ASUS Radeon R9 290X DirectCU II Graphics Card Review
The First Custom R9 290X
It has been a crazy launch for the AMD Radeon R9 series of graphics cards. When we first reviewed both the R9 290X and the R9 290, we came away very impressed with the GPU and the performance it provided. Our reviews of both products resulted in awards of the Gold class. The 290X was a new class of single GPU performance while the R9 290 nearly matched performance at a crazy $399 price tag.
But there were issues. Big, glaring issues. Clock speeds had a huge amount of variance depending on the game and we saw a GPU that was rated as "up to 1000 MHz" running at 899 MHz in Skyrim and 821 MHz in Bioshock Infinite. Those are not insignificant deltas in clock rate that nearly perfectly match deltas in performance. These speeds also changed based on the "hot" or "cold" status of the graphics card - had it warmed up and been active for 10 minutes prior to testing? If so, the performance was measurably lower than with a "cold" GPU that was just started.
That issue was not necessarily a deal killer; rather, it just made us rethink how we test GPUs. The fact that many people were seeing lower performance on retail purchased cards than with the reference cards sent to press for reviews was a much bigger deal. In our testing in November the retail card we purchased, that was using the exact same cooler as the reference model, was running 6.5% slower than we expected.
The obvious hope was the retail cards with custom PCBs and coolers would be released from AMD partners and somehow fix this whole dilemma. Today we see if that was correct.
Continue reading our review of the ASUS R9 290X DirectCU II Graphics Card!!
The ASUS R9 290X DirectCU II Graphics Card
The ASUS R9 290X DirectCU II follows in the footsteps of the previous DC2 models by introducing a custom cooler and PCB design to the Hawaii GPU.
The card is going to run cooler, quieter and overclocked out of the box; nothing to dislike about that! Of course we have to see how their claims actually hold up in our testing.
The DirectCU II R9 290X from ASUS comes out of the box in an all-black design with a pair of fans hiding a passive heatsink on the Hawaii GPU itself. The card is both longer and taller than the reference model though so users with small cases will want to make note of the specific dimensions: 11.3-in x 5.8-in x 1.6-in.
The back of the card has a plate that doubles as both a heatsink and strengthening device for the card. Thought it's not extreme, the R9 290X DirectCU II is definitely a heavier graphics cards that can use the support when installed in a standard case.
Up top you can see the massive, 10mm direct contact copper heatpipe that helps keep the GPU cooler than the reference design could ever hope to. ASUS claims that the larger heatpipe offers up to 40% better transfer efficiency than other products while the heatsink as a whole has 30% more dissipation area for air to pass over.
The left fan in this setup is based on the ASUS CoolTech fan design that offers a hybrid approach combining the axial and blower options typically seen on graphics cards. The goal is to provide high airflow while keeping acoustics low.
The output configuration includes a pair of dual-link DVI connections and a full-size port for both HDMI and DisplayPort.
Along the bottom you will find four additional heatpipes for moving heat from the GPU surface the fins of the DirectCU II cooler.
The ASUS custom 290X still only requires a 6-pin and 8-pin but ASUS has implemented an 8-phase DIGI+ VRM system with super alloy components to extend the life of the card and hopefully increase overclocking capability. To help users that hate the electronic buzzing sound some R9 290X cards have had ASUS went with concrete alloy chokes.
Even though I am huge fan of the stealth appearance of the ASUS card out of the box, ASUS does include two sets of decals that you can apply to the card in both red and gold colors. These are obviously meant to match the various motherboard options ASUS has on the market today.
Here you can see it decked out in all gold...
...here it is in red. You can mix and match if you want, or leave it all black. It's great to see ASUS include these options for the consumer.
I mentioned the ASUS R9 290X DirectCU II is larger and longer than the reference card, but here is proof how much deeper your case needs to be to fit this custom design. Small Form Factor (SFF) users might want to watch out and measure carefully before purchasing this unit.
Damn, did not expect it to actually do this well. Nice review Ryan.
Awesome:ooo
Please Jesus:((
Jesus can't help you.
Exactly, let's wait til these actually come out and what Nvidia is going to do to counter this cause we it could cause a TI drop to reasonable pricing. Thanks Ryan as always!!
