Frankly if you’re doing something sensible and you know how to use the tools at your disposal [properly] - the engine is practically perfect. That is, as perfect as it can get with hundreds of people contributing code to it’s enormous code-base, over a period of 15+ years. Since the Engine is so easily upgradeable, we probably won’t see an Unreal Engine 5 for a very very long time. UE4 is going to have a very long lifespan.
Multi-Threading isn’t necessarily always a good thing, and it won’t always help you either. Probably one of the biggest issues of Multi-Threading is that you are always going to be capped by the slowest task. If you don’t do it 100% right, you’ll have times where cores are quite literally doing nothing while they wait for another one to catch up. It doesn’t take a genius to work out that’s going to cost you more performance than it’ll give you. A lot of stuff isn’t thread-safe either. It would be bloody hard to build navmesh on one core while the other is trying to process AI for example.
In future years, I almost garauntee you that we’ll see more parallel programming moved onto the GPU instead. Some Frustrum culling is now done on the GPU in Unreal like the new Distance-Field stuff, and compute shaders have been around for a while now too.
People like to talk about Multi-Threading without really having the foggiest idea what they’re talking about. Try building something complex and Multi-Threaded of your own. Compare performance before and after. I bet you’ll **** it up.
With 30+ games going to E3 built on Unreal Engine this year alone, and plenty more being made that are both open and small-world, I think the Engine has you beat to be honest.
ARK has sold incredibly well, so it’s easy to hate them for something. Id never heard of their studio before, and now they’re making multi-million dollar profits. They’re clearly doing something right.