AMD Radeon R9 290X 4 GB

[Xzibit]

+1
you don't say..

http://s24.postimg.org/3nmdnj3d1/Screenshot_from_2013_10_25_08_01_51.png

It looks like that this card somehow throttled at 92 -94*C zone. I'm afraid that this card would still throttled at 92-94*C even using water cooling setup. Is there any possibility that this card would go into 75*C-80*C zone with driver optimization?

The PowerTune is temp controlled so you can adjust by temp but the fan will speed up to maintain the low temp.

Card operated on limits
Temp = 95C
Silent Fan = 40%
Uber Fan= 55%

If you want to go beyond the fans 55% in uber you then become limited by temp.

Its a give and take. With water the temp will be lower thus telling the card not to speed up the fan so temp nor the fan limit will be hit until you start pushing it beyond the default settings. depending on how well its cooled your then looking at a chip/memory limit rather then a temp/fan limit.

 

[TheoneandonlyMrK]

I honestly don't get what's all this fuss about power consumption and temperatures.

Who in earth buys a high end graphics card without having its cooling planned?

Reference and high end does not exist in the same context, imho.

Either watercooling or aftermarket, pick one

Yes I know that Nvidia gets reference right but they charge a well damn hefty premium for it.

Exactly , I would rather watercool it on my own anyway but im skipping this gpu as I don't have the pixels for it.

 

[The Von Matrices]

These cards are being sold mostly to gamers, so no one cares how good the cards are at GPU Compute tasks. You're making the same argument that people made to defend Fermi. AMD should have learned from nVidia's mistake and learned the GPU Compute doesn't sell desktop GPUs.

I agree that Titan was targeted to gamers as well as compute users, but your claim that Titan won't sell as a compute card by comparing it to GTX 480 is ridiculous. GTX 480 wasn't a compute card. The two cards are completely different.

There are two reason to buy a Tesla over a GeForce:
1) Unrestricted double precision compute
2) Stability features like ECC RAM
3) Better drivers which help with utilization with multiple cards and nodes

Back in the GF100 days, all three differences applied, so there was a good reason to pay $3000 for a Tesla card that could get 4x the double precision performance of a $500 GTX 480, which was also handicapped with 1.5GB of VRAM compared to 3 or 6GB on Tesla cards. This is why no one bought a GTX 480 for compute.

Today, you can get a Titan with unrestricted double precision compute; a single Titan can match a single Tesla in double precision compute. Titan also has gobs of VRAM, which helps immensely in visualization. If you want a single card for compute, nothing can beat Titan at $1000.

Sure, NVidia won't sell as many Titans at $1000 now that R9 290X exists, but to say that Titan is a dead product at $1000 is a very narrow viewpoint that omits how much of a steal it is for compute users compared to Tesla.

 

[MxPhenom 216]

I agree that Titan was targeted to gamers as well as compute users, but your claim that Titan won't sell as a compute card by comparing it to GTX 480 is ridiculous. GTX 480 wasn't a compute card. The two cards are completely different.

There are two reason to buy a Tesla over a GeForce:
1) Unrestricted double precision compute
2) Stability features like ECC RAM
3) Better drivers which help with utilization with multiple cards and nodes

Back in the GF100 days, all three differences applied, so there was a good reason to pay $3000 for a Tesla card that could get 4x the double precision performance of a $500 GTX 480, which was also handicapped with 1.5GB of VRAM compared to 3 or 6GB on Tesla cards. This is why no one bought a GTX 480 for compute.

Today, you can get a Titan with unrestricted double precision compute; a single Titan can match a single Tesla in double precision compute. Titan also has gobs of VRAM, which helps immensely in visualization. If you want a single card for compute, nothing can beat Titan at $1000.

Sure, NVidia won't sell as many Titans at $1000 now that R9 290X exists, but to say that Titan is a dead product at $1000 is a very narrow viewpoint that omits how much of a steal it is for compute users compared to Tesla.

Meh, Titan still isn't that great in compute. There are quite a few benchmarks that still show that the 7970 can beat it by quite a bit.

Real compute beasts from Nvidia are not affordable to the average consumer.

 

[qubit]

Too little too late and that noise is a complete dealbreaker.

"AMD should have invested some time and money into developing their own high-end cooler, like NVIDIA did for the GTX Titan. The noise figures of this reference card only go on to show that AMD should urgently allow its board partners to launch cards with non-reference air coolers that can handle the heat at saner noise levels." Says it all.