NV doesn't have to respond with anything. You can buy an MSI 780ti gaming card for $699 that is 1020/1085. So a FREE OC is already priced at REF MSRP. It won all benchmarks at hardwarecanucks review of this 290x card except for Dirt Showdown which nobody plays (I say that because it hasn't sold 100K copies yet according to vgchartz which is what it takes to show up on their charts, total failure). Also it won everything without OCing, which nets another 10% as it runs up to 1225 as they show.
Tried to put a link in to hardwarecanucks but caught for spam. I'm sure anyone can get there on their own anyway (newegg below also).
You should be testing at MSI speeds since you can get an OC card for ref pricing. You are testing an OC card here (290X asus) vs. REF 780ti and asking if they took the crown? NOPE. Because you can get OC 780ti for $699 already that easily wins.
The EVGA REF model is now $680 also on newegg once in the cart. More expensive than 290x maybe, but comes with gsync, 3 AAA games, cooler etc.
You really are a Nvidia fanboy. This 290x beats the 780 TI through and through. Not only does it run slightly cooler, the frame rate is the exact same. OH wait, its $150 cheaper and if Nvidia fanboys can't realize that Nvidia does need to respond with a price drop, they will be seriously hurting. I know fanboys like you will always keep them in business though because you're paying for an overpriced piece of crap. I don't know where you get that the EVGA Ref card is cooler, but it isn't. Look at the 4 degrees difference. Also AMD comes with the Never Settle Bundles, so you have zero point there with the AAA titles. Gsync is pointless. Drivers are far better on AMD now. What's that? OH wait...Games like BF4 AMD dominates Nvidia in every shape and form and this is all before mantle. Is Nvidia in the PS4 or Xbox? Nope. If Mantle does well, is every game from here on out going to adpot it? Yep. So basically Nvidia has an overpriced card that does the same things as the 290x but for way more money and....No thats pretty much it. Have a nice day.
fanboy, lol me think you can't read graphs, only one where it comes out on top. Then you have the 780ti direct cu2 which eats your AMD fanboy ass
This is the first 290X non-reference card out. If you think there aren't going to be faster non-reference 290Xs released, you've lost it.
enjoy the crow when it's served.
Yes kudos to ASUS for fixing a broken product. Ryan , put it up against the Gigabyte 780ti Ghz Edition or the EVGA SC ACX, would be interested in the results for non ref v non ref
Cheers
>g-sync is pointless
What?
You do know what g-sync does right?
I'll second what Snook said. Until G-sync gets mainstream, (meaning it would be fully compatible with most displays out there) which won't happen anytime soon, I don't really see the point of it. Most people won't be able to use it. And it's not exactly cheap. Don't get me wrong, G-sync is great and all, but it will take some time before it's really a valid argument for going green.
it doesn't come with g-sync. it comes with the ability to use g-sync. huge difference. you get to pay extra for an aftermarket kit (per ryan, limited) or buy a g-sync enabled monitor.
last podcast points out the g-sync limits also. many more than I thought it would have.
OMG, you are actually getting really upset about this review, and getting into a defensive position for Nvidia? *The fanboy is strong in this one*
How much more FPS do you get with the sticker ?
&
Which color combination gives the most FPS ?
Actually blue stickers run cooler giving you better overclocks, its science and junk. I have a friend in the "know" that says Asus is saving the blue stickers for the Rev. 2 models.
speed holes, get out that pickaxe, or shotgun, and make some Holes, at least on those refrence SKUs!
Shutup Homer.
Shut up, pleb.
Damn! Can't wait for those.
how could AMD have not come up with a similarly performing solution for their flagship cards?
AMD have cut their engineering workforce down to almost the bare-bones to stay alive, and maybe AMD should go with a flagship non-refrence design partner, and forgo the costs of thermal engineering to that partner, until AMD rehires/recommits more of its engineering assets towards thermal engineering and cooling/sound design. Yes the parts only cost $20 more, but the engineering costs much more than the materials, and if your engineering workforce is not there, or not available, it is better to get a flagship partner, and utilize that partners engineering workforce.
Alms for the cooling design challenged.
Great. But I think most of us knew it for over a month now when many used third party coolers on 290X and had excellent results.
