Mine was exceedingly slow when I had >2,000,000 files (gigabytes of versioned source code will do that). Several seconds to pull up an application by name. 10-30 seconds to pull up an infrequently-accessed document by filename. > 1 minute to pull up a PDF by content.

All of which caused a few seconds delay between every result-set returned, during which it was entirely unresponsive. It still worked faster - pressing `enter` would grab the top result when it came up - but that's unacceptably slow.