>> During dev, a game requires much more power because it hasn't been optimized yet, and you may have any number of quick and dirty hacks to get things done.
i always try to keep the framerate at an acceptable speed at all times on the target PC. so i optimize as i go along, as needed, when needed. in Caveman i only really had to deal with pairwise range checks for target section from hundreds of entities in close proximity (round robin to the rescue), and writing a data oriented render queue. as the game grows, you get a feel for which bits are probably eventually going to need optimization. when that framerate starts dropping, you take a time out, fix the problem, then move on. this tends to reduce or eliminate last minute optimization during crunch time, allowing you to concentrate on getting as many of those final features in as possible before going gold. right now, realtime collision map and terrain chunk generation are the only two things that look like they may need some attention before all is said and done. and they're not quit bad enough to make me do anything about it yet. so its still wait and see.