R9 290x Crossfire Review - performance, cooling, noise, energy [WIP]

[Tacitus]

Good evening everyone.

 

This is going to be my review of the 290x in single and crossfire mode, I have had these for about a week now. It is my intention to eventually play at 5760x1080p but as I only have two monitors right now, to see how these perform I'll be playing at 3200x1800 using virtual super resolution. I would try 4k, but for some reason my monitor doesn't seem to want to do that.

 

So, as of now, my system is as follows:

 

• i5 4670k @ stock clocks
• 2 X R9 290x PCS+ @ 1050 core, 1350 memory.
• 8GB G.Skill Ripjaws 2133mhz RAM
• Gigabyte Z97-X SLI motherboard
• Crucial MX100 256GB SSD
• Corsair Air 540, with stock fans
• EVGA SuperNova NEX 750B PSU - 750W

So, about a fortnight ago I returned my GTX 970 as being incorrectly advertised. My motives were that I could not afford a second one for SLI, not because I felt I wasn't getting what I paid for. In fact, I loved the card, and still recommend it to people frequently, but for an extra £100 I could get two 290xs, so I did:

 

 

These are two PowerColour R9 290x PCS+ models. My initial impression was very good, the cards feel well built and sturdy, the shroud is entirely metal (aluminium, I presume) the backplate is a nice addition, and the cooler look substantial.

 

 

Now, as some of you may be aware, these are two-and-a-half slot cards. This concerned me slightly as there is only a one slot gap between my cards when they are placed into my system. As a result, the gap between my cards is very narrow, and I worried the top card would be choked, overheat and throttle. I'm using a Corsair Air 540, however, so airflow should be good.

 

 

The single card experience: 

 

The single card experience has been excellent, at 1080p across a range of games which I have played with the card (mostly so far AC:IV, Dark Souls II, The Witcher 2, Dragon Age: Inquisition, and Total War: Attila) has, for all intents and purposes, performed just as my 970 did. Framerate was consistently high, no driver issues were recorded (something which worried me as the latest drivers are from last December), and temperatures were low. The card comes with an aggressive fan curve, so it was quite noisy initially, but I could lower the fan speed significantly (to around 40%) and get virtual silence, with temperatures around 70 degrees.

 

I failed to save a picture, but power consumption from the wall never went above 400W.

 

Crossfire experience:

 

The experience, I'm pleased to say has been overwhelmingly positive.

 

Performance:

 

In every game I've played (except Attila) Crossfire has worked a charm, I've experienced no stuttering, and framerates have been superb. To see how well it worked, I played each game for a while with and without crossfire at 3200X1800. No game above achieved 60FPS average, never mind minimum with one card. With two, Dark Souls 2 and the Witcher 2 achieved 60 FPS minimum, with all settings (minus ubersampling on TW2) maxed. AC:IV experienced a few drops into the 50s, but these were few and far between and did not hamper the experience. Dragon Age with one card at 3200x1800 was unplayable, averages were in the 20s, with drops down to the teens. Crossfire  delivered a fluid experience with averages in the 40s, and minimums above 30. When the game did periodically slow down it was not to the extent that it was noticeable. This is easily the most demanding game I played, and Crossfire justifies itself at these high resolutions. I could have dropped AA down to 2X, but instead kept it maxed out. This would have boosted framerates I predict into the 50s (average). Total War Attila was another story, and crossfire appears to not work at all for me there.

 

I did experience one or two issues with my crossfire setup in performance terms however. While TW2 looked beautiful, my computer froze during certain cut-scenes, though never in game. The only way to progress was to disable crossfire, get beyond the part I was at, and then re-enable it. Enabling and disabling crossfire also shifts all of my desktop icons over to the right hand side of my right monitor, which is really annoying.

 

Temperatures, Noise and Power Consumption:

 

Is the 290x in crossfire a deafening, icecap melting solution which needs its own nuclear reactor to run? In short, no.

 

Dragon Age pushed the cards the hardest by far, and by setting the fans to 70% I got top-card temperatures of around 82 degrees, with the bottom staying in the 60s. 65% fan speed seems too little, and takes the top card close to 90 degrees, which I'd rather not head toward. Considering how close the cards are together, this is impressive in my mind, as at 70% the fans are certainly audible, but if you use speakers at a moderate volume or headphones you will not hear the system. I'd heard some horror stories (including from the AMD rep at OCUK) of overheating 290xs in my case because the heat just can't be shifted, but PowerColour's PCS+ cooler does a fantastic job at keeping temperatures in line. I imagine that with better motherboard layout temperatures would be even lower.