This must be the only time ever that the same gpu that lost the performance crown took it back with just a cooler change.
It's shocking how dumb some people are....
very comprehensive review Ryan, and you have made some good points why AMD could not have done a better job with these cards when they came out of the gate is worrying. These cards do create a lot of heat and as such should have better built in cooling.
FAWK YEA am selling off my 290 ref!
AMD shut up and take my money!!!
Do you run your Skyrim tests with the official HD Textures DLC?
It's the DLC textures but no mods.
Ryan, I know it makes it harder to compare your results with other websites, but I don't know ANYONE who runs Skyrim on PC with no mods.
If you do what TekSyndicate do and run all their tests on a heavily-modded Skyrim, the game will push the GPU further. And so long as none of the mods are ENBs or anything, (so stick to texture and lighting mods) the performance should scale with texture quality, etc.
Might also make Skyrim more relevant as a benchmark, since in most of your latest reviews you've said how pretty much any new GPU can run Skyrim maxed out.
Personally, I wouldn't mind seeing these cards put through their paces with an ENB and DDOF. I noticed on one review - I think an Anandtech review - the AMD cards seemed to react worse to DDOF in Bioshock Infinite than Nvidia ones, and I'd very much like to see if that is a game-specific aberration or a trend.
The beauty of modded Skyrim as a benchmark is that you can test the cards for different kinds of advanced graphics/lighting features from mods and then compare it to vanilla Skyrim as a control.
OMG OMG..finally. Thanks for the review Ryan, was excited when you announced it on Facebook.
Great to see these cards the way they were meant, Great Performance and consistency...oh and the noise levels!! Great job Asus!
"Maybe the best news is that even with this overclock, running Metro: Last Light for an extended loop, we never saw the GPU temperature exceed 86C. The 1150 MHz clock speed was pretty much fixed (no variance) and the fan speed stayed around the 2000 RPM mark (53%). Yes, it was louder than stock but still not close to what we have seen with the AMD cards in quiet or Uber mode."
I remember your facial expression every time the subject of the 290X and its noise was brought up in the podcast ..You were Like the Captain Picard face Palm meme!!
LOl..to bad these cards wont make it to be under peoples Christmas Trees this Holiday, but At least we know they are ready!!
The Price there showing doesn't exist anymore 3 weeks ago a 280 was 300 now 420$ I get the B4 280 for 300$ which means I paid only 250$.
I wanna see ryan complain about these, maybe he will create a new metric to review these, noise? heat? powerdraw? drivers? or he will now recommend it because it is selling too well...not saying he is biased just saying he is a Journalist searching for a story down any BS avenue...
Sweet username bro.
For the AMD refrence SKUs, these induced clock/thermal feedback variances/oscillations can unnecessarly result in extra thermal wear and tear, and premature failure, and however unnoticeable they may be in game play, are still in need of review and mention, with thorough examples! And if you think AMD's fixing/mitigating the matter as quickly as possable with driver fix, was just beacuse of a few benchmarks pushing them to act quickly, then try paying the warranty, and extended/implied warranty costs, that AMD may have had to, and have incurred, should they have swept the problem under the rug!
That's why i never listen to reviews unless they only talk about stock speeds when running benchmarks. When these reviewers start talking the price is when i get turned off on the cards an go into a different direction.
So AMD has lost my business after 10yrs of only using AMD cards.
Too bad for you i guess, they loose you and gain me. I will switch my 680 our for a 290, if it does as well as the 290x.
So balance is yet again restored.
I've seen water cooling reviews and now custom fan solutions and both make this card run as advertised. It seems to me AMD barely made a minimum effort to cool the card. Once again I'm rather disappointed with the efforts by AMD. And I'm a fan......for now
I agree that the reference cooler performs poorly, but reference coolers are built for maximum compatibility. There are situations where only a blower cooler will work, and it cannot be too large either. There needs to be a bare minimum cooler that works for everyone, even if it's not very good. People who want better cooling and have enough space, airflow, etc can get a custom cooled card.
That said, the delay for custom cards is disappointing, and Nvidia certainly had a better reference cooler, although it likely cost more.
