Direct answer
Leaf documents a JVM flag -DLeaf.enableFMA in certain builds. Floating-point fused multiply-add behavior is CPU/JVM specific, and enabling a flag is not proof of a TPS improvement.
How to check it
- Check Leaf system-property documentation for the exact installed release and confirm CPU instruction support; benchmark a repeatable profile before and after one controlled startup-flag change.
- Record the original configuration or test conditions, capture a profile or a repeatable observation, and identify the affected world or player group.
- Make the measurement under the same player count and workload you will later use to verify the change.
What to change
Do not confuse FMA with adding Java heap or enabling multithreaded ticking. If your CPU or Java implementation does not benefit, leave the default unchanged.
Keep a copy of the old startup script so the flag can be removed immediately if behavior or performance regresses.
Software and version notes
Before using any example, confirm that your exact server build exposes this option and that a per-world override does not take precedence. Leaf examples from publicly available 1.21.1 documents are historical; do not assume the same keys or defaults apply to later builds.
This article is written for Leaf. Configuration keys and defaults can move between versions. The source material checked for this page includes Leaf system properties. If your server is Folia, Leaf, Purpur, Fabric, Forge, NeoForge, or another fork/loader, do not assume a Paper-only setting exists or behaves identically.
Verify the result
Repeat the same test with the same software build, player count and world activity. Compare the targeted symptom and any spark measurements, then check that intended game mechanics still work. If the change does not produce a reproducible improvement, revert to the recorded baseline.