I find it bizzare an 100+ watt difference exists between w2k8 and linux or even linux vs linux on certain test machines.

There is obviously something not right with the tests or a correctable software/firmware flaw has been invoked rendering the tests worthless.

All platforms do much the same thing. Intel is largly responsible for the throttle codes of their CPUs in both Windows and Linux platforms.

I wonder if they concidered environmental differences that would effect the energy consumed by variable speed fans?