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.