R9 290 *Complete* review list

[thilanliyan]

So are the reviewers on newegg and amazon lying? Maybe, I don't know. There are more reports of this on AMD cards, that is a true fact -

What if the particular card with a lot of reviews about coil whine actually sold a ton more cards than cards with fewer reviews claiming coil whine?

Then which card actually has "more reports"? A percentage of reports with coil whine relative to amount of cards sold would be more indicative of which cards/brand suffers more from coil whine.

EDIT: Sooo...about them 290s...loud huh?

 

[smackababy]

EDIT: Sooo...about them 290s...loud huh?

Nope, not loud at all.

Any actual owners other than me want to chime in on their experience? It seems we get a lot of "reviews said it is x decibels louder than y card! omg it's a jet engine!", yet, the only other person I know to have said they actually have one is chimax and he already has his waterblocks.

 

[Phanuel]

Nope, not loud at all.

Any actual owners other than me want to chime in on their experience? It seems we get a lot of "reviews said it is x decibels louder than y card! omg it's a jet engine!", yet, the only other person I know to have said they actually have one is chimax and he already has his waterblocks.

Sapphire 290 here;

Mine is noticeable without headphones on or during a loading screen. But beyond that, I don't hear it while playing a game with sound on. And that's with it set to 93C and 49% fan.

No coil whine that I can hear, nor is the fan harsh. It's just a lot of moving air really.

 

[Cookie Monster]

It's not the PCB, its the components, power circuitry and capacitors. @ Cookie, I've fixed a coil whine card by using hot glue on the caps before, it prevents micro-vibrations as its charged and de-charged, the resonating vibrations cause the "whine".

No matter how expensive your card is or how awesome you think its crafted, Coil Whine will exist in a small % of cards due to the vibrations occurring within the human ear's hearing range.

It could be the PCB, it could be the components, but its usually a bit of both. Experience tells me that the culprit (the physical noise source) is usually the choke/DC inductor used in the DC-DC converter which also happens to be found on every video card.

Im guessing by "capacitors" you mean the SMD electrolytic ones, some boards use the flat SMD electrolytic ones (polymer based e.g. SP-Caps from Panasonic).. Your half right that its charging and dis-charging but the kind of ripple we are talking about is very very very small as you need a very fine regulation for the GPU. In other words, the charging/dis-charging is very very very.. very small. Physically I can't see how in this application it can produce audible noise unless the converter itself is unstable.

On the other hand, coils will always have a whine (its just its physical nature because you have a very thin wire, wound around a magnetic core and these aren't normally glued properly in place but rather wound really tight) but normally inaudible to the human year if the power circuitry is designed correctly as the frequency at which the converter runs is usually well outside the audible range for a human. You can use varnish to fix this issue so that the coil wire will never vibrate or reduce its vibration. Im guessing the ones with coil whine are the ones with component tolerances outside of its optimal values or noise creeping into the video card from a different source i.e. causing this issue.. e.g. your PSU.

My other guess is that the hot glue you applied to the output capacitors perhaps also was applied unintentionally to the choke/inductor (the large cube looking components - its actually a shield made from a magnetic material for the DC inductor for those interested) as it spread/flowed after application. These are usually and literally located right next to those capacitors.

On the topic, some people are generally more sensitive to noise vs some have more tolerance! The thing is that once you start delving into the art of near silent PCs.. you will never go back just like the first time one uses an IPS panel after using TN panels. GTX680 was considered silent but for me personally, 55% heck even 50% was abit loud.

 

[Abwx]

On the other hand, coils will always have a whine (its just its physical nature because you have a very thin wire, wound around a magnetic core and these aren't normally glued properly in place but rather wound really tight) but normally inaudible to the human year if the power circuitry is designed correctly as the frequency at which the converter runs is usually well outside the audible range for a human.

Actualy the wire is quite thick since it must withstand
the switched currents wich are about 2 Amperes per coil
but also send back this current at lower voltage wich will
yield 10A or so at the GPU voltage , the magnetic field strentgh
is so intense that the coil literaly act as a motor/speaker ,
that is, the electromagnetic field move the wires , hence
the noise , of course you re right in your technical description.

