Direct answer
The “80% rule” in Folia’s FAQ is a historical rough CPU-core/thread allocation guideline, not a claim that your CPU should always run at 80% usage. Leave headroom for plugins, GC, I/O and threads you do not explicitly configure.
How to check it
- Count physical CPU cores effectively available to the Folia process; VMs and containers may be quota-limited.
- List thread requirements: Netty, chunk system I/O, chunk generation workers, GC concurrent threads and tick-region threads.
- Measure actual usage during your server’s workload. A dense hub and distributed survival worlds behave very differently.
What to change
The FAQ suggests allocating tick threads after other workers, keeping total configured threads below roughly 80% of available cores as a starting point in its historical test. This is not a stable formula; tune rather than automatically setting threads to cores × 0.8.
See How to Size Folia Tick Threads for the related configuration workflow. Do not raise concurrency beyond the machine’s capacity simply because a setting permits it.
Software and version notes
Folia’s official FAQ describes the historical test and explicitly notes these estimates are imprecise. Values are not a universal service-level target and do not transfer unchanged to Paper or other forks.
Verify the result
Compare actual tick throughput, region distribution, CPU contention and chunk load time before and after any change. Revert when the change simply increases scheduling competition.