I don't have my Mac with me (it's not the computer I regularly use anymore), so I can't test it right now, but if I recall correctly it was about 2-3x (on a Macbook Pro's internal SSD, where the bottleneck for things like this is not actually I/O but kernel filesystem code CPU time).
Increase scan performance by moving enumeration to seperate thread and switching APIs from NSDirectoryEnumerator to fts_read
Use fts_read instead of NSDirectoryEnumerator
Threaded implementation: ugly but it works and it’s fast
Split enumerator into its own file
Fix skipping and filters
memery manglement
Error handling