 

Now for the big one, power consumption. Everyone said I was cutting it fine with a 750W PSU, that I need 850W minimum and ideally 1000W. Well, the truth is, at 1100 on the core, I never saw my PC draw 700W from the wall. The highest I saw was in the 690s, but frankly, considering PSU efficiency, the system as a whole must be drawing closer to 600W. I'm not in the habit of running artificial benchmarks, but in a gaming scenario which truly stressed the cards, 750W is plenty. I never got a photo of it at peak wattage, but here is one near to it:

 

 

Concluding remarks:

 

This thread is labelled WIP for a reason: I've done limited testing, most of which is subjective. I would like in the future if there is demand for it to quantify the data in a better structured fashion, and to include more recent games as and when they release (TW3, GTA V, spring to mind).

 

Do I regret swapping out my 970 for 290x? No, in short.

Would I advocate a 290x over a 970? Not necessarily, the two are so close in performance I'd advise prospective buyers get the 290x if it is noticeably cheaper, as it was in my case.

 

Crossfire for me has worked extremely well, with a few minor exceptions. I imagine 970 SLI is quieter and less power consuming, but at the high resolutions I aspire to, the 290x seems to pull ahead somewhat. 

 

I'm happy with the performance I have gotten for £400, which is cheaper than what a single 980 costs here.

 

If anybody has any questions, comments, or feedback, please do feel to share them and I'll do my best to answer them or take on board the criticism for any future reviews I do.

 

Thanks for reading my first LTT review!

 

 

[Aniallation]

IMG_20150228_172310.jpg

[Decon]

-snip-

+1

[Meladath]

Not bad. Glad you found out the power requirements. Hopefully more people will read this and FINALLY realise that hawaii reallyy does not draw anywhere NEAR as much power as people like to make out. The 780Ti usually drew like 10-20w less than the R9 290X, which no one seemed to care about, but 290/X? YOU NEED A NUCLEAR REACTOR. Silly fanboys, oh well, rant over.

 

Looks like you have had a good experience! AMD's multi monitor support is far better than nvidia's imo, you should upgrade this thread with those results when you can.

[Enderman]

70% fan speed is A LOT considering that cooler has three fans that run at over 3000rpm max speed.

Maybe you use headphones, or maybe you just have a lot of ambient noise, but I can tell you right now that those cards are NOT quiet even at sub-50% fan speed.

 

six 80mm fans at 1500rpm? lol

[Tacitus]

Not bad. Glad you found out the power requirements. Hopefully more people will read this and FINALLY realise that hawaii reallyy does not draw anywhere NEAR as much power as people like to make out. The 780Ti usually drew like 10-20w less than the R9 290X, which no one seemed to care about, but 290/X? YOU NEED A NUCLEAR REACTOR. Silly fanboys, oh well, rant over.

 

Looks like you have had a good experience! AMD's multi monitor support is far better than nvidia's imo, you should upgrade this thread with those results when you can.

 

Thanks, and yes, it seems a few Furmark tests turned people off of the 290x. It's really not that bad.

 

When I get my third monitor (probably in April) I'll update this to include actual performance (though pixel count of 3200x1800 and 5760x1080 is very close), and experience with eyefinity!

[The Pizza Thief]

-snip-

Thanks! Nice review

[Atmos]

Not bad. Glad you found out the power requirements. Hopefully more people will read this and FINALLY realise that hawaii reallyy does not draw anywhere NEAR as much power as people like to make out. The 780Ti usually drew like 10-20w less than the R9 290X, which no one seemed to care about, but 290/X? YOU NEED A NUCLEAR REACTOR. Silly fanboys, oh well, rant over.

 

Looks like you have had a good experience! AMD's multi monitor support is far better than nvidia's imo, you should upgrade this thread with those results when you can.

 

Well, as far as nvidias newest generation of cards, the Hawaii core is more like a snappy lambo, with 4 wheel drive, v12, and turbo. It's fast, and efficient compared to the old school gt's of last generation. But compared to the half electric half petrol engine in the newer maseratis, well it looks like a brutish gas guzzling behemoth.

 

In short, it's not as bad as it could be, but compared to the competition it does use quite a bit of power.

[Tacitus]

70% fan speed is A LOT considering that cooler has three fans that run at over 3000rpm max speed.

Maybe you use headphones, or maybe you just have a lot of ambient noise, but I can tell you right now that those cards are NOT quiet even at sub-50% fan speed.

