As mentioned, I get 200% to 600% execution timings when tracing a script; is it at all possible to reduce this overhead? Especially since I'm guessing it's roughly linear with the number of function calls, it makes profiling scripts with large number of funcalls difficult.

For instance, I have a script that runs under 400ms, but does in excess of 100K calls; this leads to a traced execution time of 2s. The functions at the bottom of the call tree (those that get called a lot) pull a lot of weight for being traced, although execution in a timed cycle shows they don't pulled nearly as much weight as the profile reports.