Refund: R9 390 uses more power than a GTX 970

This whole rant is because the Micro Center website lists the MSI R9 390 as requiring a 500W PSU and I was too dumb to notice their mistake. The box clearly says 750W PSU required. My bad, sorry.

I am a Nvidia fanboy and I used to have an open mind regarding AMD. Not anymore...

I owned a R9 390 for 2 days before I stopped making excuses for the rotten experience and exchanged it for a GTX 970. I had my heart set on a used 980, I can't afford a 980Ti and the 970 seemed too close to my 2x SLI 660. When SLI functioned I was fine, but too often I was faced with a single 660 with tanked frame rates. I don't upgrade often and when I do, the performance specs have to be nearly double.

I asked @JokerProductions and he suggested the 970 over the 980. The used 980 was $350 but it had 2 8-pin connectors and I had to pass (I only have 6+8 on the PSU). Plus I was sketchy about buying a card on craigslist. Tom's Hardware recommended the R9 390 as the best value card for 1440p. When I checked it at the Micro Center website (I live 10 minutes from the main store in Ohio), I found that they had a deal on an open box R9 390 for only $288! I bought it at 10AM the next day.

At first I discovered that merely uninstalling the Nvidia drivers and utilities prior to switching teams is not enough. I used Guru3D Display Driver Unistaller, then I searched my C:/ drive for nvidia and cleaned the rest manually. The card was flaky, had lockups, weird Windows slowdowns, but no BSOD. I had more crashes in the past 2 days than I have had in a year. I did manage to work for a few hours using Photoshop and that went well. 3Dmark Firestrike ran no problem. More problems happened when I started to game.

Most games had to be re-installed or have the config files deleted to switch from Geforce to Radeon. My favorite game, Bulletstorm, would not work at all with AMD. When games did work, it was really nice. I played Just Cause 3 and DiRT 2 for a while. JC3 does not support SLI. Upgrading from 720p medium at 40 FPS to 60 FPS 1080p ultra looks pretty sweet. The problem was the excessive power draw. My Uninterruptible Power Supply battery backup started beeping a power event warning REALLY LOUD any time I played for more than a few minutes.

Long story short version:
Micro Center website lists the R9 390 required PSU = 500W.
The fine print on the MSI packaging says 750W PSU required.
My PSU is 700W but that was not the main problem, my 685VA 390W UPS was overloaded.
The GTX 970 requires a 500W PSU.

Test Bench:
Asus Z97-AR > i5 4690K 4.5Ghz @ 1.25v > 16GB DDR3 1800 > 2x SSD + 2x HDD

My UPS was reported that at idle the System + AMD was drawing 200W and the UPS maxes out at 390W.
Under load the R9 390 pushed my system to 425 Watts! The UPS was fully in the red and beeping.
I turned the GPU to silent mode and reset the motherboard BIOS to default (no o/c) 3.6Ghz @1.0v. I also tried removing every other device from the UPS but that had a minimal effect on power draw. Looking back, the only thing I did not test was plugging the PC right into the wall, but then I couldn't monitor it. I was able to reduce power draw by 50 Watts to 150W at idle (no o/c) and actually had some fun in.

The System + GTX 970 only sips 85W at idle (full o/c) and maxes out at 200W, same as where the R9 idles.
The R9 is a more powerful GPU, but the power inefficiency score is ridiculous ... let's face it.

In Firestrike the - R9 390 scores 10,437 @ 425W, the - GTX 970 scores 9,716 @ 200 Watts.
The R9 is 107% faster but uses 212% more power to do so. That's bad engineering.

Two days later with my motherboard tuned, the 970 used more power. 100W idle and 330W load, a little faster 9875 Firestrike. During the 2nd test the R9 is +6% benchmark score @ +95 Watts compared to my 970.

Compare the benchmark score on the left to the one on the right. The GTX 970 score of 9,716 is more than double the GTX 660 score of 4,363. Comparing the 660 SLI score to the new 970 GPU, 9716 / 7986 = 121% improvement. The step up in quality of look and feel is far more than the numbers can convey.

This is my system with the R9 390. The Geforce looks almost the same.

The R9 is a fine card. It just didn't make sense for me to spend another $300 on a new PSU and UPS in order to see if AMD will ever get their drivers sorted.

The Flip Side: MSI Geforce GTX 970 Gaming 4G.