It should only come in at around a 6.5-7.5 score I think. 9.3 editor's choice is being far too generous to this card. It might be priced a lot lower than a Titan, but then it's a lot less card too, so I dunno why wizzy is bowled over by its price and given it such a high score.

 

[MxPhenom 216]

Too little too late and that noise is a complete dealbreaker.

"AMD should have invested some time and money into developing their own high-end cooler, like NVIDIA did for the GTX Titan. The noise figures of this reference card only go on to show that AMD should urgently allow its board partners to launch cards with non-reference air coolers that can handle the heat at saner noise levels." Says it all.

It should only come in at around a 6.5-7.5 score I think. 9.3 editor's choice is being far too generous to this card. It might be priced a lot lower than a Titan, but then it's a lot less card too, so I dunno why wizzy is bowled over by its price and given it such a high score.

I don't know why you think its less of a card than Titan. Other than the heat and noise, for the price the card is a winner. Most people even considering it right now already plan to throw a waterblock on it, or are waiting for board partners to release their designs.

There is one issue I have with the reference design though. The fact that uber mode shouldn't be used with the stock cooler for day to day use as noted by Wiz. Who wants to hold back their $550 video card purchase? Unleash that shit, but then you have to throw a $100 water block one it to do it comfortably IMO.

 

[The Von Matrices]

Meh, Titan still isn't that great in compute. There are quite a few benchmarks that still show that the 7970 can beat it by quite a bit.

Real compute beasts from Nvidia are not affordable to the average consumer.

You need to do double precision for Titan to make any sense. I agree, AMD can beat it in single precision.

Too little too late and that noise is a complete dealbreaker.

"AMD should have invested some time and money into developing their own high-end cooler, like NVIDIA did for the GTX Titan. The noise figures of this reference card only go on to show that AMD should urgently allow its board partners to launch cards with non-reference air coolers that can handle the heat at saner noise levels." Says it all.

It should only come in at around a 6.5-7.5 score I think. 9.3 editor's choice is being far too generous to this card. It might be priced a lot lower than a Titan, but then it's a lot less card too, so I dunno why wizzy is bowled over by its price and given it such a high score.

I think that a lot of people see Titan and think that since R9 290X is half the price you can accept its flaws. What I argue is that R9 290X is still a $550 card, which is a LOT of money. You shouldn't have to put up with this when you spend that much money. NVidia did AMD a favor, because if all that existed was the $650 GTX 780 and there was no comparison to the $1000 Titan, I think the conclusions would be much different.

 

[sweet]

Too little too late and that noise is a complete dealbreaker.

"AMD should have invested some time and money into developing their own high-end cooler, like NVIDIA did for the GTX Titan. The noise figures of this reference card only go on to show that AMD should urgently allow its board partners to launch cards with non-reference air coolers that can handle the heat at saner noise levels." Says it all.

It should only come in at around a 6.5-7.5 score I think. 9.3 editor's choice is being far too generous to this card. It might be priced a lot lower than a Titan, but then it's a lot less card too, so I dunno why wizzy is bowled over by its price and given it such a high score.

It only scored 9.3 because of the heat and noise. For its superior performance alone, it would have been a 10. Furthermore, another 10 score for its ideal price would be reasonable.

 

[MxPhenom 216]

It only scored 9.3 because of the heat and noise. For its superior performance alone, it would have been a 10. Furthermore, another 10 score for its ideal price would be reasonable.

I do not think Wizzard gives 10 scores, except for the DirectCU Top GTX670 a while back, but I think that might have been a mistake. A lot of people were buying those cards due to that score, and then a lot of them were failing IIRC.

 

[sweet]

I do not think Wizzard gives 10 scores, except for the DirectCU Top GTX670 a while back, but I think that might have been a mistake. A lot of people were buying those cards due to that score, and then a lot of them were failing IIRC.

Completely agree! A perfect 10 should only be rewarded to a flagship, not a mere harvest chip.

 

[MxPhenom 216]

Completely agree! A perfect 10 should only be rewarded to a flagship, not a mere harvest chip.

I disagree. No piece of computer hardware should ever get a 10. There will always be some sort of flaw.

 

[sweet]

I disagree. No piece of computer hardware should ever get a 10. There will always be some sort of flaw.

That's the point. The flaw of any non-flagship GPU is that it will be beaten in performance. Therefore, the perfect 10, if ever exists, should only belong to the performance king.

 

[xorbe]

or his testing room is acting like a parabolic reflector to his neighbors.

