Direct answer

Paper automatically chooses chunk-generation worker count when chunk-system.worker-threads is negative (default -1 in the referenced Paper global config). Start with automatic mode, measure exploration performance and only override if there is a proven CPU scheduling benefit.

How to check it

  1. Read config/paper-global.yml rather than assuming defaults from a forum guide.
  2. Note actual physical cores, container quotas and other workloads sharing the host.
  3. Compare new world/chunk loading with an already pre-generated area; recording spark samples can help distinguish generation from I/O or tick bottlenecks.

What to change

The documented automatic worker calculation is limited by physical core count. Manually increasing workers can starve gameplay or GC; do not simultaneously change worker and I/O thread counts. For details, see Chunk Worker Threads.

# Paper global configuration:
chunk-system:
  worker-threads: -1  # leave automatic unless measured evidence exists

Software and version notes

The setting is described in PaperMC global configuration. Check your exact server release before using the same key on Folia or other forks.

Verify the result

Measure comparable exploration, CPU load and MSPT. If custom threads do not show repeatable gains, restore -1.

References