But a lot of the time it doesn't matter. And it's easy-ish to go from calling update on everything to moving objects in or out of a "currently visible or at le

But a lot of the time it doesn't matter. And it's easy-ish to go from calling update on everything to moving objects in or out of a "currently visible or at least close enough that we bother simulating them and may need updates" bucket (which might be what's on screen, or what's in a room, or in a level, or within X amount of distance, or any number of other things) at whatever granularity you need to get the performance you want. The general advice of splitting the updates into frame-sized chunks solves the important beginner issue, and the performance tweak of reducing the number of objects you update is something you'll likely encounter much later. E.g. I'm currently doing a simple platformer with my son for fun, and just a very crude "is any part of this objects default starting position within x distance from what is currently on screen" is more than sufficient to constrain the update. Before that I just had it even naively render every single object in a level, because it was fast enough, so I didn't bother filtering anything. I only needed to start thinking about performance in the first place because of a combination of me going over-board with various effects and a fully waveform-based software synth and this all being written in Ruby (using the DragonRuby game toolkit [1]). If it'd been written in a faster, compiled language, or I'd been more reasonable in my expectations we'd likely never have needed to think about it. But 10+ parallax layers, wave-formed bases synthesis, and a bunch of pixel effects might be nothing in C, but keeping 60 FPS in Ruby that way takes a little bit of care. That said, we got to 1000+ objects updated every frame before I needed any filtering at all. I'm sure we'll eventually do more - not all objects have any animations or move, so they'd be easy to filter out too, but then I have to ensure every type of object is categorized accordingly, so I won't bother doing that until there's a need to, and it takes calling an update or render method that does nothing for a lot of objects before it starts dragging down the frame rate. [1] https://dragonruby.itch.io/dragonruby-gtk https://dragonruby.itch.io/dragonruby-gtk