 

[Cookie Monster]

Actualy the wire is quite thick since it must withstand
the switched currents wich are about 2 Amperes per coil
but also send back this current at lower voltage wich will
yield 10A or so at the GPU voltage , the magnetic field strentgh
is so intense that the coil literaly act as a motor/speaker ,
that is, the electromagnetic field move the wires , hence
the noise , of course you re right in your technical description.

Yeah your correct. Thinking the frequency at which these converters switch in the video card must be quite high, perhaps in the MHz i.e. the inductors can be quite small i.e. you don't need a high inductance DC choke i.e. you dont need many turns and hence can use thick wires (even then its quite small given the size of chokes used on video cards) to handle the current rating. But anyhow, those wires will always want to physically move.

And your almost right with the current.. except it is more like 100~200As @full load@the required GPU voltage for most video cards Hence why the load is split amongst the different phases.

On a side note, its an interesting topic actually because for instance the GPU itself maybe consuming 220W of power but if the efficiency of the power circuitry is at 90%, your pulling ~244W from the PCI-e pins/slot (input supply). Those power measurements that reviewers do from the PCI-e 6pin/8pin and perhaps the slot is the total input power but they don't necessarily indicate the real consumption of the GPU core as the power efficiency of the converter and temperature can also determine the final consumption (there are more into this variable but lets keep it simple).

 

[Cookie Monster]

Just thinking about this topic.. I still don't understand why they went with the underperforming reference cooling system and set the threshold at 95C. Them saying its perfectly fine is just marketing talk and probably true in a sense that the card will be OK til its 2~3 year warranty is over

Electrical components are normally rated at 85C and the high temperature ones at 105/125C.. but come at a higher cost. High temperature often lead to more leakages, more lossy characteristics of IC/passive components e.g capacitors especially the electrolytic ones. It also shortens the lifespan of the components considerably unless everything on there is cooled. A more efficient design makes everything else so much easier and this could have been achieved with a better reference cooling system as shown by nVIDIA. Although not an expert with cooling systems, just from the gist of it there can be lots of improvements that can be made to improve the blower type cooling system. The reference GTX780/Titan heatsink is the evidence of this.

At the end of the day, you end up with a product that could cost $449 but consumes less power because its cooled better which in return might give you less throttling/more performance, its less noisy and you don't have to worry about buying 3rd party coolers/go water cooling. I hope they take note and deliver a well rounded product next time instead of falling short on one/two things.

 

[Silverforce11]

Just thinking about this topic.. I still don't understand why they went with the underperforming reference cooling system and set the threshold at 95C. Them saying its perfectly fine is just marketing talk and probably true in a sense that the card will be OK til its 2~3 year warranty is over

Oh, it will run for years just fine. We see that with the 480s. The problem is the huge increase in power use at high temps. There's already several review sites that have done custom cooling, showing R290/X using ~40W less when it runs at 70C opposed to 95C, that is, its not throttling clock speeds AND using less power at the same time. The reference cooler is trash for this reason alone, why would I want my GPU to be ~30% more power inefficient when it does not have to be??

There was an AMD interview on PCPER, this question was put to them, they responded by saying their OEM and partners prefer blower reference cards, and they had to meet all demands thus they kept the design the same. Which is really silly. It's marketing's way of saying: "Well.. our crap HSF combo has worked fine for us and our partners for many generations already, we're too lazy to design a better one eventhough we knew R290/X was much more power hungry, the same crap HSF would force it to run at 95C and its still not stable.. but meh, we don't care, 95C is fine!!!"

 

[3DVagabond]

Just thinking about this topic.. I still don't understand why they went with the underperforming reference cooling system and set the threshold at 95C. Them saying its perfectly fine is just marketing talk and probably true in a sense that the card will be OK til its 2~3 year warranty is over

Electrical components are normally rated at 85C and the high temperature ones at 105/125C.. but come at a higher cost. High temperature often lead to more leakages, more lossy characteristics of IC/passive components e.g capacitors especially the electrolytic ones. It also shortens the lifespan of the components considerably unless everything on there is cooled. A more efficient design makes everything else so much easier and this could have been achieved with a better reference cooling system as shown by nVIDIA. Although not an expert with cooling systems, just from the gist of it there can be lots of improvements that can be made to improve the blower type cooling system. The reference GTX780/Titan heatsink is the evidence of this.

