Thanks for an interesting Review:) As being into the low power route it is always very nice to find a new champion. Why do you think this is, btw? Is it due to less features or is it something else, something ineherent with the ITX factor or something else?
This review gave rise to a few questions regarding your test-method, though.
1. The Power supply used on the testbench is an untested unit for which we have no idea about the the efficiency drop off as power decreases. Sure it is 80+, but that only means that at 20% power draw it is 80%+ efficient. This is a fair bit from 20% though. Even the AC draw is only 7.5%.
This means that more efficient mobos will be penalized compared to less efficient ones due to the efficiency of the power supply. 1W or 2W difference may not seem much at the 30W level, but relatively speaking, it is a rather lot. Specially if the system is meant to be powered up 24/7.
I can immediately see 2 solutions to this problem. The first is to get a powersupply that we know is efficient at these power draws, but that would mean that the new results would not be comparable with older tests.
The other solution is to make a very good mapping of the used powersupply at those load levels so that it is possible to convert the meassured AC into DC values. However, if the AC values would platuae (same AC value for different DC values because that efficiency drops at the same rate as relative power level), then a new power supply would be needed regardless.
2. From the article for the Zotac board.
We tested the board with the CPU at stock settings with Cool'n'Quiet enabled, and underclocked to 1.5GHz and undervolted to the minimum stable voltage (0.875V).
From the M3A78 Pro article
X2 4850e @ 1.5Ghz undervolted to 1.100V, Cool-N-Quiet disabled
In the M2N78 Pro article the underclocked/undervolted CPU was also at 1.1V as well as in the MA74GM-S2 review.
Is there a reason why you chose to go with the minimum stable voltage this time? But more importantly, why not do that every time? We are all running our CPU's somewhere at the efficiency curve anyways, not some arbitrary set point, like stock clock/voltage. And finding that one motherboard will let you undervolt better than another might sway the conclusions...
Apart from the above, there are a few differences in this review. In the Zotac review, CnQ was enabled, but at least in the M3A78 review quoted above, CnQ was disabled when testing underclock/volt. There has also been a few comparisons between chipsets at the underclocked level, but not in this one. However, due to the differences in how the CPU was configured, such a comparison would not have been informative.
As it stands, though, it was nice to see what is attainable in terms of powerlevel with a CPU at the efficiency curve. It gives a nice idea of how much heat one will have to plan for.