Ryan writes: "It looks great, overclocks well and ASUS is only asking $20 more than the current MSRP. You can't really demand much more from an enthusiast class GPU today!"
Good Review!
It is not good enough for NVidia fans. Why? Because this card uses AMD chip and its price is very reasonable. NVidia fans wont buy a good AMD product. They love to bash AMD.
Very nice to see a backplate on a card costing only $20 over reference. Hopefully Asus releases a 3-slot card as well; if any GPU needs it, it's the 290, although this card is certainly enough for anyone who can only spare 2 slot.
But surely you'll need to compare it to cards like the Asus 780 Ti DCUII to be fair, Either way the 290x and no doubt 290 are again very desirable. 290 will be a bargain and a custom version will be faster than stock 290X IMO. I still don't which one I'd buy out fo custom 290X or custom 780 Ti however. Need to wait for more custom 290(X) solutions to compare.
I hope to test these soon!
Well, I might not be able to get one this year. I do know I'll be asking for some New Egg gift cards. 8)
What were the clocks of the gtx 780 ti? Were they limited to 876 mhz? Your results seem kind of low for bioshock infinite. Bioshock seems like a game nvidia hardware do better at.
GTX780Ti running at non-boost on a PCper review of an AMD card???? now, that is funny.
The clocks weren't limited at all. They were running in their stock functionality. I don't have the specific clock recorded for Bioshock Infinite though.
EDIT: Actually I do!! Here is the GPUZ data from our Bioshock testing on the GTX 780 Ti.
http://screencast.com/t/eQfWxYNF3
The numbers you see there are GPU clock, memory clock, GPU temp, fan %, fan RPM.
thanks, I was right again! woohoo. not so much a knock against you as it was dismay at his question. essentially thermal throttling at 1006.2?
In an ironic twist, the 780Ti would bet the 290X if Nv allowed the gpu a higher thermal limit..say 90c.
What thermal throttle are you talking about buddy???
The card was between 1006 and 993mhz. That is well above the minimum stock boost specifications of the 780Ti. 7mhz variance is nothing.
At similar clocks the 290x and the 780Ti are very close. Deal with it.
Also the 290x was not running at its thermal limit. It was well below that.
Yah, agreed here. NVIDIA's ratings and clock speed claims are obviously being met by our results.
my claim isn't against clocks, simply asking why that limit. was it thermal? see, nice and clean.
stated like this: why didn't it boost above 1006.2? that seems the max you got from it.
it was a question buddy???
if it's not thermal (780Ti buddy??), why didn't gpu boost take the clock higher?
again, a question buddy??? seeking information. deal with what? I don't have a dog in this fight other than contempt for Nv.
let it be known, from this day hence. I predicted this result on three separate occasions, on this PCper site. nice being right.
You shall forever be known as the prophet of thermal/performance results.
Also, since we're on the internet, SS or it didn't happen.
no SS, 24th comment from top: http://www.pcper.com/reviews/Graphics-Cards/NVIDIA-GeForce-GTX-780-Ti-3G...
Look over there it's SUPER ALLOY POWER MAN to the rescue... He will save us.
Great to see the 290X taking on and often beating the 780TI, not to mention the titan.
Someone above me mentioned about the Bioshock results looking low. I believe i can explain that and maybe Ryan can confirm im correct. Pcper are using the stock Ultra preset settings for Bioshock. This actually favours AMD cards. Most sites, but not all use the custom DDOF preset. This actually adds zero image quality improvement, it actually looks worse. What it does do though is favour Nvidia cards and hits fps quite hard for no improvement. If you see AMD winning in Bioshock, they used Ultra preset. If you see Nvidia winning they used the custom preset, this is assuming the cards are fairly evenly matched ala 770vs7970 or 780vs290 etc.
You are correct, thank you LtMatt!
I will admit that I do not regret purchasing the reference card. In Quiet mode it is not too annoying, but throttles too much. Seeing tomshardware test reference card with aftermaket cooler showed how much potential was hidden.
Seeing this review just verifies that what AMD did with it's cooling was a shot in the foot, giving some negative feedback on a good card.
One thing I really like about this ASUS design compared to other cooler designs for R9 290x that are floating on the web is that this still blows the air out of the case (at least in part) Gigabyte and MSI do not.