I'm not just a Nvidia fanboy, I'm an EVGA fanboy. I have owned their cards and motherboards forever. I considered the EVGA SSC, but I liked the look of the MSI card better. After I exchanged the defective Radeon card at Micro Center and installing my new Geforce, I was up and running in 10 minutes. No crashes, no errors, no hiccups that I have to justify as newbie issues. Smooth like butter. I colored the LED on the MSI logo with orange tape. The Geforce is lighter, thinner (The R9 was ~2.5 slots), not as high and lacks a backplate :(

Oh yeah... I thought the coil whine on my GTX 660 was loud. The coil whine on the Radeon was way louder than the 660's and the 970 is much quieter than the 660's. I tried putting the insulated panel back on my be quiet! Silent Base 800, put the 1/4" thick tempered glass over the steel and the R9 390 was still too loud. Even the fans were loud. The same fans are silent on the 970 because they don't have to work as hard.

That's why Nvidia GPU's are the best. They just work.

1 Like

I'm going to bet that every issue you listed is the exact reason this card was in the Open Box discount bin.


Well that is an interesting story. The good point you made here is the power draw difference (amd has to step it up in that department). I only have a couple issues w/ your argument. First you moved from Nvidia over to AMD forcing you to uninstall drivers and the like (which you may or may not have gotten them all). Second, is it was an open box GPU as opposed to brand new (may or may not effect anything). To say they just work is a bit ignorant imo (especially saying that you're an Nvidia fanboy which leads me to believe this is your first AMD card in quite some time...). In order to prove the last statement you made (which is what has me most perturbed) you would need to use a much more controlled situation. If, for example, you were to say freshly build a PC wouldn't an AMD card just work as well? (assuming of course it's not DOA). To just arbitrarily say that Nvidia GPUs are the best is very short sighted... they are the best for you maybe, but, overall they aren't and neither is AMD. Both sides have their faults and that is why we watch channels like Tek Syndicate and the like... to learn what their experiences w/ the hardware is and to make our own judgement based on those opinions. Clearly you will stick w/ Nvidia but i am fine supporting the little guy to hopefully spur even more innovation and to stop the shenanigans that Nvida has been pulling over the last couple years.

(I apologize for the lack of paragraphing here)


The R9 390 at worst should only be using 40W more than a GTX 970. Something was fucky with that card and I'm going to put all my money on it being the ugly duckling of 390s. Again, discount bin.


What PSU are you using?

It seems odd a 700w PSU not having two 8 pins or at least 2 6+2 pins? Most quality 500w PSUs have 2x 6+2s :P

What version of windows are you running?

As a side-note... AMD's drivers are fine, I use both teams daily...

But happy that you're happy :)

1 Like

I was saying there is a difference... i didn't necessarily agree that the difference OP experienced was normal (its clearly not) but the point was made, its a valid point to an extent, and i acknowledged it.


Try and see it from my perspective, that of a happy 390x user that barely makes any noise whatsover (except for when benchmarking under unrealistic loads) and gives me very nice framerates in practically everything I play at 1440p.

If I had bought a discounted open box 970 and ranted and raved about how bad nvidia was I am sure you would have made the exact same argument as Fouquin did.

I am pretty sure I did a good job of cleaning out the Nvidia drivers. I checked regedit where I was shocked to find all the left over drivers, then I checked again after I cleaned them.

My PSU is old but solid. It is an OCZ Stealth Stream 2 700W non-modular.
Windows 10 Pro x64.

The card probably would have worked fine if I spent $100 on a new PSU and another $150 on a bigger battery backup. It is faster than the 970 and I'm sure it's a beast when overclocked. Bottom line is the 390 uses 225 Watts (measured) more than a 970. It's written on the box (+250W). My mistake was not verifying that Micro Center made a mistake by listing 500W on the website before paying for a GPU that had 750W printed in teeny letters.

This is my 3rd bad experience with AMD. I'm very happy with my new Geforce.

It only irks me a little when nVidia is idolized as the king of power efficiency, when in reality most of their cards are hardly more efficient than the competition. So the whole, "AMD needs to step up their game" just seems dismissive of the massive leaps and bounds they've made in efficiency. A Nano can outperform a respectably overclocked GTX 980 and use equal or less power in the process, for example.


A quick comparison on pcpartpicker shows that a sapphire r9 390 Nitro only uses 130W more than an MSI 970 twin frozr v... (275 vs 145)

That is a comparison of TDP, which is a thermal measurement not a power input measurement. It's saying, "This card produces ____ watts in measurable thermal output."

No the power draw between the R9 390 Gaming and GTX 970 Gaming is very close. Here's a nice chart from PCLab with a ton of cards tested:

And here's a quick cut showing exactly where MSI's GTX 970 Gaming sits:

It has the worst efficiency of any GTX 970 on the list. Now I'm not using this as any way to slander the OP's post, just pointing out that the R9 390 and GTX 970 are actually very close to each other in power efficiency across the third party vendors.