This actually happened to me once. When I tracked down the "bong" sound that started at 5:30 am for 2 week ... it was a neighbor playing a video game, with a decent sized plastic computer speaker across two parking lots at his window frame (aimed inside), and it wasn't even that loud there, but it resonated in my bedroom. True story. He was nice and used headphones from then on.

 

[Loosenut]

That's the point. The flaw of any non-flagship GPU is that it will be beaten in performance. Therefore, the perfect 10, if ever exists, should only belong to the performance king.

All kings eventually lose their crowns

 

[mastershake575]

Too little too late

As a current AMD card owner, even I agree

Don't get me wrong, the 290x is fast and the price is good but it isn't the 780 killer that most people hyped it as (hence why people are comparing it to the Titan instead of the 780).

Stock vs stock and overclocked vs overclocked, its basically the same damn thing as the 780 while being $75 cheaper (NVidia is dropping the price of the 780 to $550-575 in the next few weeks to pretty much cancel things out all while releasing the 780 Ti for $650 that will take single GPU crown).

I'm not extremely disappointed with this card ($550 is a good price) its just a $550-575 780 and a $650 780 Ti will pretty much negate all the pros of the card (price cut 780 will make the 290x not really a better value of the money since it will be identical and a 780Ti will clearly take the single GPU crown while costing not that much more)

After reading this review im for damn sure waiting for the 20nm GPU"s next year (the GTX 870/pirate island equivalent should offer better performance than the 290x/780 for only $400).

 

[okidna]

Good review like always, but I'm curious about something.

From 290X and 280X review :

The card requires a 6-pin and 8-pin PCI-Express power connector. This configuration is good for up to 375 W of power draw.

From 270X review :

The card requires two 6-pin PCI-Express power connectors. This configuration is good for up to 300 W of power draw.

From 780 and TITAN review :

The card requires one 6-pin and one 8-pin PCI-Express power cable for operation. This power configuration is good for up to 300 W of power draw.

From 760 review :

The card requires two 6-pin PCI-Express power connectors. This configuration is good for up to 225 W of power draw.

Is there any difference between AMD's and NVIDIA's power connectors configuration & specification? Or the difference coming from PCI-E 3.0 specification?

 

[sweet]

As a current AMD card owner, even I agree

Don't get me wrong, the 290x is fast and the price is good but it isn't the 780 killer that most people hyped it as (hence why people are comparing it to the Titan instead of the 780).

Stock vs stock and overclocked vs overclocked, its basically the same damn thing as the 780 while being $75 cheaper (NVidia is dropping the price of the 780 to $550-575 in the next few weeks to pretty much cancel things out all while releasing the 780 Ti for $650 that will take single GPU crown).

I'm not extremely disappointed with this card ($550 is a good price) its just a $550-575 780 and a $650 780 Ti will pretty much negate all the pros of the card (price cut 780 will make the 290x not really a better value of the money since it will be identical and a 780Ti will clearly take the single GPU crown while costing not that much more)

After reading this review im for damn sure waiting for the 20nm GPU"s next year (the GTX 870/pirate island equivalent should offer better performance than the 290x/780 for only $400).

WRONG! They are not the same when overclocked. 290X is a beast when overclocked given that you can cool it down.

http://xtreview.com/addcomment-id-31366-view-Radeon-R9-290X-vs-GeForce-GTX-Titan-benchmarks.html

In the chart above, MSI Lightning 780 actually run at 1300 MHz range, while 290X OC at 1100 MHz often throttles down near 1000 MHz. You can still observe a big different in performance here.

If you believe in TPU chart and claim that 780 is just 1% below 290 silent, let me tell you this fact: The overall chart of TPU is in fact, deceiving. In some adnormal case such as Starcraft 2, Splinter Cell: Blacklist, a.k.a games favoring old DirectX version, AMD's cards generally doesn't perform at their full potential. Only modern DirectX 11 games, which AMD's GCN cards are designed for, can take the best of AMD. The overall result, sadly, is heavily affected by the old tech games, and that doesn't feel right.

 

[Xzibit]

As a current AMD card owner, even I agree

Don't get me wrong, the 290x is fast and the price is good but it isn't the 780 killer that most people hyped it as (hence why people are comparing it to the Titan instead of the 780).

Stock vs stock and overclocked vs overclocked, its basically the same damn thing as the 780 while being $75 cheaper (NVidia is dropping the price of the 780 to $550-575 in the next few weeks to pretty much cancel things out all while releasing the 780 Ti for $650 that will take single GPU crown).

