At first I thought the problem was that you were taking timestamps while other operations were happening, so I re-wrote your benchmark VI (attached below) to eliminate that possibility. But I still saw that Match Pattern was faster even when I made this change. I'm not sure why it's behaving differently now than when I benchmarked it in the past. Perhaps it has to do with your string size, or the fact that the match string is at the end...it may even have to do with the fact that I'm running a quad core machine now, whereas I was running a single core when I wrote that nugget 3 1/2 years ago.
Either way, thanks for pointing this out. I guess over time, even nuggets can become obsolete. 😉