If I was to get another R9 290x this would be the card I get. And most likely will to finally have Dual GPU setup - current GTX 680 + R9 290x don't count
Great review. This is what I was expecting. Asus has done a great job. I just wish it was available right now. Nvidia would have been caught out again requiring them to make another price revision.
Oh Well. At least hoping that the card would be available in decent numbers at the quoted price.
It would be great if you guys could do a review of this card in comparison to the Gigabyte 780 ti GHZ edition once you get your hands on an actual card. There's talk about that version of the ti murdering the performance charts for only a little more than a reference 780 ti.
There are unlimited options for promoting the products.
I am sure there are many more examples and the above can include both ready made
and those crafted by loving owners too. Jo Han Mok is the #1 International bestselling author of the E-code.
Here is my blog post; my site (Sherlyn)
why are you still testing on BF3 , next to no one is still playing it. Battlefield 4 has been out for a few months at this stage , any chance you could get some benchmarks with that instead ?
Error in the article.
"To put it in a different perspective, the average frame rate of the ASUS card was 1047.57 MHz over the 25 minute period"
Should be:
To put it in a different perspective, the average CLOCK SPEED of the ASUS card was 1047.57 MHz over the 25 minute period
Thanks!!
I would still like to see the 780 ti at it's maximum overclocking potential (1200MHz core seems easily achievable, and 1250-1300MHz is possible) compared to the overclocked ASUS R9 290X @1150MHz. Then the benchmarking results will be fair. Maybe Linus will cover this in one of his videos.
I would also like to point out that the AMD cards are selling well above their MSRP's, perhaps due to the bitcoin miners right now. So the $570 R9 290X actually costs $620-$630 (if you can even find it). The 780 ti, MSRP @$700 comes with $170 worth of free games this holiday season. Overall the 780 ti still offers better price/performance ratio.
In sweden we have high availability of the R9 290X, and the prices has gone down somewhat.
Interesting, I'll keep an eye on it for sure.
Availability of the reference model shouldn't be a problem anymore? Well at least not here in sweden. The prices has gone down to (somewhat).
What monitoring software are you using to record the gpu load differences? I ask since I have a custom cooling solution on my 290x and during gameplay of current games, the hottest the gpu ever gets is 70 degrees but the gpu load goes from 0% to 100% literally hundreds of times during a gaming session (looks like a seismograph). Never do I see a solid line at 1000Mhz. Never! I wonder if it is the software that I am using (MSI Afterburner) to monitor or my card. Please note that my gampeplay is silky smooth but the reporting numbers that concern me. So if I could get some insight on the software you are using, that would be great.
We use GPU-Z.
Snook:
The throtteling isn't an issue really. First you have to consider that the fan is only running 59% in Ryan's bench. Second the powertarget and temp target can easily be up'ed by flashing the bios to one of skyn3t's rev's.
I have a 780 Classy running completely stable (with ACX air cooler) @ 1270/1750 with max temp 75C (skyn3t ln2 rev.3).
As from what I have read/seen; the 780Ti (specially the Evga classy/SC versions) hits about the same clock speeds as mine on air.
thanks for the response sans the "buddy???".
those are really good speeds too. my question was just what is happening to limit it to 1006.2? so powertarget seems as good as anything.
thanks again.
hmmm.. i wish the price is too good to be true.. in Malaysia all these price are inrease a titan will cost u as much as 1125.79$ and this radeon R9 290x will cost 852.18$.. we do not know even why they cost so much.. i wish they were cheap like other countries.. the price of gtx 770 can buy a gtx 780 in other countries. and a gtx 680 still cost 608.70$ if guys dont believe me just go to any online retail shop pages.. and also gtx 760 is the same price of a gtx 660 ti that is 369$ thats a rip offf!!!!!
Why do you compare a non-reference AMD to a reference Nvidia card. Would it not be better to compare non-reference to non-reference? It would make it easier to see which is the better card and by how much.
Yah, we are waiting for some more custom GTX 780 Ti cards to arrive.
Good review!
Thank you!
