Multiple cores

I am not a dev but afaik, the game is written in two codes, C++ and LUA. The engine is written in C++ and can be multi threaded. LUA on other hand is a single thread code so the bottleneck is this. My guess is to further multi thread the game, you have to move CPU-intensive LUA methods to C++, multi-thread them and expose these methods to LUA. Well, at least that what make sense to me, LUA usually is a mean for modders to mod the game, not use as base to build the game upon. Doing things on the engine side can be complicated and time consuming, evidently Zero-K doesn't have people who can afford to do that, otherwise the engine would have been very different from SpringRTS already.

I've been trying to add multithreading to several of my projects in the past.
It takes a huge amount of performance testing to figure out what helps and what hurts.

The main issue for ZK (AFAICT) is how advanced the game simulation is.
ZK is basically doing what sandbox FPS games do for 10 objects, but have thousands of objects instead.
Thus, it's kinda hard to compare the performance to other RTS games, or even FPS games.

For LUA, if someone can point out where the bottleneck is (with profiling), then I'm sure it wouldn't be too hard to rewrite that part. [Spoiler]

Sounds like its worth a conversation with BAR people about everything mentioned here.

quote: is there any plans to look at Coil as an option for ZK?

We never really left. Engine dev essentially moved to the BAR fork. ZK was on the last usable Spring versions up until the end, and after a brief break while the BAR fork was a bit unstable, we switched onto the BAR fork. We've been on the same version as BAR for a while now, at least a year I think. I recall mentioning this in release notes, but nothing as glossy as their multithreading hype page.