Big colonies make the game repeat the same work again and again: finding jobs for haulers, rebuilding routes, searching for trees, cleaning up memory. Late Game Performance removes that repeated work, so you get fewer lag spikes and freezes. It is built to change how quickly the game finds its answers, not what the answers are.
Played in a real late-game colony. In a save of about 350 beavers it gave noticeably fewer lag spikes (played as 0.4.8; 0.4.9 runs the same code). See exactly what is and isn't tested.
Every decision a hauler makes, every road you finish and every lumberjack's search for a tree makes the game redo calculations whose results mostly have not changed. In a small colony that costs nothing. In a colony of hundreds of beavers it adds up, and you feel it as the game stuttering.
This mod keeps the results the game has already worked out, does the heavy parts on other CPU cores, and skips lookups that could not change the answer. Beavers make the same choices they would have made anyway, just sooner.
It is aimed at big late-game colonies. A small colony has little repeated work to remove, so it is unlikely to notice much.
Every time a hauler looks for work, the game rebuilds and sorts a list of every hauling job in the district. The mod keeps each list until something that feeds it changes. It is the same list, in the same order, that the game would have built.
After you finish a path, stair or platform, the game has to rebuild the route map of every building with an entrance, and does it on the main thread. The mod spreads that work over up to 7 worker threads, and the maps come out exactly as the game would have built them.
To find a tree, the game measures the path to every marked tree, though most of them are still growing. The mod leaves out the measurements that cannot change the answer. Lumberjacks, gatherers and farmers still get the game's own result.
The game cleans up memory with everything stopped, which in a big colony can freeze it for most of a second, about once a minute. One tick-box in the mod's settings page switches on incremental collection, which does the same work a few milliseconds at a time.
Every speed-up is built to give the game's own answer, and nothing that affects the simulation is a setting, so every player on the same version runs the same code. Compatibility with multiplayer has been the top priority. Details
The mod writes a timing line to the game log: ticks per second, longest frame, time per system, memory clean-ups and saves. You can measure your own colony instead of taking anyone's word for it.
In multiplayer every player's game has to calculate exactly the same thing. A performance mod that changed even one answer would be worse than no mod, so keeping BeaverBuddies games in step has come before every speed-up.
It has been played in multiplayer on two computers, and tested extensively against the game's own code. The full list is below.
Late Game PerformancePlayer.log reads the same on every computer running the same version.Figures from the mod's own test harness and from play. None of them is a promise about your colony: the gain depends on its size and shape.
A model forest of 2,000 marked trees, 50 of them grown. Lower is better
| Case | Lookups |
|---|---|
| Game's own search | 2,000 |
| With the mod | 51 |
A test model of the game's search, not a live game. How much it saves in your colony depends on your forest: a lot when most marked trees are still growing, nothing when every marked tree is grown.
A 22,600-tile road network in the test harness. Lower is better
| Case | Time |
|---|---|
| One at a time | about 1.8 s |
| On 7 workers | about 0.23 s |
In the game the workers run alongside the tick, and the main thread still builds a map itself if it needs one before a worker has reached it. The harness road network is heavier than a real colony's.
One 35-minute multiplayer session, same save and same mods, on two computers
| Without incremental collection | With it | |
|---|---|---|
| Freezes from memory clean-up | 39, median 675 ms each (about 30 seconds in total) | 1 frame over 50 ms outside of saves |
| In play | Drastic spikes at high tick rates | Much steadier frame rate at high tick rates |
Two different computers in one session, so hardware counts as well as the setting. The memory in use was 1.6 to 2.7 GB on both.
Tested extensively, but it is still a young mod, so here is the honest picture. Try it on a copy of a save first.
Verified
Not yet verified