I'm not extremely disappointed with this card ($550 is a good price) its just a $550-575 780 and a $650 780 Ti will pretty much negate all the pros of the card (price cut 780 will make the 290x not really a better value of the money since it will be identical and a 780Ti will clearly take the single GPU crown while costing not that much more)

After reading this review im for damn sure waiting for the 20nm GPU"s next year (the GTX 870/pirate island equivalent should offer better performance than the 290x/780 for only $400).

I'll be waiting it out aswell. Usually I wait until the next revisions show up and get what I want on Sale.

Seams to do just fine against a 780 HOF clocked at 1006-1059mhz in FireStrike so its not too bad eh.
Galaxy GTX 780 HOF3GB @ $689.99

Then you consider how the 290X achillies heal is the ref cooler which is probably throttling in the test well under 1ghz more like 830-900mhz.

I say not bad for $140 difference.

Now if the HOF cooler was on the 290X then you'd see the 290X OC "uber" number get closer to the R280 DC2TOP CF number.

If it almost best a 780 HOF on both FireStrike scores while still in reference form its not a killer to the ref 780 ? I guess its perspective then.

 

[AsRock]

Would of been nice to see if the positional sound though HDMI worked as it should and if you could tell if some thing was inn front or behind you like you can with older AMD cards.

 

[The Von Matrices]

Would of been nice to see if the positional sound though HDMI worked as it should and if you could tell if some thing was inn front or behind you like you can with older AMD cards.

If you're referring to TrueAudio, AMD has no driver for it yet, so as of right now it can't be used.

 

[arterius2]

LOL people keep contradicting themselves on this forum.

First, they'll start off with the argument that the 290X is about ~$100 cheaper than the GTX780 (and that the price killed the 780)

and then they'll say "Oh who cares about temp, noise and power consumption", -but you know what? most people do! Then after that, they'll say "Just throw a waterblock on it".

ummm, if you truly didnt' care about noise and temp why are you throwing a waterblock on it, and how much does a waterblock cost? around $150, so basically, 290x + waterblock cost around $700, that is only if you already have a waterloop setup, if you don't, well you are likely to run into the $1000s+ to have everything setup just to run this card within acceptable parameters.

and don't give me this 'that everyone who buys high end already have a waterloop setup', that is just BS. most people don't run waterloops and those who do are in the small minorities, and the people who makes these claims don't even have waterloops themselves.

and when all arguments fail against them, they'll bring out the performance card, where they claim 290x "absolutely destroyed" titan, um why are you comparing to titan first of all(it LOST to titan in silent mode, and if u are comparing in uber mode, then don't give me this its only 2db louder, because its 12db louder in uber mode), it barely even "destroyed" 780, they are well within 1~5% of each other trading blows. that is not what "destroy" means, "destroy" means beating something by at least 15-20%, otherwise, a few fps difference is barely even noticeable to the human eye.

so to sum it up:
1) they claim its cheaper
2) they ignore noise, temp and power consumption
3) then they say throw a water block on it ignoring the price of a water-cool kit (+$150 for the block and +$400 for the whole setup)
4) selective/inconsistency in comparison of performance of 290x (in uber mode), but using silent mode to compare noise and temp.
5) overhype with words such as "destroy" "kill" "massacre" when benchmarks show they are fairly equal matched.
6) selectively/inconsistency in comparison of price to titan when situation favors them. but compares to 780 when other situations fits them(performance).

 

[1d10t]

+1
you don't say..

http://s24.postimg.org/3nmdnj3d1/Screenshot_from_2013_10_25_08_01_51.png

It looks like that this card somehow throttled at 92 -94*C zone. I'm afraid that this card would still throttled at 92-94*C even using water cooling setup. Is there any possibility that this card would go into 75*C-80*C zone with driver optimization?

naaahh...it wouldn't touch 92'C mark if you drowned them under water
Universal GPU block such as my XSPC Rasa have done perfectly well job keeping 7970 below 60'C...even if you furmark'ed them they still even touch 80'C.

Good review like always, but I'm curious about something.
Is there any difference between AMD's and NVIDIA's power connectors configuration & specification? Or the difference coming from PCI-E 3.0 specification?

Just like processor counterparts,AMD usually set the TDP higher.What you see is maximum in rare occasion,such as furmark or extreme overvolting.
Another living proof,a mere Seasonic P760W could handle my previous 7970 CF heavy OC'ed without hiccups

If you're referring to TrueAudio, AMD has no driver for it yet, so as of right now it can't be used.

