Well first of all the 390 does not idle like a champ, you never read reviews? It's one of the worst "new" graphics cards in idling (and multi monitor is a lot worse, and I guess you're using that too). Second SB-E isn't so bad at power consumption, only with too high overclocks, FX is a damn whole lot worse. Well, whatever, you can do what you want, I don't think 390 Crossfire is a good idea. I'd rather save the money for a new GPU to replace the 390 in a few months or next year.
First of all, if your defense of that statement comes from "because reviews said so," and I'm speaking from my own personally testing, good luck. I'm more than happy to explain myself beyond "because something on the internet says so."
From my own testing using the Kill-A-Watt built into my UPS is that the 390 at idle does not use more power if you use 1, 2, or 3 displays. I did, however, notice that occasionally when doing 2D things like moving windows or doing something that substantially alters the screen that memory would occasionally clock up from the 150Mhz idle to a full stock 1500Mhz. This will increase draw from the wall by up to 50 watts on my machine, even if the GPU core is still at idle. I suspect that has something to do with making sure that certain 2D workloads don't cause visual corruption due to memory bandwidth being too low at idle clocks when driving multiple displays. Either way, while I trust W1zz to do good reviews, I suspect what he's doing is taking the highest number that gets recorded to indicate idle usage which isn't necessarily an accurate assessment of the minimum, average or, median draw at idle. Just like when a GPU is fully loaded, W1zz gives two numbers, peak usage and average usage. I would highly suggest that he do the same with idle consumption numbers as one number just simply doesn't explain the entire picture.
With said said, with two 6870s my idle from the wall was basically 200 watts regardless of what I was doing in 2d, removing one 6870 brought it to 175, and replacing the last 6870 with the 390 brought it to 150 except in some situations where it will spike to 200 and come right back down to 150, so yeah, I would call that an upgrade for me but, if memory clocks up that usage easily will read out 200 watts for very brief moments but, doesn't reflect what the draw is most of the time. So yes, the 390 does idle better you think it does and no, reviews don't offer a complete picture of exactly what's going on at idle. To get that we would at least need a "peak idle consumption" along with an "average idle consumption" to minimally describe the distribution. From my own experiences at idle, memory stays clocked down far more often that it stays clocked up.
This isn't to say that the 390 can't suck down power. It most definitely can and it will consume more than both 6870s combined when overclocked when I overclock the 390 and at stock it's super close to the same draw.
My point is that power consumption numbers at idle are pretty limited and there are no numbers for load that's anything but full tilt. The problem is that a lot of time, my GPU runs at idle or at partial load so it is my opinion that current numbers with regard to power consumption are only valuable if there are enough numbers to describe consumption over time which very few reviews seem to provide. W1zz does provide that by giving average and peak load but, nothing exists for idle when I think it should for the same reasons it's included for loaded values.
