) is minimal if at all there. The majority of graphics cards have hardware transform and lighting, which means they can process ridiculous amounts of polygons per second. Additionally there is overhead for submitting a mesh to the graphics card to render, so being really thrifty on polygons is probably only making your game look blocky.)
“Overhead” has a lot of subtly different meanings depending on the context.
The line you’re quoting from the documentation is a very unfortunate reality: draw calls.
Consider the following hypothetical situation:
You have a spoon mesh. It has a metal material for the tip, and a plastic material for the handle. When this is in your scene and visible to a camera it will cost you 2 draw calls. The spoon geometry is relatively simple, so this cost is negligable.
Suddenly you have a flash of inspiration and decide you want spoons everywhere. You duplicate the spoon instance so there are now 1000 spoons in the scene. This is going to cost you a minimum of 2000 draw calls. Things are starting to slow down and your game isn’t even doing anything complicated. Hell, the spoons are made of really simple geometry.
The problem here is the overhead described in the documentation. There is a minimum cost of rendering an object, regardless of complexity.
A solution to the above problem is to combine the geometry. You will end up with a single, complex mesh representing your 1000 spoons. However, you’re only eating 2 draw calls (for the 2 different materials). You’d be surprised to see how much faster this is than drawing the 1000 spoons separately.
Overhead typically refers to cost that is there no matter what you do. There is a certain cost to running the Unity engine, no matter how simple your scene is. There is a certain cost to using PhysX. This cost is about the same weather you use one RigidBody or 64K.
Draw calls are a common example. Its cheaper to render a bunch of different items on one big texture then it is to use a dozen smaller textures. The total data use is the same. But there is overhead in loading each individual texture. This is where batching comes in.
In game programming terms cost commonly refers to the effect on performance doing a certain task has. Cost is often dependent on where your constraints are, common things referred to as cost include CPU time, GPU time, draw calls, and memory