It turns out that just turning Dstrace on causes a large performance slow down. Even at level 1. It is the act of converting from efficient memory structures to human readable XML that really eats the CPU cycles.
Then throw on writing out to disk, and there are more cycles being burned. While multi CPU may help it is not clear that a single process thread can span multiple CPU's and I am not sure how multithreaded IDM is. Also, most of IDM is running in a JVM, so can a single JVM use multiple CPU's? Dunno.
My examples from the article were actually on Linux (SLES 10 SP1, eDir 8.8 SP1).
95% of the time, the hit is perfectly acceptable, but when you get hit that small part of the time with large numbers of updates, the slow down is a pain.