Yes i realize this now... my bad

I have a Masters in engineering and performed extensive testing and comparison on my system. My test showed 425 W for AMD and 200 W for Nvidia. About 225W difference, very close to the box claim of 250W. Your results may vary.

I'm not hating on AMD. I'm just saying for me I have confirmed the rumors are true.
AMD = Cheaper, Faster, Hotter, Thirstier. Nvidia = More Expensive.

As for who is better? I've made my choice.

You definitely don't need a 750w PSU for a 390... that's a bunch of garbage... a quality 500w PSU would handle that system... a terrible 700w PSU would...

Never used a OCZ PSU... usually stick to Seasonic or XFX for the Japanese caps, sometimes delve into EVGA, or Corsair for cheap semi-modulars... though both their high-end models are pretty nice...

But I'd put money on a combination of poor driver uninstall/reinstall, the GPU you bought being defective, or the PSU being garbage... it's got 44% 1-star reviews on Amazon :P

We're all glad you got an education in Electrical Mechanical Engineering... but yes Hawaii was a thirsty GPU... the newer Fiji line of AMD GPUs are more power-efficient than anything nVidia has on the market for enthusiast home systems... as if power efficiency on a home rig was a thing?

Like I said, I run both... I put R9 290's and 390X's in our engineering rigs at work and my home system runs EVGA 780's in SLI... I'm not biased... but your reasons for your choice were :)

Again... glad you're happy... all that matters

1 Like

That's probably 225W under a peak load, not too much during a gaming, load, gaming load difference should be about 100W at worst

Also you can potentially just undervolt your card and get it to be very effecient, on par with maxwell at times going from this recent test from tom's hardware undervolting a Fury card,4425.html

and like others said, it was an open box 390, that was probably having issues hence the return, and if it was a driver thing, reinstalling your OS would have been best

1 Like

The green team's drivers have been absolute dogshit lately, so the grass is not necessarily always greener on the other side.

Also, as others have pointed out, there's probably something wrong with that open box card.

The OCZ700SXS2 is a FSP unit, if anyone was interested. Professional reviews indicate that it's a decent budget power supply, but consumers on newegg report a high failure rate.

Well the powerdraw diffrence was allready obvious even before you jumped the AMD ship right?
Unless you realy didnt know about the powerdraw diffrence between a 970 and 390 of course.

1 Like

The power draw tests were during a gaming load. I use the Hitman benchmark and I also checked it with Just Cause 3 and Furmark. I honestly didn't even consider the AMD card until I read the article in Tom's. I was going with Joker's pick. When I checked Micro Center the AMD was $60 less and they both were listed as 500 Watts. I did not know the power difference between cards.

I'm not a noob and I still made a mistake picking components. I started my current rig on Thanksgiving 2014 and I have been carefully shopping for a new GPU for 15 months. The AMD card would have been a great choice if I got a new PSU also. However I most of the time I was nowhere near max Wattage and the experience was still very glitchy. I tried everything but reinstall Windows. I did re-install the test programs. Then I said "This is too much effort. I'm exchanging it in the morning."

Not one person mentioned the root cause of my problem. My 685VA battery backup is inadequate.
If my UPS didn't start beeping at 400W I would never have guessed that the glitches could be caused by power issues as much as the card or the drivers. I used to have a much better APC UPS. The second time I replaced the battery it stopped working. The new battery backup was barely adequate but it could not handle the Radeon. The R9 had 8 min run time and the Geforce 970 has 30 min run time.

Now that I have tested both, I think the R9 390 is a GTX 980 for the price of a 970.

Yeah the higherend Hawai / Grenada based gpu´s, like the 290 (X) / 390 (x) are pretty powerhungry yes.
Nvidia maxwell cards have a great performance per watt value definitely.

The UPS should basically function as a surge protector so long as you have power... some of them regulate power flow and fluctuations, but the battery shouldn't ever do anything so long as you're connected to a main... the battery in a UPS is irrelevant so long as you have power... It's not like the wall isn't providing enough power for the system and it needs to draw MORE from the battery to run :P

Regardless... if you had the slightest inclination there was any power issue at all, the first troubleshooting step would have been to remove the UPS variable and run it straight from the wall jack...

I kind of find it hard to believe the PSU was ever an issue (assuming it was working properly) being that I had 780's in SLI with a 4790k running on a Corsair TX750 for months... I also don't see why the UPS would affect it...