The victimof:  Since people do their power calculations in different ways, to compare apples and apples, it's preferable to just report power at the wall divided by total number of cards.  That's all that matters anyway for determining efficiency.  By that measure, I've never seen a rig that pulls less than ~120W/card at a decent hash rate.  Za1n's 714W result is impressive, well done.  I find that believable, but I suspect Za1n is at the outer limit of reducing power without a substantial reduction in hashrate.

And none of this actually addresses the substantive point I made, i.e. that at a given power draw (say 1440W for a typical 110V circuit), it's more profitable to mine monero than eth.