That review looks about as bogus as they get.
The reviewer actually provides graphs making it clear that he didn't allow time for temperatures to stabilize before recording the data.
There is no mention of ambient temperature at the time of the different tests.
Saddest of all is this graph:
The power dissipation when the CPU is operating at 2.15V should be 65% higher than when the CPU is operating at 1.75V. The temperature differential between idle and load should be 65% higher for the 2.15V case than the 1.75V case. Yet as best I can tell eyeballing that graph, the idle/load temperature differentials are equal at the two voltage settings. I'm skeptical that this test was actually done.