In testing hardware and/or software, processor events can be logged during
a test. Using the events, we can determine an amount of processor
activity not used to execute software considered to be involved in the
test. Such noise may then be accounted for in determining and
prioritizing regressions, as well as to improve test results generally.