If you referring True Audio as codec,then its driverless.If AMD put True Audio more like hardware audio stream,more likely they will release a driver for it.Personally i have no clue about that and i'm not interested since the day of "passthrough"

-= edited=-

ummm, if you truly didnt' care about noise and temp why are you throwing a waterblock on it, and how much does a waterblock cost? around $150, so basically, 290x + waterblock cost around $800, that is only if you already have a waterloop setup, if you don't, well you are likely to run into the $1000s+ to have everything setup just to run this card within acceptable parameters.

Because WE LIKE IT!
Watercooling doesn't always mean performance,it more like aesthetic.Ask anyone around here who drowned their rigs,did they want to go back to the air?

 

[SIGSEGV]

so to sum it up:
1) they claim its cheaper
2) they ignore noise, temp and power consumption
3) then they say throw a water block on it ignoring the price of a water-cool kit (+$150 for the block and +$400 for the whole setup)
4) selective/inconsistency comparison in performance of 290x (in uber mode), but using silent mode to compare noise and temp.
5) overhype with words such as "destroy" "kill" "massacre" when benchmarks show they are fairly equal matched.
6) compares to titan when situation favors them(price). then compares to 780 when other situations fits them(performance).

meh..

naaahh...it wouldn't touch 92'C mark if you drowned them under water
Universal GPU block such as my XSPC Rasa have done perfectly well job keeping 7970 below 60'C...even if you furmark'ed them they still even touch 80'C.

even my GTX680 DC II T furmark'ed wouldn't touch 50*C

so, is it safe to say that this card wouldn't hit 65*C under water ?

 

[NutZInTheHead]

Nice one AMD.
Now all Nvidia needs to do is drop the price to something equal to or a tad-bit lower than the R9 290X.
And the price of GTX 780 needs to be around $499.

 

[sweet]

LOL people keep contradicting themselves on this forum.

First, they'll start off with the argument that the 290X is about ~$100 cheaper than the GTX780 (and that the price killed the 780)

and then they'll say "Oh who cares about temp, noise and power consumption", and then after that, they'll say "Just throw a waterblock on it".

ummm, if you truly didnt' care about noise and temp why are you throwing a waterblock on it, and how much does a waterblock cost? around $150, so basically, 290x + waterblock cost around $800, that is only if you already have a waterloop setup, if you don't, well you are likely to run into the $1000s+ to have everything setup just to run this card within acceptable parameters.

and don't give me this 'that everyone who buys high end already have a waterloop setup', that is just BS. most people don't run waterloops and those who do are in the small minorities, and the people who makes these claims don't even have waterloops themselves.

and when all arguments fail against them, they'll bring out the performance war, where 290x "absolutely destroyed" titan, um why are you comparing to titan first of all(it LOST to titan in silent mode, and if u are comparing in uber mode, then don't give me this its only 2db louder, because its 12db louder in uber mode), it barely even "destroyed" 780, they are well within 1~5% of each other trading blows. that is not what "destroy" means, "destroy" means beating something by at least 20%, otherwise, a few fps difference is barely even noticeable to the human eye.

Learn math please. 550$ + 150$ = 700$, not 800$. And a 600$ custom card is enough to fix the heat problem.
And I can quote my post above, because you are too lazy to scroll the page up

WRONG! They are not the same when overclocked. 290X is a beast when overclocked given that you can cool it down.

http://xtreview.com/images/radeon290X_002.jpg
http://xtreview.com/addcomment-id-31366-view-Radeon-R9-290X-vs-GeForce-GTX-Titan-benchmarks.html

In the chart above, MSI Lightning 780 actually run at 1300 MHz range, while 290X OC at 1100 MHz often throttles down near 1000 MHz. You can still observe a big different in performance here.

If you believe in TPU chart and claim that 780 is just 1% below 290 silent, let me tell you this fact: The overall chart of TPU is in fact, deceiving. In some adnormal case such as Starcraft 2, Splinter Cell: Blacklist, a.k.a games favoring old DirectX version, AMD's cards generally doesn't perform at their full potential. Only modern DirectX 11 games, which AMD's GCN cards are designed for, can take the best of AMD. The overall result, sadly, is heavily affected by the old tech games, and that doesn't feel right.

The fact is: Titan manages to beat 290X silent mode with the help of old tech games. Starcraft 2: DirectX 9, World of Warcraft and Splinter Cell: Blacklist : DirectX 9 engine with some DX11 tweak. If you exclude these games, Titan will be more miserable in the overall result.