At the end of the day, you end up with a product that could cost $449 but consumes less power because its cooled better which in return might give you less throttling/more performance, its less noisy and you don't have to worry about buying 3rd party coolers/go water cooling. I hope they take note and deliver a well rounded product next time instead of falling short on one/two things.

TSMC's 28nm HP process is rated for 125°C (military spec). 95°C isn't high stress at all.

 

[thilanliyan]

^All that ASUS and MSI Military class components mumbo jumbo is starting to make sense now!!

 

[Abwx]

Yeah your correct. Thinking the frequency at which these converters switch in the video card must be quite high, perhaps in the MHz i.e. the inductors can be quite small i.e. you don't need a high inductance DC choke i.e. you dont need many turns and hence can use thick wires (even then its quite small given the size of chokes used on video cards) to handle the current rating. But anyhow, those wires will always want to physically move.

And your almost right with the current.. except it is more like 100~200As @full load@the required GPU voltage for most video cards Hence why the load is split amongst the different phases.

On a side note, its an interesting topic actually because for instance the GPU itself maybe consuming 220W of power but if the efficiency of the power circuitry is at 90%, your pulling ~244W from the PCI-e pins/slot (input supply). Those power measurements that reviewers do from the PCI-e 6pin/8pin and perhaps the slot is the total input power but they don't necessarily indicate the real consumption of the GPU core as the power efficiency of the converter and temperature can also determine the final consumption (there are more into this variable but lets keep it simple).

Nice to see insightfull tech explanations , agree for everything
safe the coils current , actualy each coil provide a part of theses
200A , if there s 8 then it s 12.5A/coil , a single coil cant deal
with such high current , hence the multiphase power supplies,
but still , as you point it there s some losses in the supply
circuit that are indeed included in the total comsumption
of the card.

 

[Zanovar]

Some interesting reading on whine for sure,ive had it on a sapphire dualx 7970 and it was terrible,might have had it on other cards also but so slight not noticed it.but the screech on the duelx was unnacceptable(cicada-like).

 

[RussianSensation]

Oh, it will run for years just fine. We see that with the 480s. The problem is the huge increase in power use at high temps. There's already several review sites that have done custom cooling, showing R290/X using ~40W less when it runs at 70C opposed to 95C, that is, its not throttling clock speeds AND using less power at the same time. The reference cooler is trash

Great points. Having said that, the reference blower issue is greatly overblown for several reasons. Firstly, a lot of PC gamers use closed headphones or 2.1-5.1 systems. Once you crank the sounds & music in a game, you can hardly hear the videocard, even the likes of GTX480/R9 290X. Secondly, since HD4870 days, it was understood that AMD's reference cooler just doesn't cut the mustard. Why is this news now after reference HD5870-6970-7970? Talking about beating a dead horse. Thirdly, for 95% of DIY PC builders, despite the myths that keep getting perpetuated on forums about open air coolers increasing case temperatures by bazillion degrees, an open air cooler is superior -- it offers 20-30*C temperature drops and most importantly the ability to run high overclocked settings at reasonable temperatures and noise levels. This is NOT possible with any NV or AMD reference cooler. Fourthly, after-market cards come with their own perks - upgraded PCB/VRMs/Chokes that help to reduce the chance of coil whine, factory pre-overclocks -- all that for a mere $10-30 over reference cards.

Unless NV lowers prices, we are on a course for a match-up of 2x after-market R9 290s for $800 USD laying waste to a $700 GTX780Ti.

Hopefully AMD will finally redesign the reference design cooler for 20nm cards so that this issue is stopped being brought up and the focus can shift on how NV's $100-300 price premiums for 0-16% more performance can be justified due to PhysX, CUDA, 3D Vision, etc.

