I appreciate the test and timings!
First: did you know there is a high quality option which you can turn off? I’ve run tests with tens of thousands of labs and still come in under 16ms per game update (with high quality disabled).
Regarding further optimization: mods are limited to single-threaded execution on the CPU, and it’s all in Lua which is interpreted. As far as I know, I’ve addressed the low-hanging fruit: reducing work done per frame, distributing work across frames, and the big one: minimizing per-frame memory allocation. You are more than welcome to play around with the code and see if you can find more!