 

six 80mm fans at 1500rpm? lol

 

I use a pair of speakers, not headphones. Even in something like DS2 which doesn't really have intrusive and constant music playing it really wasn't that loud. You might look at charts and graphs, and YouTube videos where people shove their microphone right up against the card, but in all honesty in my case on my floor, gaming at 70% fan speed, the noise is largely unobtrusive. 

[Tacitus]

Well, as far as nvidias newest generation of cards, the Hawaii core is more like a snappy lambo, with 4 wheel drive, v12, and turbo. It's fast, and efficient compared to the old school gt's of last generation. But compared to the half electric half petrol engine in the newer maseratis, well it looks like a brutish gas guzzling behemoth.

 

In short, it's not as bad as it could be, but compared to the competition it does use quite a bit of power.

 

I agree with this to some extent. I unfortunately never took any photos, but actually, the 970 does use more power than people think it does. There's some in-built and universally accepted belief that the 970 could be powered by a hamster on a wheel, but the card certainly uses more than the 165W TDP. In reality I'd say overall the system used about 10-15% less power with 1 970 than with 1 290x. Not that big a deal.

[Tacitus]

Thanks! Nice review

 

Thanks

[Atmos]

I agree with this to some extent. I unfortunately never took any photos, but actually, the 970 does use more power than people think it does. There's some in-built and universally accepted belief that the 970 could be powered by a hamster on a wheel, but the card certainly uses more than the 165W TDP. In reality I'd say overall the system used about 10-15% less power with 1 970 than with 1 290x. Not that big a deal.

 

Oh yea, every card on earth draws a lot more power than on the box when at full load. The 970 and 980 are absolutely no exceptions. Sure, under normal load it uses half the power of a 780, and has a bit more snap, but under full load the difference is far, less impressive.

[Meladath]

I agree with this to some extent. I unfortunately never took any photos, but actually, the 970 does use more power than people think it does. There's some in-built and universally accepted belief that the 970 could be powered by a hamster on a wheel, but the card certainly uses more than the 165W TDP. In reality I'd say overall the system used about 10-15% less power with 1 970 than with 1 290x. Not that big a deal.

 

This so much. The nvidia TDP is hilariously unrealistic, ESPECIALLY when you overclock. (The problem here lies both with what TDP is, and how nvidia knows they can manipulate this to fool people into thinking things). Most reviews seem to show around 180-200W power usage on a 970 and 200-220W on a 980. This is ofc better than the 250-280 the 780Ti/290X used, but its so far from the 165W figure they state its not even funny.

[Faceman]

NZXT G10 + H55 for the top card.

[Tacitus]

Oh yea, every card on earth draws a lot more power than on the box when at full load. The 970 and 980 are absolutely no exceptions. Sure, under normal load it uses half the power of a 780, and has a bit more snap, but under full load the difference is far, less impressive.

 

 

This so much. The nvidia TDP is hilariously unrealistic, ESPECIALLY when you overclock. (The problem here lies both with what TDP is, and how nvidia knows they can manipulate this to fool people into thinking things). Most reviews seem to show around 180-200W power usage on a 970 and 200-220W on a 980. This is ofc better than the 250-280 the 780Ti/290X used, but its so far from the 165W figure they state its not even funny.

 

The issue is certainly made worse by people overestimating how much power a 290x uses. I think some people out there legitimately believe it uses almost twice as much power as a 970.

[Tacitus]

NZXT G10 + H55 for the top card.

 

I've been looking at it, and I might do it. Do you know where I can find out if my card uses a reference PCB, and if not, where I can find out if it'll fit my card?

 

Cheers

[Agazed]

666

[Atmos]

The issue is certainly made worse by people overestimating how much power a 290x uses. I think some people out there legitimately believe it uses almost twice as much power as a 970.

 

I wouldn't doubt you. Won't believe how many people come in to where I work, and eventually say something like, "No i can't buy amd, it will make my powerbill go way up and my computer will be too hawt for me to like." It's something I've heard enough that I just mechanically spit out something about amd uses on average about $5.00 a year more than an equivalent intel cpu, or that amd cards use only a bit more power than nvidia cards of equal caliber.

 

Only a couple days ago I had some console peasant mlg 15yr old come up when trying to pick out parts for his first gaming rig, and he actually told me, "No I can't use an ATI AMD card, I've seen them get so hot they catch on fire and burn peoples house down." then when discussing ram he followed up with, "But if I run out of ram wont it get really really hot because it's holding more data than it can and it does that for too long it could catch on fire right???" Regardless, it took me nearly half my shift just to get the kiddo setup with a 4440, asus 760 4gb and a handful of green color led strips, along with other very generic first time computer building things. Led fans, led strips, a case with an orange racing stripe down the middle... That kinda stuff.