 

[Silverforce11]

Once you crank the sounds & music in a game, you can hardly hear the videocard, even the likes of GTX480/R9 290X. Secondly, since HD4870 days, it was understood that AMD's reference cooler just doesn't cut the mustard. Why is this news now after reference HD5870-6970-7970? Talking about beating a dead horse. Thirdly, for 95% of DIY PC builders, despite the myths that keep getting perpetuated on forums about open air coolers increasing case temperatures by bazillion degrees, an open air cooler is superior -- it offers 20-30*C temperature drops and most importantly the ability to run high overclocked settings at reasonable temperatures and noise levels. This is NOT possible with any NV or AMD reference cooler. Fourthly, after-market cards come with their own perks - upgraded PCB/VRMs/Chokes that help to reduce the chance of coil whine, factory pre-overclocks -- all that for a mere $10-30 over reference cards.

1. A lot of gamers use headphones, true.. but also a lot do not.

2. AMD's reference cooler has been horrible for a long long time, they are just too damn slack to fix it. It was acceptable in the old days, but these days, a $550 card deserves a better reference cooler, period.

3. Agreed on custom designs, open air design is even better in a SFF case, as long as airflow is decent: http://hardforum.com/showthread.php?t=1717132

4. I nearly snapped and ordered a Reference R290 + H220 + Aqua Computer R290 Waterblock.. but the cheapskate in me over-ruled, to hold out a little longer and wait for some AIB models to come.

 

[ICDP]

AMD's reference cooler has been horrible for a long long time, they are just too damn slack to fix it. It was acceptable in the old days, but these days, a $550 card deserves a better reference cooler, period.

This 100%.

R9 290 is £310 and 290X is £420 here in the UK. I wouldn't touch either with a barge pole with those horrendous clown coolers. The fact that many out there think like me is completely lost on AMD. They have completely ruled out R9 290/X cards as a viable GPU solution for many people. The sooner AMD realise cheap cooling = lost sales the better.

 

[raghu78]

This 100%.

R9 290 is £310 and 290X is £420 here in the UK. I wouldn't touch either with a barge pole with those horrendous clown coolers. The fact that many out there think like me is completely lost on AMD. They have completely ruled out R9 290/X cards as a viable GPU solution for many people. The sooner AMD realise cheap cooling = lost sales the better.

lets address this in 2 ways. first the people who run the card at its advertised stock settings. for them the stock cooler is moderately loud but not disturbingly loud. this has been commented in detail by brett of hardocp and scott of techreport who say they were perfectly comfortable using these cards for long durations on a open test bench. in a closed case the noise is going to be even lesser. considering the price at which the R9 290 is selling its a no-brainer. also there are users on ocn running the R9 290 cards at 1100 mhz with stock cooler at 75%. the cooler is loud at those speeds but many users are comfortable with the noise tradeoff in exchange for the outstanding price/perf. for the enthusiast who wants to overclock the card to the max and still have quiet operation the options are to slap a accelero xtreme iii or gelid icy vision. btw how quiet are those GTX 780 custom cards when you push them to the max. They definitely are louder than the accelero xtreme iii coolers

http://www.tomshardware.com/reviews/radeon-r9-290-review-benchmark,3659-19.html

http://www.overclock.net/t/1437634/installation-guide-tips-of-rev-2-icy-vision-on-r9-290x

http://www.overclock.net/t/1439731/bored-of-waiting-for-none-ref-290x-check-this-simple-cheap-fix

http://www.youtube.com/watch?v=U1RYjETWR7U

 

[ICDP]

lets address this in 2 ways.

Let's not. I don't give a crap what other people think about the fan. I don't like it because the same abysmal fan was on the 7970 reference cards. For me it was loud and unusable then and it's clearly loud and unusable now.

Everybody is different and AMD with this utterly crap cooler have instantly excluded me and the many people like me who refuse to have a hot and loud GPU. Their continued lack of movement on custom cooled AIB versions is becoming a joke and so many sales have already been lost to Nvidia.

AMD's marketing is a joke and they had Nvidia for the taking if they had simply had a decent cooler.

 

[DooKey]

Let's not. I don't give a crap what other people think about the fan. I don't like it because the same abysmal fan was on the 7970 reference cards. For me it was loud and unusable then and it's clearly loud and unusable now.

