To mitigate the problem of running software impacting test results, I've been running PT in a "pre-installation environment" (PE) via a tool called BartPE? It creates a bootable CD that actually boots a working O/S, under which programs of the users' choice can br executed. Here's a link. http://www.nu2.nu/pebuilder/
PT 5 works under it just fine, I have incorporated it into my personal BartPE build. The difference in numbers is most interesting.
The jist of it is that memory throughput and raw calc ops were all significantly higher under BartPE (based on XP) than they were under XP normal, between 40% and 70% better.
Of course without proper drivers, graphics performance was way down, and unless I feed it a floppy while booting (the old F6 to specify storage drivers thing) my RAID volumes are off-limits.
Regardless, using this as a way to quantify parasitic drag imposed on a system by all of its installed software, I can imagine none better. It also levels the field when comparing baselines between machines.
In general it adds considerable depth to your results; if you haven't checked it out, you may want to take a look.
(btw, I tried to follow your documented method of applying a license to removable media. It didn't seem to work, it nags when I run it from BartPE, but it doesn't refuse to run, so I guess a non-issue.