[Tacitus]

-snip-

 

Sounds fun

[Faceman]

I've been looking at it, and I might do it. Do you know where I can find out if my card uses a reference PCB, and if not, where I can find out if it'll fit my card?

 

Cheers

 

It is compatible, and, even better news, your PCS+ R9 290X comes with built in VRM heatsinks!

 

 

Check out my How to Water Cool a CPU and Single GPU For Under $200 USD Guide.  There is a lot of information there that I could copy and paste here if you like, but I'd rather not hijack your review.  I know that you have two cards, but if we can just eliminate the heat from one, it will go a long way towards lowering the temperatures of both, as well as drastically reducing the noise.

 

Start with one, then go to two if you like it enough.  The thing about the G10 + AIO, is that you HAVE to set your radiators to exhaust.  Now, you do have a really good and versatile case, but that doesn't change the fact that if you wanted to go dual G10s, you would need to set them to front exhaust.  I recommend starting with one as rear exhaust for the time being, see how you like it.  There are plenty of people in the G10 Owner's Club who have dual GPUs, and do the G10 mod to just their top card and it does so much for them.

 

I recommend the NZXT G10 Bracket + Corsair H55 AIO + 2nd 120mm fan for Push/Pull.  This shouldn't cost you too much, and the improvements in both temperature and noise will be incredible.

[Faceman]

I wouldn't doubt you. Won't believe how many people come in to where I work, and eventually say something like, "No i can't buy amd, it will make my powerbill go way up and my computer will be too hawt for me to like." It's something I've heard enough that I just mechanically spit out something about amd uses on average about $5.00 a year more than an equivalent intel cpu, or that amd cards use only a bit more power than nvidia cards of equal caliber.

 

Only a couple days ago I had some console peasant mlg 15yr old come up when trying to pick out parts for his first gaming rig, and he actually told me, "No I can't use an ATI AMD card, I've seen them get so hot they catch on fire and burn peoples house down." then when discussing ram he followed up with, "But if I run out of ram wont it get really really hot because it's holding more data than it can and it does that for too long it could catch on fire right???" Regardless, it took me nearly half my shift just to get the kiddo setup with a 4440, asus 760 4gb and a handful of green color led strips, along with other very generic first time computer building things. Led fans, led strips, a case with an orange racing stripe down the middle... That kinda stuff.

 

Dude, don't you know racing stripes cause your computer to catch fire because flames?

 

[Tacitus]

-snip-

 

 

Thanks a lot, and that's really handy.

 

I would probably only do one, as the coolers are really good anyway and the bottom card has kept really low temps. Should I wish to do both eventually though, would I not be able to just put 2 small radiators on the top of the case where there's room for a long rad?

[SkilledRebuilds]

-Snipped-

It is compatible

 

I recommend the NZXT G10 Bracket + Corsair H55 AIO + 2nd 120mm fan for Push/Pull.  This shouldn't cost you too much, and the improvements in both temperature and noise will be incredible.

Sorry to hijack,,.,.Curious if you know quickly offhand if the Gigabyte 290X (triple fan) is compatible? - http://www.gigabyte.com.au/products/product-page.aspx?pid=4919#ov

/It's got 4 main screws on the back, with two smaller ones further down where the VRM's are.

I have a really hot core (after reseating/pasting) up to 87*c and vrm @ 104*c and I want it lowered...

Msg me if possible to keep this thread clean if need be... Many thanks.

[Faceman]

Thanks a lot, and that's really handy.

 

I would probably only do one, as the coolers are really good anyway and the bottom card has kept really low temps. Should I wish to do both eventually though, would I not be able to just put 2 small radiators on the top of the case where there's room for a long rad?

 

You could, but tube length will be your problem, and it won't look neat.  Others who do dual GPUs in the Air 540 always have it set to front exhaust.  Start with one as rear exhaust, that is all you will probably need if your bottom card is already really cool.

[Tacitus]

You could, but tube length will be your problem, and it won't look neat.  Others who do dual GPUs in the Air 540 always have it set to front exhaust.  Start with one as rear exhaust, that is all you will probably need if your bottom card is already really cool.

 

Ok, great, I may consider doing it in a month or two. I hope you don't mind if I PM you to ask any questions down the line!