Everybody is different and AMD with this utterly crap cooler have instantly excluded me and the many people like me who refuse to have a hot and loud GPU. Their continued lack of movement on custom cooled AIB versions is becoming a joke and so many sales have already been lost to Nvidia.

AMD's marketing is a joke and they had Nvidia for the taking if they had simply had a decent cooler.

As the owner of 2x 290 I agree that the fans on the 290s are crap compared to the reference Nvidia cards like the 780/780ti/Titan. These cards compared to my old Titans are noisy as heck. Fortunately for AMD I like video cards and am willing to put up with the noise. Most people I know aren't so forgiving.

I find it's best to ignore the serial AMD marketing posters and do what you are doing and hold AMDs feet to the fire regarding their crap coolers.

 

[SlowSpyder]

Let's not. I don't give a crap what other people think about the fan. I don't like it because the same abysmal fan was on the 7970 reference cards. For me it was loud and unusable then and it's clearly loud and unusable now.

Everybody is different and AMD with this utterly crap cooler have instantly excluded me and the many people like me who refuse to have a hot and loud GPU. Their continued lack of movement on custom cooled AIB versions is becoming a joke and so many sales have already been lost to Nvidia.

AMD's marketing is a joke and they had Nvidia for the taking if they had simply had a decent cooler.

And other sales are lost to AMD from Nvidia due to Nvidia's price premium for given performance levels. No card is perfect for everyone. AMD's $400 card gives you the same performance and 33% more memory as Nvidia's $500 card, but is louder. No card is perfect, pick your poison.

 

[Silverforce11]

And other sales are lost to AMD from Nvidia due to Nvidia's price premium for given performance levels. No card is perfect for everyone. AMD's $400 card gives you the same performance and 33% more memory as Nvidia's $500 card, but is louder. No card is perfect, pick your poison.

TBH, they could have enabled custom versions to launch at the same time, these are the same designs for the R280X, as its been shown to fit perfect and able to keep R290X to 70C and staying quiet.

These custom variants are often only $10 extra compared to AMD's reference card.

So AMD missed the mark, they could have had a $400 card that some people would buy for need of a blower, or they could have had $410 cards that most would buy.. instead they gave no 2nd choice where the demand is. People are waiting and waiting.. many would probably jumped the gun and got a 780 or 780ti.

The really interesting result with custom cooling, it shows that R290/X running at 70C may end up using the same power as the 7970Ghz, since benches show a saving of 40W when its not running at 95C. It really is a big deal and again highlights, AMD = great engineers, horrible management. Custom cards at launch for a $10 premium would have had NV by the balls and SQUEEZED hard.

 

[raghu78]

And other sales are lost to AMD from Nvidia due to Nvidia's price premium for given performance levels. No card is perfect for everyone. AMD's $400 card gives you the same performance and 33% more memory as Nvidia's $500 card, but is louder. No card is perfect, pick your poison.

true. with an asus r9 290 dcii top or his iceq x2 at USD 400 - 430 the R9 290 would get very close to satisfying all aspects - performance , price/perf and low noise operation. in fact the power consumption also has been shown to be lower by 30w on R9 290 when a custom cooler was slapped. so the difference in power too reduces.

there are quite a few users doing the accelero xtreme iii or gelid icy vision or prolimatech mk-26 cooler mod.

http://translate.google.com/transla...celero_xtreme_iii_meets_amd_r9_290/index6.php

 

[RussianSensation]

Let's not. I don't give a crap what other people think about the fan. I don't like it because the same abysmal fan was on the 7970 reference cards. For me it was loud and unusable then and it's clearly loud and unusable now.

Why did you buy a reference 7970 blower card if you already knew that based on 5870/6970, the 7970 blower would be loud? Why would you buy a reference 7970 given how great it overclocked and knowing full well that a reference cooler would not allow 7970 to operate at low noise + temperature levels?

Fast forward to 290 and you are voicing the same complaints. There are other great alternatives if you do not want to wait for after-market 290s. Gigabyte Windforce Ghz Edition for $510 and 1071mhz Boost clocks is an awesome card.

