When many players and mobs gather, the server has more to work out on every tick. If one thread handles the whole world, a single busy spot can slow everyone down.
How Divine works
A plain-language look at what Divine does when a server gets busy. It describes goals and design, not guarantees.
Divine splits the world into areas that run side by side, an approach that comes from Folia. In a crowded area, it also hands safe calculations, like working out which players can see each other, to spare CPU cores, then applies the results in a fixed order.
The aim: a crowd in one place has less effect on players far away, and busy areas stay smoother. The game rules themselves don’t change.
Early test results
Automated tests comparing a development build of Divine with Folia. Lower tick time means more room before lag.
250 players standing in one spot
About 37% lower in all 3 runs150 players moving around one spot
About the same. Results varied between runs.50 players with 300 mobs nearby
About the sameTest conditions
- Tested on 26 September 2026 on one Windows PC, with Java 25 and 4 GB of memory for each server.
- Players were automated test bots. Each test ran 3 times, with a 20-second warm-up and 45 seconds of measurement, in rotating order.
- Compared with Folia 26.2 (build 7) on the same world and settings. Divine was a development build, not a release.
- Tick time is the median time the busiest area of the world took per tick.
- The percentage compares each run with its paired Folia run, then takes the median.
Show the numbers as a table
| Test | Folia (ms) | Divine (ms) | Median difference | Same direction |
|---|---|---|---|---|
| 250 players standing in one spot | 5.45 | 3.45 | −37.4% | 3 of 3 runs |
| 150 players moving around one spot | 7.55 | 7.05 | −6.6% | Mixed |
| 50 players with 300 mobs nearby | 2.93 | 3.03 | −1.0% | Mixed |
What these results don’t show
These were short tests on one computer with test bots, so real servers will behave differently. Longer release tests haven’t been run yet.