Refund: R9 390 uses more power than a GTX 970

Well this fine. Yet another post, noise, noise, noise, nVidia is best. That's cool looks like there was a massive failing of many thign here. Including basic I formation.

So I built a PC for a friend of mine 2 years ago, with an MSI r9 290 (basically same card as the 390) and a 480w high quality Be Quiet PSU (E9-CM 480w). I never heard any complaint from him. So what you describe is not normal and there is a problem with the card.

I guess you should get a better PSU, would definitely be safer even when you continue to use the 970.
And the difference in power draw should be because of a faulty card, I guess they could've given you a refund at Microcenter.

Something that nobody as mentioned is that VA =////= watts.
You're taking one big brainfart if you think a 650VA UPS is going to play nicely with a high end computer, the computer is a capacitive load which stuff rated in VA really don't like.
You can probably cut the VA rating in half for such a set up, my 1000VA transformer didn't like the 500w computer load I was putting on it for testing, got pretty hot rather fast.

Lesson well learned. Don't buy open box unless you know the reason why people send it back? Sometimes it's a impatient buyer who bought 2 cards, sometimes it's an OC'er who broke the crap out of the card, haha.

One thing you fail to menion is your PSU brand/model. Keep in mind the 12V rail's loading capacity. For example the starting amps is 83, and your PSU can only provide 70amps etc..

I guess the card needed to burn in and now my motherboard is tuned and o/c'd again. Now I am getting 100W at idle (up from 85W) and 330W w/ the 970 under load. 425W - 330 = 95W difference, much less than 225W initially reported but still significant. My readings are not accurate, but I think the differences between them can have some meaning. I also discovered my 24" monitor uses about 50W.

R9 390 +108% score for 129% more power.

Don't forget about DX12 though, current Nvidia cards seem to be struggling a lot with it. which is another reason I stick to recommending AMD cards for the most part.

Also you could have tried undervolting, dunno if that's been mentioned already

I don't usually mess around with overclocking. I did use the MSI Gaming app to switch to Silent Mode. That reduced the core clock and the power a little.

undervolting is keeping the same performance but gaining power efficieny, it worked well on a Fury card here, but not all cards can go so low on the voltage, it made it as efficient as a 980, basically turned it into a nano

http://www.tomshardware.com/reviews/msi-afterburner-undervolt-radeon-r9-fury,4425.html

For what it's worth, in my experience, a UPS can have significant impact on performance. I tried to run my X99 rig off of one that easily supported a previous system and it began beeping in under a minute of gaming. Now, I'm no expert, but I believe some if not all models of UPS (certainly mine) it in fact does depend on the battery in some way even while you're plugged in. If I plug into my UPS right now, with a full charge and in perfect order, and load up a benchmark and let the system run, it will shut off in a matter of minutes.

This may have nothing to do with your specific issue, but it may have played some small role, who knows. Just wanted to mention this. My wife's Haswell i5/970 rig can run on the UPS 24/7 with no issues, because it's rated for it, but my rig pulls double the power, and although it's 'supposedly' within spec, the quality of my UPS is obviously extremely suspect.

Also, both of my systems report slightly different power consumption figures when on the UPS as compared with being off the UPS, very tiny amount, but it's there. At this point I'm very curious what impact a UPS can have on power delivery, both efficiency and also how 'clean' the power is, with a mediocre UPS, and a very high end UPS. Gonna have to go do some research because I don't have $1,500 lying around for a UPS that can support my rig :/