It's been less than a month since R9 290 was released. If you paid attention to previous launches from AMD, it takes about 1-1.5 months before after-market versions are released. Even for NV, some cards like EVGA Classified or MSI Lightning took a whopping 5-6 months to be released from the time a reference 780 launched. I don't see anyone on the forum defending AMD's lousy reference cooler. Instead, we are just saying that a lot of PC gamers who have built modern systems know that based on facts, not myths, an open air cooler is superior to a blower in noise levels and temperatures in nearly every situation, excluding very small cases, or going quad-fire/quad-SLI. There people running MSI Lighting 7970 Tri-Fire and GTX780 Gigabyte Windforce TriSLI without problems, and this also is getting ignored.

A lot of people buy reference blowers because they are still stuck in the 90s where they used cases with terrible airflow and keep using them, or they keep accepting the perpetuating myth that an open air cooler dumps so much heat in the case that CPU and component temperatures go through the roof.

There are cases for $100 that can cope with 2x 300W after-market GPUs dumping heat straight into the case with little side-effects to motherboard, GPU and CPU temperatures.

There are plenty of reviews that have scientific evidence that disprove the idea that after-market open air coolers raise temperatures bazillion degrees and they still keep getting ignored because the myth that blowers are superior keeps getting perpetuated every year...

http://www.hardware.fr/articles/853-18/thermographie-infrarouge-systemes.html

A reference card cannot compete in 95% of systems. You can buy an after-market card and drop the fan speeds to 30% and let GPU temps ramp up to 80C of the reference card. Even if the GPU dumps 300W of heat into the case, you can install a tower mounted CPU heatsink or an AIO water cooling kit and it completely takes care of your CPU temps. Since after-market GPU coolers have so much headroom, you can force GPU fan speeds very low and the GPU will still not overheat.

Based on facts, not myths, a blower card will perform worse than a high quality open air cooler in any modern case with good airflow. Blowers are aimed primarily at OEMs where the cases are poor, airflow is terrible or super high end systems where you have no room for dual-slot open air coolers. For DIY market, however, there are 100s of cases and CPU towers to choose from which would allow one to build a system with 2 or even 3 flagship GPUs with open air cooler and have a quieter system overall vs. 2-3 reference cards. Once you take into account GPU overclocking, even the reference blower from NV has no chance of competing with MSI Gaming, Gigabyte Windforce 450W 3x or Asus DCUIIs.

 

[blackened23]

Why did you buy a reference 7970 blower card if you already knew that based on 5870/6970, the 7970 blower would be loud? Why would you buy a reference 7970 given how great it overclocked and knowing full well that a reference cooler would not allow 7970 to operate at low noise + temperature levels?

To answer this question you raise specifically:

It isn't the same issue. Using the 7970 overclocked at low noise levels with the reference shroud didn't require a huge drop-off in performance - and that is precisely what is happening with the 290. This isn't the same situation, which is why the 290 in stock form is really disappointing. If you want good acoustics, you lose a ton of performance. This was not the case with the 7970.

I could overclock my old 7970s and never, not once, did it lose performance for reasonably quiet operation and meager fan speeds. Same for the 6970 and 5870. Conversely you have Ryan Smith with charts showing 11-22% loss of performance for quiet fan speeds with the 290. Sorry, AMD should not have released the product with this type of trade-off IMO. It would be cool for someone somewhere to admit "hey we screwed up with the reference design" but apparently everyone here loves this new cooler AMD is using. How can that cooler have been worth all the bad press that AMD has received for throttling and noise? Was it worth it? I'd say no. Someone at AMD has to think that in hindsight, using that same reference design under the hood was the stupidest decision they could have made. It's just disappointing because the 290 is a good chip hobbled by that design.

 

[Skurge]

Why did you buy a reference 7970 blower card if you already knew that based on 5870/6970, the 7970 blower would be loud? Why would you buy a reference 7970 given how great it overclocked and knowing full well that a reference cooler would not allow 7970 to operate at low noise + temperature levels?

Fast forward to 290 and you are voicing the same complaints. There are other great alternatives if you do not want to wait for after-market 290s. Gigabyte Windforce Ghz Edition for $510 and 1071mhz Boost clocks is an awesome card.

