Direct answer

An idle furnace array is not equivalent to an active one. Container transfers, fuel handling, item transport and any plugin events can dominate the cost.

How to check it

  1. Test the array with input items and fuel flowing, recording how many furnaces and hoppers are loaded at once.
  2. Collect /spark profiler start --timeout 120 while the active array causes the suspected slowdown.
  3. Pause item insertion briefly to separate smelting work from hopper inventory checks and collection systems.

Example: A 128-furnace line that is idle most of the day should be compared with the same line at full input and output, rather than treating all placed furnaces as continuously expensive.

What to change

Reduce redundant hopper connections, unnecessary clocked item feeders and always-on sorter lanes. Measure items processed per minute alongside MSPT to avoid trading performance for broken throughput.

Software and version notes

Do not globally disable Paper hopper move events to optimize a single furnace array; doing so can bypass plugin handling. Farm rates and hopper timing can be affected by Spigot-specific settings.

Verify the result

Compare equal batches of smelted items before and after; check output collection, overflow handling and permissions.

References