We can't distribute the Lynnfield testing utility, but we used to use CPU Burn-in, which did a pretty good job. (http://www.softpedia.com/get/System/Ben ... rnin.shtml
In HWMonitor I take the individual core readouts after an hour of idling and an hour of running the Lynnfield utility, respectively (after an hour the temps have stabilized) and average them together. So, for example, if after an hour of the Lynnfield utility running at 80% I see core temps of 60, 58, 55, and 57C, the "Burn" temperature there is 57.5C. I do the same thing after an hour of no activity. And that's how the magic happens.