Its takes a custom cooled overclocked 290x to match a stock reference 780ti. We all know how well a 780ti overclocks so they should of tested with apples against apples. anyway still a fail on Amd part.
with that huge 44Mhz OC :/ yea, them apples. sour apples.
it didn't match it beat it BTW
Exactly. You should be able to hot 1150 mhz without issues so add another 100 mhz on the factory OC. Even if 780 ti hits 1250 stable, it is still 130 dollars more expensive. And as a previous Swede explained, the price gouging is a US-centric issue. And I do not even think it is that related to mining. We have seen a similar mining craze here in Sweden.
Anyway, this card crushes the stock Ti on price/perf ratio. Custom coolers will narrow the gap but unlikely to close it. But Maxwell for desktops is coming in 3 months! Hoprfully AMD can counter faster and prevent Nvidia from taking their customary high prices.
Exactly. You should be able to hot 1150 mhz without issues so add another 100 mhz on the factory OC. Even if 780 ti hits 1250 stable, it is still 130 dollars more expensive. And as a previous Swede explained, the price gouging is a US-centric issue. And I do not even think it is that related to mining. We have seen a similar mining craze here in Sweden.
Anyway, this card crushes the stock Ti on price/perf ratio. Custom coolers will narrow the gap but unlikely to close it. But Maxwell for desktops is coming in 3 months! Hoprfully AMD can counter faster and prevent Nvidia from taking their customary high prices.
http://www.pcper.com/image/view/35024?return=node%2F59148
The graph with the temps comparing the Quiet mode and the Performance mode idle and load temps is mislabeled. Idle is labeled as Load.
Oh, darn, that's the Sound level graph, not the temp one. Sorry!
Oh, darn, that's the Sound level graph, not the temp one. Sorry!
How many nVidia fanboys here. Here I'am switching from ASUS GTX670 DCII to the Asus 290 (probably not X) when available
What would be really nice is to see how quiet we could get the fans if the target temp was set at say 90C, because the GPU is meant to handle 95C without a hitch, for the life of it. So it seems we are either leaving performance or quietness on the table by limiting the temp under 80C in stock settings.
when will this be available?
I was thinking about getting a 780, but this is changing my mind completely for multiple reasons.
I was going to buy a GPU this month since I am in america right now and ofcourse booya euro vs dollar shit.
I might of been interested in this card but eh I am not going to pay 200-300 dollars more just because I live in europe so I think I will be buying a 780 Ti then.
the ONLY 290X non ref that is for sale is on newegg right now and it's the gigabyte one.
and that one is 699 dollars so there is no point to buy it over say a MSI 780Ti or hell a 749 EVGA classified which has been proven to overclock to 1300+ easily on air (with some bios fiddling)
Bluh this sucks. replacing my 680 is turning out to be a pain in the ass.
One thing that would really interest me since Sapphire is the biggest contender for Asus in the AMD custom cooling sector:
How does the db(A) of the Asus r9 290x in performance/quiet mode compare to the Sapphire r9 290x or the Asus r9 280x?
Thanks in advance!
The xfx R9-290A-ENBC is 499.00 at ncix.ca and 539.00 newegg.ca Canadian dollers. The custom 290 price is around 570.00 to 600.00 and that is not even the 290x. I was going to ask do you guys know what the sp is of the fans becuase it is static pressure and a good heatsink on the card that would chill it down good right? I was thinking a blower fan design is the best for static pressure right or wrong ? and what if you just purchased a usb extender 10$ and hmdi extender 10$ and put it in a closet with 80% or more fan speed on a stock 290 blower design. I have seen most custom r9 290s is 70$ or 100$ more and the same if you buy a aftermarket cooler it's around 100$ for it.
Newegg just raised the price from $699 to $749. In fact if you search 290x cards, they show $699 but when you open the product page the $749 price appears This is the case for most 290x's.
Now, is it juѕt mе,or Ԁo you have somе trouble finding truth
іn that. hаd undergone а revival ɗuring оur nations' bicentennial
ɑnd noww has. Iո оrder to improve іtѕ products in readiness fօr the Beijing Olympics, thе
company reported іn June that ovеr 20,000 pieces of ɗifferent equipments ɦad been tested over
the lɑst one yeаr and the testing was supposed to bе ߋvеr by July.
Post new comment