It's been less than a month since R9 290 was released. If you paid attention to previous launches from AMD, it takes about 1-1.5 months before after-market versions are released. Even for NV, some cards like EVGA Classified or MSI Lightning took a whopping 5-6 months to be released from the time a reference 780 launched. I don't see anyone on the forum defending AMD's lousy reference cooler. Instead, we are just saying that a lot of PC gamers who have built modern systems know that based on facts, not myths, an open air cooler is superior to a blower in noise levels and temperatures in nearly every situation, excluding very small cases, or going quad-fire/quad-SLI. There people running MSI Lighting 7970 Tri-Fire and GTX780 Gigabyte Windforce TriSLI without problems, and this also is getting ignored.

A lot of people buy reference blowers because they are still stuck in the 90s where they used cases with terrible airflow and keep using them, or they keep accepting the perpetuating myth that an open air cooler dumps so much heat in the case that CPU and component temperatures go through the roof.

There are cases for $100 that can cope with 2x 300W after-market GPUs dumping heat straight into the case with little side-effects to motherboard, GPU and CPU temperatures.

There are plenty of reviews that have scientific evidence that disprove the idea that after-market open air coolers raise temperatures bazillion degrees and they still keep getting ignored because the myth that blowers are superior keeps getting perpetuated every year...

http://www.hardware.fr/articles/853-18/thermographie-infrarouge-systemes.html

A reference card cannot compete in 95% of systems. You can buy an after-market card and drop the fan speeds to 30% and let GPU temps ramp up to 80C of the reference card. Even if the GPU dumps 300W of heat into the case, you can install a tower mounted CPU heatsink or an AIO water cooling kit and it completely takes care of your CPU temps. Since after-market GPU coolers have so much headroom, you can force GPU fan speeds very low and the GPU will still not overheat.

Based on facts, not myths, a blower card will perform worse than a high quality open air cooler in any modern case with good airflow. Blowers are aimed primarily at OEMs where the cases are poor, airflow is terrible or super high end systems where you have no room for dual-slot open air coolers. For DIY market, however, there are 100s of cases and CPU towers to choose from which would allow one to build a system with 2 or even 3 flagship GPUs with open air cooler and have a quieter system overall vs. 2-3 reference cards. Once you take into account GPU overclocking, even the reference blower from NV has no chance of competing with MSI Gaming, Gigabyte Windforce 450W 3x or Asus DCUIIs.

I have to add that a case like mine, dual open cards, the case struggles to keep cool and CPU temps rose quite and a bit and with ivy bridge being as hot as it is, it is a legitimate concern as I had to ditch my aircooler and get a Corsair H60 to keep my overclock where it was.

This was with both a 6990 with a twin turbo and sapphire 7950s.

 

[Skurge]

To answer this question you raise specifically:

It isn't the same issue. Using the 7970 overclocked at low noise levels with the reference shroud didn't require a huge drop-off in performance - and that is precisely what is happening with the 290. This isn't the same situation, which is why the 290 in stock form is really disappointing. If you want good acoustics, you lose a ton of performance. This was not the case with the 7970.

I could overclock my old 7970s and never, not once, did it lose performance for reasonably quiet operation and meager fan speeds. Same for the 6970 and 5870. Conversely you have Ryan Smith with charts showing 11-22% loss of performance for quiet fan speeds with the 290. Sorry, AMD should not have released the product with this type of trade-off IMO. It would be cool for someone somewhere to admit "hey we screwed up with the reference design" but apparently everyone here loves this new cooler AMD is using. How can that cooler have been worth all the bad press that AMD has received for throttling and noise? Was it worth it? I'd say no. Someone at AMD has to think that in hindsight, using that same reference design under the hood was the stupidest decision they could have made. It's just disappointing because the 290 is a good chip hobbled by that design.

I'm with you all the way about the ref cooler being crap and should be better, but where are you getting your numbers from? Looking at the AT review again. I see less than 10% difference, in a lot of games, there isn't a difference.

Looking at all the reviews, I will be running my 290X in quiet mode as the performance difference is pretty minuscule.